diff --git a/.Rbuildignore b/.Rbuildignore index 91114bf..f5a7e6b 100644 --- a/.Rbuildignore +++ b/.Rbuildignore @@ -1,2 +1,11 @@ ^.*\.Rproj$ ^\.Rproj\.user$ +^LICENSE\.md$ +^exec$ +^test-dataset\.RDS$ +^sandbox$ +^example\.R$ +^\.github$ +^_pkgdown\.yml$ +^docs$ +^pkgdown$ diff --git a/.github/.gitignore b/.github/.gitignore new file mode 100644 index 0000000..2d19fc7 --- /dev/null +++ b/.github/.gitignore @@ -0,0 +1 @@ +*.html diff --git a/.github/workflows/R-CMD-check.yaml b/.github/workflows/R-CMD-check.yaml new file mode 100644 index 0000000..d46a617 --- /dev/null +++ b/.github/workflows/R-CMD-check.yaml @@ -0,0 +1,52 @@ +# Workflow derived from https://github.com/r-lib/actions/tree/v2/examples +# Need help debugging build failures? Start at https://github.com/r-lib/actions#where-to-find-help +on: + push: + branches: [main, master] + pull_request: + branches: [main, master] + +name: R-CMD-check.yaml + +permissions: read-all + +jobs: + R-CMD-check: + runs-on: ${{ matrix.config.os }} + + name: ${{ matrix.config.os }} (${{ matrix.config.r }}) + + strategy: + fail-fast: false + matrix: + config: + - {os: macos-latest, r: 'release'} + - {os: windows-latest, r: 'release'} + - {os: ubuntu-latest, r: 'devel', http-user-agent: 'release'} + - {os: ubuntu-latest, r: 'release'} + - {os: ubuntu-latest, r: 'oldrel-1'} + + env: + GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }} + R_KEEP_PKG_SOURCE: yes + + steps: + - uses: actions/checkout@v4 + + - uses: r-lib/actions/setup-pandoc@v2 + + - uses: r-lib/actions/setup-r@v2 + with: + r-version: ${{ matrix.config.r }} + http-user-agent: ${{ matrix.config.http-user-agent }} + use-public-rspm: true + + - uses: r-lib/actions/setup-r-dependencies@v2 + with: + extra-packages: any::rcmdcheck + needs: check + + - uses: r-lib/actions/check-r-package@v2 + with: + upload-snapshots: true + build_args: 'c("--no-manual","--compact-vignettes=gs+qpdf")' diff --git a/.gitignore b/.gitignore index fc0099d..be5d964 100644 --- a/.gitignore +++ b/.gitignore @@ -3,3 +3,4 @@ .RData *.DS_Store* *.Rproj +docs diff --git a/DESCRIPTION b/DESCRIPTION index 143d25b..12d3f8d 100644 --- a/DESCRIPTION +++ b/DESCRIPTION @@ -1,31 +1,36 @@ Package: CVN Title: Covariate-varying Networks Version: 1.1 -Date: 2022-09-05 Author: Louis Dijkstra [aut, cre] Maintainer: Louis Dijkstra -Description: A package for inferring high-dimensional - Gaussian graphical networks that change with - multiple discrete covariates +Description: A package for inferring high-dimensional Gaussian graphical + networks that change with multiple discrete covariates +License: GPL (>= 3) +URL: https://github.com/bips-hb/CVN +BugReports: https://github.com/bips-hb/CVN/issues +Depends: + R (>= 4.0.2), Imports: -Depends: - R (>= 4.0.2), - Rcpp, - doSNOW, - visNetwork, - parallel, - progress, - Matrix, - huge, - glasso, - crayon, - ggplot2, - reshape2, - dplyr -LinkingTo: Rcpp -License: GPL-3 + crayon, + doSNOW, + dplyr, + foreach, + ggplot2, + glasso, + magrittr, + Matrix, + parallel, + progress, + Rcpp, + reshape2, + snow, + stats, + utils, + visNetwork +LinkingTo: + Rcpp Encoding: UTF-8 LazyData: true -RoxygenNote: 7.2.3 -URL: https://github.com/bips-hb/CVN -BugReports: https://github.com/bips-hb/CVN/issues +RoxygenNote: 7.3.2 +Suggests: + igraph diff --git a/LICENSE.md b/LICENSE.md new file mode 100644 index 0000000..175443c --- /dev/null +++ b/LICENSE.md @@ -0,0 +1,595 @@ +GNU General Public License +========================== + +_Version 3, 29 June 2007_ +_Copyright © 2007 Free Software Foundation, Inc. <>_ + +Everyone is permitted to copy and distribute verbatim copies of this license +document, but changing it is not allowed. + +## Preamble + +The GNU General Public License is a free, copyleft license for software and other +kinds of works. + +The licenses for most software and other practical works are designed to take away +your freedom to share and change the works. By contrast, the GNU General Public +License is intended to guarantee your freedom to share and change all versions of a +program--to make sure it remains free software for all its users. We, the Free +Software Foundation, use the GNU General Public License for most of our software; it +applies also to any other work released this way by its authors. You can apply it to +your programs, too. + +When we speak of free software, we are referring to freedom, not price. Our General +Public Licenses are designed to make sure that you have the freedom to distribute +copies of free software (and charge for them if you wish), that you receive source +code or can get it if you want it, that you can change the software or use pieces of +it in new free programs, and that you know you can do these things. + +To protect your rights, we need to prevent others from denying you these rights or +asking you to surrender the rights. Therefore, you have certain responsibilities if +you distribute copies of the software, or if you modify it: responsibilities to +respect the freedom of others. + +For example, if you distribute copies of such a program, whether gratis or for a fee, +you must pass on to the recipients the same freedoms that you received. You must make +sure that they, too, receive or can get the source code. And you must show them these +terms so they know their rights. + +Developers that use the GNU GPL protect your rights with two steps: **(1)** assert +copyright on the software, and **(2)** offer you this License giving you legal permission +to copy, distribute and/or modify it. + +For the developers' and authors' protection, the GPL clearly explains that there is +no warranty for this free software. For both users' and authors' sake, the GPL +requires that modified versions be marked as changed, so that their problems will not +be attributed erroneously to authors of previous versions. + +Some devices are designed to deny users access to install or run modified versions of +the software inside them, although the manufacturer can do so. This is fundamentally +incompatible with the aim of protecting users' freedom to change the software. The +systematic pattern of such abuse occurs in the area of products for individuals to +use, which is precisely where it is most unacceptable. Therefore, we have designed +this version of the GPL to prohibit the practice for those products. If such problems +arise substantially in other domains, we stand ready to extend this provision to +those domains in future versions of the GPL, as needed to protect the freedom of +users. + +Finally, every program is threatened constantly by software patents. States should +not allow patents to restrict development and use of software on general-purpose +computers, but in those that do, we wish to avoid the special danger that patents +applied to a free program could make it effectively proprietary. To prevent this, the +GPL assures that patents cannot be used to render the program non-free. + +The precise terms and conditions for copying, distribution and modification follow. + +## TERMS AND CONDITIONS + +### 0. Definitions + +“This License” refers to version 3 of the GNU General Public License. + +“Copyright” also means copyright-like laws that apply to other kinds of +works, such as semiconductor masks. + +“The Program” refers to any copyrightable work licensed under this +License. Each licensee is addressed as “you”. “Licensees” and +“recipients” may be individuals or organizations. + +To “modify” a work means to copy from or adapt all or part of the work in +a fashion requiring copyright permission, other than the making of an exact copy. The +resulting work is called a “modified version” of the earlier work or a +work “based on” the earlier work. + +A “covered work” means either the unmodified Program or a work based on +the Program. + +To “propagate” a work means to do anything with it that, without +permission, would make you directly or secondarily liable for infringement under +applicable copyright law, except executing it on a computer or modifying a private +copy. Propagation includes copying, distribution (with or without modification), +making available to the public, and in some countries other activities as well. + +To “convey” a work means any kind of propagation that enables other +parties to make or receive copies. Mere interaction with a user through a computer +network, with no transfer of a copy, is not conveying. + +An interactive user interface displays “Appropriate Legal Notices” to the +extent that it includes a convenient and prominently visible feature that **(1)** +displays an appropriate copyright notice, and **(2)** tells the user that there is no +warranty for the work (except to the extent that warranties are provided), that +licensees may convey the work under this License, and how to view a copy of this +License. If the interface presents a list of user commands or options, such as a +menu, a prominent item in the list meets this criterion. + +### 1. Source Code + +The “source code” for a work means the preferred form of the work for +making modifications to it. “Object code” means any non-source form of a +work. + +A “Standard Interface” means an interface that either is an official +standard defined by a recognized standards body, or, in the case of interfaces +specified for a particular programming language, one that is widely used among +developers working in that language. + +The “System Libraries” of an executable work include anything, other than +the work as a whole, that **(a)** is included in the normal form of packaging a Major +Component, but which is not part of that Major Component, and **(b)** serves only to +enable use of the work with that Major Component, or to implement a Standard +Interface for which an implementation is available to the public in source code form. +A “Major Component”, in this context, means a major essential component +(kernel, window system, and so on) of the specific operating system (if any) on which +the executable work runs, or a compiler used to produce the work, or an object code +interpreter used to run it. + +The “Corresponding Source” for a work in object code form means all the +source code needed to generate, install, and (for an executable work) run the object +code and to modify the work, including scripts to control those activities. However, +it does not include the work's System Libraries, or general-purpose tools or +generally available free programs which are used unmodified in performing those +activities but which are not part of the work. For example, Corresponding Source +includes interface definition files associated with source files for the work, and +the source code for shared libraries and dynamically linked subprograms that the work +is specifically designed to require, such as by intimate data communication or +control flow between those subprograms and other parts of the work. + +The Corresponding Source need not include anything that users can regenerate +automatically from other parts of the Corresponding Source. + +The Corresponding Source for a work in source code form is that same work. + +### 2. Basic Permissions + +All rights granted under this License are granted for the term of copyright on the +Program, and are irrevocable provided the stated conditions are met. This License +explicitly affirms your unlimited permission to run the unmodified Program. The +output from running a covered work is covered by this License only if the output, +given its content, constitutes a covered work. This License acknowledges your rights +of fair use or other equivalent, as provided by copyright law. + +You may make, run and propagate covered works that you do not convey, without +conditions so long as your license otherwise remains in force. You may convey covered +works to others for the sole purpose of having them make modifications exclusively +for you, or provide you with facilities for running those works, provided that you +comply with the terms of this License in conveying all material for which you do not +control copyright. Those thus making or running the covered works for you must do so +exclusively on your behalf, under your direction and control, on terms that prohibit +them from making any copies of your copyrighted material outside their relationship +with you. + +Conveying under any other circumstances is permitted solely under the conditions +stated below. Sublicensing is not allowed; section 10 makes it unnecessary. + +### 3. Protecting Users' Legal Rights From Anti-Circumvention Law + +No covered work shall be deemed part of an effective technological measure under any +applicable law fulfilling obligations under article 11 of the WIPO copyright treaty +adopted on 20 December 1996, or similar laws prohibiting or restricting circumvention +of such measures. + +When you convey a covered work, you waive any legal power to forbid circumvention of +technological measures to the extent such circumvention is effected by exercising +rights under this License with respect to the covered work, and you disclaim any +intention to limit operation or modification of the work as a means of enforcing, +against the work's users, your or third parties' legal rights to forbid circumvention +of technological measures. + +### 4. Conveying Verbatim Copies + +You may convey verbatim copies of the Program's source code as you receive it, in any +medium, provided that you conspicuously and appropriately publish on each copy an +appropriate copyright notice; keep intact all notices stating that this License and +any non-permissive terms added in accord with section 7 apply to the code; keep +intact all notices of the absence of any warranty; and give all recipients a copy of +this License along with the Program. + +You may charge any price or no price for each copy that you convey, and you may offer +support or warranty protection for a fee. + +### 5. Conveying Modified Source Versions + +You may convey a work based on the Program, or the modifications to produce it from +the Program, in the form of source code under the terms of section 4, provided that +you also meet all of these conditions: + +* **a)** The work must carry prominent notices stating that you modified it, and giving a +relevant date. +* **b)** The work must carry prominent notices stating that it is released under this +License and any conditions added under section 7. This requirement modifies the +requirement in section 4 to “keep intact all notices”. +* **c)** You must license the entire work, as a whole, under this License to anyone who +comes into possession of a copy. This License will therefore apply, along with any +applicable section 7 additional terms, to the whole of the work, and all its parts, +regardless of how they are packaged. This License gives no permission to license the +work in any other way, but it does not invalidate such permission if you have +separately received it. +* **d)** If the work has interactive user interfaces, each must display Appropriate Legal +Notices; however, if the Program has interactive interfaces that do not display +Appropriate Legal Notices, your work need not make them do so. + +A compilation of a covered work with other separate and independent works, which are +not by their nature extensions of the covered work, and which are not combined with +it such as to form a larger program, in or on a volume of a storage or distribution +medium, is called an “aggregate” if the compilation and its resulting +copyright are not used to limit the access or legal rights of the compilation's users +beyond what the individual works permit. Inclusion of a covered work in an aggregate +does not cause this License to apply to the other parts of the aggregate. + +### 6. Conveying Non-Source Forms + +You may convey a covered work in object code form under the terms of sections 4 and +5, provided that you also convey the machine-readable Corresponding Source under the +terms of this License, in one of these ways: + +* **a)** Convey the object code in, or embodied in, a physical product (including a +physical distribution medium), accompanied by the Corresponding Source fixed on a +durable physical medium customarily used for software interchange. +* **b)** Convey the object code in, or embodied in, a physical product (including a +physical distribution medium), accompanied by a written offer, valid for at least +three years and valid for as long as you offer spare parts or customer support for +that product model, to give anyone who possesses the object code either **(1)** a copy of +the Corresponding Source for all the software in the product that is covered by this +License, on a durable physical medium customarily used for software interchange, for +a price no more than your reasonable cost of physically performing this conveying of +source, or **(2)** access to copy the Corresponding Source from a network server at no +charge. +* **c)** Convey individual copies of the object code with a copy of the written offer to +provide the Corresponding Source. This alternative is allowed only occasionally and +noncommercially, and only if you received the object code with such an offer, in +accord with subsection 6b. +* **d)** Convey the object code by offering access from a designated place (gratis or for +a charge), and offer equivalent access to the Corresponding Source in the same way +through the same place at no further charge. You need not require recipients to copy +the Corresponding Source along with the object code. If the place to copy the object +code is a network server, the Corresponding Source may be on a different server +(operated by you or a third party) that supports equivalent copying facilities, +provided you maintain clear directions next to the object code saying where to find +the Corresponding Source. Regardless of what server hosts the Corresponding Source, +you remain obligated to ensure that it is available for as long as needed to satisfy +these requirements. +* **e)** Convey the object code using peer-to-peer transmission, provided you inform +other peers where the object code and Corresponding Source of the work are being +offered to the general public at no charge under subsection 6d. + +A separable portion of the object code, whose source code is excluded from the +Corresponding Source as a System Library, need not be included in conveying the +object code work. + +A “User Product” is either **(1)** a “consumer product”, which +means any tangible personal property which is normally used for personal, family, or +household purposes, or **(2)** anything designed or sold for incorporation into a +dwelling. In determining whether a product is a consumer product, doubtful cases +shall be resolved in favor of coverage. For a particular product received by a +particular user, “normally used” refers to a typical or common use of +that class of product, regardless of the status of the particular user or of the way +in which the particular user actually uses, or expects or is expected to use, the +product. A product is a consumer product regardless of whether the product has +substantial commercial, industrial or non-consumer uses, unless such uses represent +the only significant mode of use of the product. + +“Installation Information” for a User Product means any methods, +procedures, authorization keys, or other information required to install and execute +modified versions of a covered work in that User Product from a modified version of +its Corresponding Source. The information must suffice to ensure that the continued +functioning of the modified object code is in no case prevented or interfered with +solely because modification has been made. + +If you convey an object code work under this section in, or with, or specifically for +use in, a User Product, and the conveying occurs as part of a transaction in which +the right of possession and use of the User Product is transferred to the recipient +in perpetuity or for a fixed term (regardless of how the transaction is +characterized), the Corresponding Source conveyed under this section must be +accompanied by the Installation Information. But this requirement does not apply if +neither you nor any third party retains the ability to install modified object code +on the User Product (for example, the work has been installed in ROM). + +The requirement to provide Installation Information does not include a requirement to +continue to provide support service, warranty, or updates for a work that has been +modified or installed by the recipient, or for the User Product in which it has been +modified or installed. Access to a network may be denied when the modification itself +materially and adversely affects the operation of the network or violates the rules +and protocols for communication across the network. + +Corresponding Source conveyed, and Installation Information provided, in accord with +this section must be in a format that is publicly documented (and with an +implementation available to the public in source code form), and must require no +special password or key for unpacking, reading or copying. + +### 7. Additional Terms + +“Additional permissions” are terms that supplement the terms of this +License by making exceptions from one or more of its conditions. Additional +permissions that are applicable to the entire Program shall be treated as though they +were included in this License, to the extent that they are valid under applicable +law. If additional permissions apply only to part of the Program, that part may be +used separately under those permissions, but the entire Program remains governed by +this License without regard to the additional permissions. + +When you convey a copy of a covered work, you may at your option remove any +additional permissions from that copy, or from any part of it. (Additional +permissions may be written to require their own removal in certain cases when you +modify the work.) You may place additional permissions on material, added by you to a +covered work, for which you have or can give appropriate copyright permission. + +Notwithstanding any other provision of this License, for material you add to a +covered work, you may (if authorized by the copyright holders of that material) +supplement the terms of this License with terms: + +* **a)** Disclaiming warranty or limiting liability differently from the terms of +sections 15 and 16 of this License; or +* **b)** Requiring preservation of specified reasonable legal notices or author +attributions in that material or in the Appropriate Legal Notices displayed by works +containing it; or +* **c)** Prohibiting misrepresentation of the origin of that material, or requiring that +modified versions of such material be marked in reasonable ways as different from the +original version; or +* **d)** Limiting the use for publicity purposes of names of licensors or authors of the +material; or +* **e)** Declining to grant rights under trademark law for use of some trade names, +trademarks, or service marks; or +* **f)** Requiring indemnification of licensors and authors of that material by anyone +who conveys the material (or modified versions of it) with contractual assumptions of +liability to the recipient, for any liability that these contractual assumptions +directly impose on those licensors and authors. + +All other non-permissive additional terms are considered “further +restrictions” within the meaning of section 10. If the Program as you received +it, or any part of it, contains a notice stating that it is governed by this License +along with a term that is a further restriction, you may remove that term. If a +license document contains a further restriction but permits relicensing or conveying +under this License, you may add to a covered work material governed by the terms of +that license document, provided that the further restriction does not survive such +relicensing or conveying. + +If you add terms to a covered work in accord with this section, you must place, in +the relevant source files, a statement of the additional terms that apply to those +files, or a notice indicating where to find the applicable terms. + +Additional terms, permissive or non-permissive, may be stated in the form of a +separately written license, or stated as exceptions; the above requirements apply +either way. + +### 8. Termination + +You may not propagate or modify a covered work except as expressly provided under +this License. Any attempt otherwise to propagate or modify it is void, and will +automatically terminate your rights under this License (including any patent licenses +granted under the third paragraph of section 11). + +However, if you cease all violation of this License, then your license from a +particular copyright holder is reinstated **(a)** provisionally, unless and until the +copyright holder explicitly and finally terminates your license, and **(b)** permanently, +if the copyright holder fails to notify you of the violation by some reasonable means +prior to 60 days after the cessation. + +Moreover, your license from a particular copyright holder is reinstated permanently +if the copyright holder notifies you of the violation by some reasonable means, this +is the first time you have received notice of violation of this License (for any +work) from that copyright holder, and you cure the violation prior to 30 days after +your receipt of the notice. + +Termination of your rights under this section does not terminate the licenses of +parties who have received copies or rights from you under this License. If your +rights have been terminated and not permanently reinstated, you do not qualify to +receive new licenses for the same material under section 10. + +### 9. Acceptance Not Required for Having Copies + +You are not required to accept this License in order to receive or run a copy of the +Program. Ancillary propagation of a covered work occurring solely as a consequence of +using peer-to-peer transmission to receive a copy likewise does not require +acceptance. However, nothing other than this License grants you permission to +propagate or modify any covered work. These actions infringe copyright if you do not +accept this License. Therefore, by modifying or propagating a covered work, you +indicate your acceptance of this License to do so. + +### 10. Automatic Licensing of Downstream Recipients + +Each time you convey a covered work, the recipient automatically receives a license +from the original licensors, to run, modify and propagate that work, subject to this +License. You are not responsible for enforcing compliance by third parties with this +License. + +An “entity transaction” is a transaction transferring control of an +organization, or substantially all assets of one, or subdividing an organization, or +merging organizations. If propagation of a covered work results from an entity +transaction, each party to that transaction who receives a copy of the work also +receives whatever licenses to the work the party's predecessor in interest had or +could give under the previous paragraph, plus a right to possession of the +Corresponding Source of the work from the predecessor in interest, if the predecessor +has it or can get it with reasonable efforts. + +You may not impose any further restrictions on the exercise of the rights granted or +affirmed under this License. For example, you may not impose a license fee, royalty, +or other charge for exercise of rights granted under this License, and you may not +initiate litigation (including a cross-claim or counterclaim in a lawsuit) alleging +that any patent claim is infringed by making, using, selling, offering for sale, or +importing the Program or any portion of it. + +### 11. Patents + +A “contributor” is a copyright holder who authorizes use under this +License of the Program or a work on which the Program is based. The work thus +licensed is called the contributor's “contributor version”. + +A contributor's “essential patent claims” are all patent claims owned or +controlled by the contributor, whether already acquired or hereafter acquired, that +would be infringed by some manner, permitted by this License, of making, using, or +selling its contributor version, but do not include claims that would be infringed +only as a consequence of further modification of the contributor version. For +purposes of this definition, “control” includes the right to grant patent +sublicenses in a manner consistent with the requirements of this License. + +Each contributor grants you a non-exclusive, worldwide, royalty-free patent license +under the contributor's essential patent claims, to make, use, sell, offer for sale, +import and otherwise run, modify and propagate the contents of its contributor +version. + +In the following three paragraphs, a “patent license” is any express +agreement or commitment, however denominated, not to enforce a patent (such as an +express permission to practice a patent or covenant not to sue for patent +infringement). To “grant” such a patent license to a party means to make +such an agreement or commitment not to enforce a patent against the party. + +If you convey a covered work, knowingly relying on a patent license, and the +Corresponding Source of the work is not available for anyone to copy, free of charge +and under the terms of this License, through a publicly available network server or +other readily accessible means, then you must either **(1)** cause the Corresponding +Source to be so available, or **(2)** arrange to deprive yourself of the benefit of the +patent license for this particular work, or **(3)** arrange, in a manner consistent with +the requirements of this License, to extend the patent license to downstream +recipients. “Knowingly relying” means you have actual knowledge that, but +for the patent license, your conveying the covered work in a country, or your +recipient's use of the covered work in a country, would infringe one or more +identifiable patents in that country that you have reason to believe are valid. + +If, pursuant to or in connection with a single transaction or arrangement, you +convey, or propagate by procuring conveyance of, a covered work, and grant a patent +license to some of the parties receiving the covered work authorizing them to use, +propagate, modify or convey a specific copy of the covered work, then the patent +license you grant is automatically extended to all recipients of the covered work and +works based on it. + +A patent license is “discriminatory” if it does not include within the +scope of its coverage, prohibits the exercise of, or is conditioned on the +non-exercise of one or more of the rights that are specifically granted under this +License. You may not convey a covered work if you are a party to an arrangement with +a third party that is in the business of distributing software, under which you make +payment to the third party based on the extent of your activity of conveying the +work, and under which the third party grants, to any of the parties who would receive +the covered work from you, a discriminatory patent license **(a)** in connection with +copies of the covered work conveyed by you (or copies made from those copies), or **(b)** +primarily for and in connection with specific products or compilations that contain +the covered work, unless you entered into that arrangement, or that patent license +was granted, prior to 28 March 2007. + +Nothing in this License shall be construed as excluding or limiting any implied +license or other defenses to infringement that may otherwise be available to you +under applicable patent law. + +### 12. No Surrender of Others' Freedom + +If conditions are imposed on you (whether by court order, agreement or otherwise) +that contradict the conditions of this License, they do not excuse you from the +conditions of this License. If you cannot convey a covered work so as to satisfy +simultaneously your obligations under this License and any other pertinent +obligations, then as a consequence you may not convey it at all. For example, if you +agree to terms that obligate you to collect a royalty for further conveying from +those to whom you convey the Program, the only way you could satisfy both those terms +and this License would be to refrain entirely from conveying the Program. + +### 13. Use with the GNU Affero General Public License + +Notwithstanding any other provision of this License, you have permission to link or +combine any covered work with a work licensed under version 3 of the GNU Affero +General Public License into a single combined work, and to convey the resulting work. +The terms of this License will continue to apply to the part which is the covered +work, but the special requirements of the GNU Affero General Public License, section +13, concerning interaction through a network will apply to the combination as such. + +### 14. Revised Versions of this License + +The Free Software Foundation may publish revised and/or new versions of the GNU +General Public License from time to time. Such new versions will be similar in spirit +to the present version, but may differ in detail to address new problems or concerns. + +Each version is given a distinguishing version number. If the Program specifies that +a certain numbered version of the GNU General Public License “or any later +version” applies to it, you have the option of following the terms and +conditions either of that numbered version or of any later version published by the +Free Software Foundation. If the Program does not specify a version number of the GNU +General Public License, you may choose any version ever published by the Free +Software Foundation. + +If the Program specifies that a proxy can decide which future versions of the GNU +General Public License can be used, that proxy's public statement of acceptance of a +version permanently authorizes you to choose that version for the Program. + +Later license versions may give you additional or different permissions. However, no +additional obligations are imposed on any author or copyright holder as a result of +your choosing to follow a later version. + +### 15. Disclaimer of Warranty + +THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. +EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES +PROVIDE THE PROGRAM “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER +EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF +MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE +QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE +DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. + +### 16. Limitation of Liability + +IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY +COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS THE PROGRAM AS +PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, +INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE +PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE +OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE +WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE +POSSIBILITY OF SUCH DAMAGES. + +### 17. Interpretation of Sections 15 and 16 + +If the disclaimer of warranty and limitation of liability provided above cannot be +given local legal effect according to their terms, reviewing courts shall apply local +law that most closely approximates an absolute waiver of all civil liability in +connection with the Program, unless a warranty or assumption of liability accompanies +a copy of the Program in return for a fee. + +_END OF TERMS AND CONDITIONS_ + +## How to Apply These Terms to Your New Programs + +If you develop a new program, and you want it to be of the greatest possible use to +the public, the best way to achieve this is to make it free software which everyone +can redistribute and change under these terms. + +To do so, attach the following notices to the program. It is safest to attach them +to the start of each source file to most effectively state the exclusion of warranty; +and each file should have at least the “copyright” line and a pointer to +where the full notice is found. + + + Copyright (C) + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . + +Also add information on how to contact you by electronic and paper mail. + +If the program does terminal interaction, make it output a short notice like this +when it starts in an interactive mode: + + Copyright (C) + This program comes with ABSOLUTELY NO WARRANTY; for details type 'show w'. + This is free software, and you are welcome to redistribute it + under certain conditions; type 'show c' for details. + +The hypothetical commands `show w` and `show c` should show the appropriate parts of +the General Public License. Of course, your program's commands might be different; +for a GUI interface, you would use an “about box”. + +You should also get your employer (if you work as a programmer) or school, if any, to +sign a “copyright disclaimer” for the program, if necessary. For more +information on this, and how to apply and follow the GNU GPL, see +<>. + +The GNU General Public License does not permit incorporating your program into +proprietary programs. If your program is a subroutine library, you may consider it +more useful to permit linking proprietary applications with the library. If this is +what you want to do, use the GNU Lesser General Public License instead of this +License. But first, please read +<>. diff --git a/NAMESPACE b/NAMESPACE index e434aeb..5d28b41 100644 --- a/NAMESPACE +++ b/NAMESPACE @@ -2,6 +2,7 @@ S3method(plot,cvn) S3method(print,cvn) +export("%>%") export(CVN) export(check_correctness_input) export(create_edges_visnetwork) @@ -32,4 +33,18 @@ export(updateZ_wrapper) export(visnetwork) export(visnetwork_cvn) import(Rcpp) +import(ggplot2) +import(visNetwork) +importFrom(Matrix,Matrix) +importFrom(doSNOW,registerDoSNOW) +importFrom(foreach,"%dopar%") +importFrom(foreach,foreach) +importFrom(magrittr,"%>%") +importFrom(snow,makeCluster) +importFrom(snow,stopCluster) +importFrom(stats,cov) +importFrom(stats,optim) +importFrom(stats,runif) +importFrom(stats,var) +importFrom(utils,combn) useDynLib(CVN) diff --git a/R/CVN-package.R b/R/CVN-package.R new file mode 100644 index 0000000..095e3fb --- /dev/null +++ b/R/CVN-package.R @@ -0,0 +1,33 @@ +#' @keywords internal +"_PACKAGE" + +## usethis namespace: start +#' @import ggplot2 +#' @import Rcpp +#' @import visNetwork +#' @importFrom foreach %dopar% +#' @importFrom foreach foreach +#' @importFrom Matrix Matrix +#' @importFrom snow makeCluster +#' @importFrom snow stopCluster +#' @importFrom doSNOW registerDoSNOW +#' @importFrom stats cov +#' @importFrom stats optim +#' @importFrom stats runif +#' @importFrom stats var +#' @importFrom utils combn +#' @useDynLib CVN +## usethis namespace: end +NULL + +# Silence global variable warning +# Ideally this would be solved more cleanly, but this suffices +utils::globalVariables(c( + "from", "to", "id", + "lambda1", "lambda2", + "gamma1", "gamma2", + "Theta", "Sigma", "data", + "Var1", "Var2", + "value", + "aic", "bic" +)) diff --git a/R/CVN.R b/R/CVN.R index f1c044c..85a9037 100644 --- a/R/CVN.R +++ b/R/CVN.R @@ -1,62 +1,62 @@ #' Estimating a Covariate-Varying Network (CVN) -#' -#' Estimates a covariate-varying network model (CVN), i.e., \eqn{m} -#' Gaussian graphical models that change with (multiple) external covariate(s). +#' +#' Estimates a covariate-varying network model (CVN), i.e., \eqn{m} +#' Gaussian graphical models that change with (multiple) external covariate(s). #' The smoothing between the graphs is specified by the \eqn{(m \times m)}-dimensional -#' weight matrix \eqn{W}. The function returns the estimated precision matrices -#' for each graph. -#' @section Reusing Estimates: When estimating the graph for different values of -#' \eqn{\lambda_1} and \eqn{\lambda_2}, we use the graph estimated (if available) -#' for other \eqn{\lambda_1} and \eqn{\lambda_2} values closest to them. -#' -#' @param data A list with matrices, each entry associated with a single graph. -#' The number of columns should be the same for each matrix. +#' weight matrix \eqn{W}. The function returns the estimated precision matrices +#' for each graph. +#' @section Reusing Estimates: When estimating the graph for different values of +#' \eqn{\lambda_1} and \eqn{\lambda_2}, we use the graph estimated (if available) +#' for other \eqn{\lambda_1} and \eqn{\lambda_2} values closest to them. +#' +#' @param data A list with matrices, each entry associated with a single graph. +#' The number of columns should be the same for each matrix. #' Number of observations can differ -#' @param W The \eqn{(m \times m)}-dimensional symmetric +#' @param W The \eqn{(m \times m)}-dimensional symmetric #' weight matrix \eqn{W} -#' @param lambda1 Vector with different \eqn{\lambda_1} LASSO penalty terms +#' @param lambda1 Vector with different \eqn{\lambda_1} LASSO penalty terms #' (Default: \code{1:2}) -#' @param <- Vector with different \eqn{\lambda_2} global smoothing parameter values +#' @param lambda2 Vector with different \eqn{\lambda_2} global smoothing parameter values #' (Default: \code{1:2}) -#' @param gamma1 A vector of \eqn{\gamma_1}'s LASSO penalty terms, where -#' \eqn{\gamma_1 = \frac{2 \lambda_1}{m p (1 - p)}}. If \code{gamma1} -#' is set, the value of \code{lambda1} is ignored. (Default: \code{NULL}). +#' @param gamma1 A vector of \eqn{\gamma_1}'s LASSO penalty terms, where +#' \eqn{\gamma_1 = \frac{2 \lambda_1}{m p (1 - p)}}. If \code{gamma1} +#' is set, the value of \code{lambda1} is ignored. (Default: \code{NULL}). #' @param gamma2 A vector of \eqn{\gamma_2}'s global smoothing parameters, where -#' that \eqn{\gamma_2 = \frac{4 \lambda_2}{m(m-1)p(p-1)}}. If \code{gamma2} +#' that \eqn{\gamma_2 = \frac{4 \lambda_2}{m(m-1)p(p-1)}}. If \code{gamma2} #' is set, the value of \code{lambda2} is ignored.(Default: \code{NULL}). #' @param rho The \eqn{\rho} penalty parameter for the global ADMM algorithm (Default: \code{1}) -#' @param eps If the relative difference between two update steps is -#' smaller than \eqn{\epsilon}, the algorithm stops. +#' @param eps If the relative difference between two update steps is +#' smaller than \eqn{\epsilon}, the algorithm stops. #' See \code{\link{relative_difference_precision_matrices}} #' (Default: \code{1e-4}) #' @param maxiter Maximum number of iterations (Default: \code{100}) -#' @param truncate All values of the final \eqn{\hat{\Theta}_i}'s below \code{truncate} will be +#' @param truncate All values of the final \eqn{\hat{\Theta}_i}'s below \code{truncate} will be #' set to \code{0}. (Default: \code{1e-5}) -#' @param rho_genlasso The \eqn{\rho} penalty parameter for the ADMM algorithm +#' @param rho_genlasso The \eqn{\rho} penalty parameter for the ADMM algorithm #' used to solve the generalized LASSO (Default: \code{1}) -#' @param eps_genlasso If the relative difference between two update steps is -#' smaller than \eqn{\epsilon}, the algorithm stops. +#' @param eps_genlasso If the relative difference between two update steps is +#' smaller than \eqn{\epsilon}, the algorithm stops. #' (Default: \code{1e-10}) -#' @param maxiter_genlasso Maximum number of iterations for solving +#' @param maxiter_genlasso Maximum number of iterations for solving #' the generalized LASSO problem (Default: \code{100}) -#' @param truncate_genlasso All values of the final \eqn{\hat{\beta}} below -#' \code{truncate_genlasso} will be set to \code{0}. +#' @param truncate_genlasso All values of the final \eqn{\hat{\beta}} below +#' \code{truncate_genlasso} will be set to \code{0}. #' (Default: \code{1e-4}) -#' @param n_cores Number of cores used (Default: max. number of cores - 1, or +#' @param n_cores Number of cores used (Default: max. number of cores - 1, or #' the total number penalty term pairs if that is less) #' @param normalized Data is normalized if \code{TRUE}. Otherwise the data is only #' centered (Default: \code{FALSE}) #' @param warmstart If \code{TRUE}, use the \code{\link[huge]{huge}} package for estimating #' the individual graphs first (Default: \code{TRUE}) -#' @param minimal If \code{TRUE}, the returned \code{cvn} is minimal in terms of -#' memory, i.e., \code{Theta}, \code{data} and \code{Sigma} are not +#' @param minimal If \code{TRUE}, the returned \code{cvn} is minimal in terms of +#' memory, i.e., \code{Theta}, \code{data} and \code{Sigma} are not #' returned (Default: \code{FALSE}) -#' @param verbose Verbose (Default: \code{TRUE}) -#' -#' @return A \code{CVN} object containing the estimates for all the graphs -#' for each different value of \eqn{(\lambda_1, \lambda_2)}. General results for +#' @param verbose Verbose (Default: \code{TRUE}) +#' +#' @return A \code{CVN} object containing the estimates for all the graphs +#' for each different value of \eqn{(\lambda_1, \lambda_2)}. General results for #' the different values of \eqn{(\lambda_1, \lambda_2)} can be found in the data frame -#' \code{results}. It consists of multiple columns, namely: +#' \code{results}. It consists of multiple columns, namely: #' \item{\code{lambda1}}{\eqn{\lambda_1} value} #' \item{\code{lambda2}}{\eqn{\lambda_2} value} #' \item{\code{converged}}{whether algorithm converged or not} @@ -68,15 +68,15 @@ #' \item{\code{id}}{The id. This corresponds to the indices of the lists} #' \item{\code{bic}}{Bayesian information criterion} #' The estimates of the precision matrices and the corresponding adjacency matrices -#' for the different values of \eqn{(\lambda_1, \lambda_2)} can be found -#' \item{\code{Theta}}{A list with the estimated precision matrices \eqn{\{ \hat{\Theta}_i(\lambda_1, \lambda_2) \}_{i = 1}^m}, +#' for the different values of \eqn{(\lambda_1, \lambda_2)} can be found +#' \item{\code{Theta}}{A list with the estimated precision matrices \eqn{\{ \hat{\Theta}_i(\lambda_1, \lambda_2) \}_{i = 1}^m}, #' (only if \code{minimal = FALSE})} -#' \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the -#' estimated precision matrices in \code{Theta}. The entries -#' are \code{1} if there is an edge, \code{0} otherwise. +#' \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the +#' estimated precision matrices in \code{Theta}. The entries +#' are \code{1} if there is an edge, \code{0} otherwise. #' The matrices are sparse using package \code{\link[Matrix]{Matrix}}} #' In addition, the input given to the CVN function is stored in the object as well: -#' \item{\code{Sigma}}{Empirical covariance matrices \eqn{\{\hat{\Sigma}_i\}_{i = 1}^m}, +#' \item{\code{Sigma}}{Empirical covariance matrices \eqn{\{\hat{\Sigma}_i\}_{i = 1}^m}, #' (only if \code{minimal = FALSE})} #' \item{\code{m}}{Number of graphs} #' \item{\code{p}}{Number of variables} @@ -84,12 +84,12 @@ #' \item{\code{data}}{The \code{data}, but then normalized or centered (only if \code{minimal = FALSE})} #' \item{\code{W}}{The \eqn{(m \times m)}-dimensional weight matrix \eqn{W}} #' \item{\code{maxiter}}{Maximum number of iterations for the ADMM} -#' \item{\code{rho}}{The \eqn{\rho} ADMM's penalty parameter} -#' \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} +#' \item{\code{rho}}{The \eqn{\rho} ADMM's penalty parameter} +#' \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} #' \item{\code{truncate}}{Truncation value for \eqn{\{ \hat{\Theta}_i \}_{i = 1}^m}} #' \item{\code{maxiter_genlasso}}{Maximum number of iterations for the generarlzed LASSO} -#' \item{\code{rho_genlasso}}{The \eqn{\rho} generalized LASSO penalty parameter} -#' \item{\code{eps_genlasso}}{The stopping criterion \eqn{\epsilon} for the generalized LASSO} +#' \item{\code{rho_genlasso}}{The \eqn{\rho} generalized LASSO penalty parameter} +#' \item{\code{eps_genlasso}}{The stopping criterion \eqn{\epsilon} for the generalized LASSO} #' \item{\code{truncate_genlasso}}{Truncation value for \eqn{\beta} of the generalized LASSO} #' \item{\code{n_lambda_values}}{Total number of \eqn{(\lambda_1, \lambda_2)} value combinations} #' \item{\code{normalized}}{If \code{TRUE}, \code{data} was normalized. Otherwise \code{data} was only centered} @@ -97,56 +97,57 @@ #' \item{\code{minimal}}{If \code{TRUE}, \code{data}, \code{Theta} and \code{Sigma} are not added} #' \item{\code{hits_border_aic}}{If \code{TRUE}, the optimal model based on the AIC hits the border of \eqn{(\lambda_1, \lambda_2)}} #' \item{\code{hits_border_bic}}{If \code{TRUE}, the optimal model based on the BIC hits the border of \eqn{(\lambda_1, \lambda_2)}} -#' @examples +#' @examples #' data(grid) #' m <- 9 # must be 9 for this example -#' -#' #' Choice of the weight matrix W. -#' #' (uniform random) +#' +#' #' Choice of the weight matrix W. +#' #' (uniform random) #' W <- matrix(runif(m*m), ncol = m) #' W <- W %*% t(W) #' W <- W / max(W) #' diag(W) <- 0 -#' +#' #' # lambdas: #' lambda1 = 1:2 #' lambda2 = 1:2 -#' -#' (cvn <- CVN::CVN(grid, W, lambda1 = lambda1, lambda2 = lambda2, eps = 1e-3, maxiter = 1000, verbose = TRUE)) +#' +#' (cvn <- CVN::CVN(grid, W, lambda1 = lambda1, lambda2 = lambda2, +#' eps = 1e-3, maxiter = 1000, verbose = TRUE)) #' @export -CVN <- function(data, W, lambda1 = 1:2, lambda2 = 1:2, - gamma1 = NULL, gamma2 = NULL, +CVN <- function(data, W, lambda1 = 1:2, lambda2 = 1:2, + gamma1 = NULL, gamma2 = NULL, rho = 1, eps = 1e-4, - maxiter = 100, - truncate = 1e-5, + maxiter = 100, + truncate = 1e-5, rho_genlasso = 1, - eps_genlasso = 1e-10, - maxiter_genlasso = 100, - truncate_genlasso = 1e-4, - n_cores = min(length(lambda1)*length(lambda2), parallel::detectCores() - 1), - normalized = FALSE, - warmstart = TRUE, - minimal = FALSE, - verbose = TRUE) { - + eps_genlasso = 1e-10, + maxiter_genlasso = 100, + truncate_genlasso = 1e-4, + n_cores = min(length(lambda1)*length(lambda2), parallel::detectCores() - 1), + normalized = FALSE, + warmstart = TRUE, + minimal = FALSE, + verbose = TRUE) { + # Check correctness input ------------------------------- - CVN::check_correctness_input(data, W, lambda1, lambda2, gamma1, gamma2, rho) - + check_correctness_input(data, W, lambda1, lambda2, gamma1, gamma2, rho) + # Extract variables ------------------------------------- - m <- length(data) # total number of graphs + m <- length(data) # total number of graphs p <- ncol(data[[1]]) # total number of variables - n_obs <- sapply(data, function(X) nrow(X)) # no. of observations per graph - + n_obs <- sapply(data, function(X) nrow(X)) # no. of observations per graph + # convert the lambda values to gamma values or the other way around - if (is.null(gamma1) && is.null(gamma2)) { - gamma1 <- 2*lambda1 / (m*p*(p-1)) - gamma2 <- 4*lambda2 / (m*(m-1)*p*(p-1)) - } else { - lambda1 <- (gamma1*m*p*(p-1)) / 2 + if (is.null(gamma1) && is.null(gamma2)) { + gamma1 <- 2*lambda1 / (m*p*(p-1)) + gamma2 <- 4*lambda2 / (m*(m-1)*p*(p-1)) + } else { + lambda1 <- (gamma1*m*p*(p-1)) / 2 lambda2 <- (gamma2*m*(m-1)*p*(p-1)) / 4 } - + # When the weight matrix is completely zero, there is no smoothing # between graphs. Therefore, the value of lambda2 is irrelevant. # In this case, we fix lambda2 to 0 and inform the user of this fact @@ -154,236 +155,241 @@ CVN <- function(data, W, lambda1 = 1:2, lambda2 = 1:2, warning("Since weight matrix W is zero, there is no smoothing. lambda2 is fixed to 0") lambda2 <- 0 } - + # Set-up cluster --------------------------- - if (n_cores > 1) { + if (n_cores > 1) { cl <- makeCluster(n_cores) registerDoSNOW(cl) opts <- list(progress = function(i) {i}) # empty place holder for foreach } - - if (verbose) { + + if (verbose) { cat(sprintf("Estimating a CVN with %d graphs...\n\n", m)) cat(sprintf("Number of cores: %d\n", n_cores)) - if (warmstart) { - cat("Uses a warmstart...\n\n") - } else { - cat("No warmstart...\n\n") + if (warmstart) { + cat("Uses a warmstart...\n\n") + } else { + cat("No warmstart...\n\n") } - + # Set-up a progress bar --------------------------------- - if (n_cores > 1) { + if (n_cores > 1) { pb <- progress::progress_bar$new( format = "estimating CVN [:bar] :percent eta: :eta", total = length(lambda1)*length(lambda2) + 1, clear = FALSE, width= 80, show_after = 0) pb$tick() - + progress <- function(i) pb$tick() opts <- list(progress = progress) # used by doSNOW } } - + # Center or normalize data ------------------------------- data <- lapply(data, function(X) scale(X, scale = normalized)) - + # Compute the empirical covariance matrices -------------- Sigma <- lapply(1:m, function(i) cov(data[[i]])*(n_obs[i] - 1) / n_obs[i]) - + # initialize results list ------------ global_res <- list( - Theta = list(), - adj_matrices = list(), + Theta = list(), + adj_matrices = list(), Sigma = Sigma, - m = m, - p = p, + m = m, + p = p, n_obs = n_obs, data = data, - W = W, - maxiter = maxiter, - rho = rho, + W = W, + maxiter = maxiter, + rho = rho, eps = eps, - truncate = truncate, - rho_genlasso = rho_genlasso, - eps_genlasso = eps_genlasso, - maxiter_genlasso = maxiter_genlasso, - truncate_genlasso = truncate_genlasso, - n_lambda_values = length(lambda1) * length(lambda2), + truncate = truncate, + rho_genlasso = rho_genlasso, + eps_genlasso = eps_genlasso, + maxiter_genlasso = maxiter_genlasso, + truncate_genlasso = truncate_genlasso, + n_lambda_values = length(lambda1) * length(lambda2), normalized = normalized, - warmstart = warmstart, + warmstart = warmstart, minimal = minimal ) - + # data frame with the results for each unique (lambda1,lambda2) pair - res <- data.frame(expand.grid(lambda1 = lambda1, - lambda2 = lambda2, - converged = FALSE, - value = NA, - n_iterations = NA, + res <- data.frame(expand.grid(lambda1 = lambda1, + lambda2 = lambda2, + converged = FALSE, + value = NA, + n_iterations = NA, aic = NA)) - res$gamma1 <- 2*res$lambda1 / (m*p*(p-1)) - res$gamma2 <- 4*res$lambda2 / (m*(m-1)*p*(p-1)) - + res$gamma1 <- 2*res$lambda1 / (m*p*(p-1)) + res$gamma2 <- 4*res$lambda2 / (m*(m-1)*p*(p-1)) + res$id <- 1:nrow(res) - + # estimate the graphs for the different values of (lambda1, lambda2) -------- - estimate_lambda_values <- function(i) { + estimate_lambda_values <- function(i) { # Create initial values for the ADMM algorithm ---------- Z <- rep(list(matrix(0, nrow = p, ncol = p)), m) # m (p x p)-dimensional zero matrices Y <- rep(list(matrix(0, nrow = p, ncol = p)), m) - + # if warmstart, the individual graphs are first estimated using the GLASSO. - if (warmstart) { - # We use the lambda1 value - Theta <- lapply(Sigma, function(S) { + if (warmstart) { + # We use the lambda1 value + Theta <- lapply(Sigma, function(S) { est <- glasso::glasso(s = S, rho = res$lambda1[i]) est$w }) - } else { + } else { Theta <- lapply(Sigma, function(S) diag(1/diag(S))) } - + # Initialize variables for the algorithm ----------------- if (sum(W) == 0) { # if the weight matrix is empty, the value a = eta_1^2 + 1 - a <- (res$lambda1[i] / rho)^2 + 1 - } else { + a <- (res$lambda1[i] / rho)^2 + 1 + } else { # Determine the value of the diagonal matrix A such that A - D'D > 0 (positive definite) a <- CVN::matrix_A_inner_ADMM(W, res$lambda1[i] / rho, res$lambda2[i] / rho) + 1 } - + # Estimate the graphs ------------------------------------- - eta1 <- res$lambda1[i] / rho - eta2 <- res$lambda2[i] / rho - CVN::estimate(m, p, W, Theta, Z, Y, a, eta1, eta2, Sigma, n_obs, - rho, rho_genlasso, - eps, eps_genlasso, - maxiter, maxiter_genlasso, truncate = truncate, - truncate_genlasso = truncate_genlasso, - verbose = verbose) + eta1 <- res$lambda1[i] / rho + eta2 <- res$lambda2[i] / rho + CVN::estimate(m, p, W, Theta, Z, Y, a, eta1, eta2, Sigma, n_obs, + rho, rho_genlasso, + eps, eps_genlasso, + maxiter, maxiter_genlasso, truncate = truncate, + truncate_genlasso = truncate_genlasso, + verbose = verbose) } - + # go over each pair of penalty terms if (n_cores > 1) { # parallel est <- foreach(i = 1:(length(lambda1) * length(lambda2)), .options.snow = opts) %dopar% { estimate_lambda_values(i) } - } else { # sequential - est <- lapply(1:(length(lambda1) * length(lambda2)), function(i) { + } else { # sequential + est <- lapply(1:(length(lambda1) * length(lambda2)), function(i) { estimate_lambda_values(i) }) } - + # Process results ----------------------------------------- - for (i in 1:(length(lambda1)*length(lambda2))) { + for (i in 1:(length(lambda1)*length(lambda2))) { global_res$Theta[[i]] <- est[[i]]$Z - global_res$adj_matrices[[i]] <- est[[i]]$adj_matrices - + global_res$adj_matrices[[i]] <- est[[i]]$adj_matrices + res$converged[i] <- est[[i]]$converged res$value[i] <- est[[i]]$value res$n_iterations[i] <- est[[i]]$n_iterations - + # determine the AIC - res$aic[i] <- CVN::determine_information_criterion(Theta = est[[i]]$Z, - adj_matrices = est[[i]]$adj_matrices, - Sigma = Sigma, + res$aic[i] <- CVN::determine_information_criterion(Theta = est[[i]]$Z, + adj_matrices = est[[i]]$adj_matrices, + Sigma = Sigma, n_obs = n_obs, - type = "AIC") - + type = "AIC") + # determine the BIC - res$bic[i] <- CVN::determine_information_criterion(Theta = est[[i]]$Z, - adj_matrices = est[[i]]$adj_matrices, - Sigma = Sigma, + res$bic[i] <- CVN::determine_information_criterion(Theta = est[[i]]$Z, + adj_matrices = est[[i]]$adj_matrices, + Sigma = Sigma, n_obs = n_obs, - type = "BIC") + type = "BIC") } - - # stop the progress bar - if (n_cores > 1 && verbose) { + + # stop the progress bar + if (n_cores > 1 && verbose) { pb$terminate() } - + # stop the cluster - if (n_cores > 1) { - stopCluster(cl) + if (n_cores > 1) { + stopCluster(cl) } - + # determine whether the optimal model based on the AIC or BIC hits the border # of lambda1 and/or lambda2 hit <- CVN::hits_end_lambda_intervals(res) global_res$hits_border_aic <- hit$hits_border_aic global_res$hits_border_bic <- hit$hits_border_bic - + # Collect all the results & input --------------------------- - global_res$results <- res - - class(global_res) <- c("cvn", "list") - - if (minimal) { - global_res <- strip_cvn(global_res) + global_res$results <- res + + class(global_res) <- c("cvn", "list") + + if (minimal) { + global_res <- strip_cvn(global_res) } - + return(global_res) } #' Print Function for the CVN Object Class #' #' @export -print.cvn <- function(cvn, ...) { # TODO +print.cvn <- function(x, ...) { # TODO cat(sprintf("Covariate-varying Network (CVN)\n\n")) - - if (all(cvn$results$converged)) { - cat(green(sprintf("\u2713 all converged\n\n"))) + + if (all(x$results$converged)) { + cat(crayon::green(sprintf("\u2713 all converged\n\n"))) } else { - cat(red(sprintf("\u2717 did not converge (maxiter of %d not sufficient)\n\n", cvn$maxiter))) + cat(crayon::red(sprintf("\u2717 did not converge (maxiter of %d not sufficient)\n\n", x$maxiter))) } - + # print following variables - cat(sprintf("Number of graphs (m) : %d\n", cvn$m)) - cat(sprintf("Number of variables (p) : %d\n", cvn$p)) - cat(sprintf("Number of lambda pairs : %d\n\n", cvn$n_lambda_values)) - - cat(sprintf("Weight matrix (W):\n")) - print(Matrix(cvn$W, sparse = T)) - + cat(sprintf("Number of graphs (m) : %d\n", x$m)) + cat(sprintf("Number of variables (p) : %d\n", x$p)) + cat(sprintf("Number of lambda pairs : %d\n\n", x$n_lambda_values)) + + # If x is an object returned by CVN::glasso() it doesn't have a $W element and this errors + # Causes example in glasso.R to error during R CMD check + # TODO: Check if glasso() behaves correctly, possibly adjust print method for CVN:glasso subclass? + if (!is.null(x$W)) { + cat(sprintf("Weight matrix (W):\n")) + print(Matrix::Matrix(x$W, sparse = T)) + } + cat(sprintf("\n")) - print(cvn$results) + print(x$results) } #' Strip CVN -#' -#' Function that removes most of the items to make the CVN object -#' more memory sufficient. This is especially important when the +#' +#' Function that removes most of the items to make the CVN object +#' more memory sufficient. This is especially important when the #' graphs are rather larger -#' +#' #' @param cvn \code{cvn} object #' #' @return Reduced cvn where \code{Theta}, \code{data} and \code{Sigma} #' are removed #' @export -strip_cvn <- function(cvn) { - - if ('Theta' %in% names(cvn)) { +strip_cvn <- function(cvn) { + + if ('Theta' %in% names(cvn)) { cvn <- within(cvn, rm(Theta)) } - - if ('data' %in% names(cvn)) { + + if ('data' %in% names(cvn)) { cvn <- within(cvn, rm(data)) } - - if ('Sigma' %in% names(cvn)) { + + if ('Sigma' %in% names(cvn)) { cvn <- within(cvn, rm(Sigma)) } - - # set the variable keeping track of whether the cvn is + + # set the variable keeping track of whether the cvn is # striped to TRUE cvn$minimal <- TRUE - + return(cvn) } #' Plot Function for CVN Object Class -#' +#' #' @export -plot.cvn <- function(cvn, ...) { - CVN::visnetwork_cvn(cvn, ...) +plot.cvn <- function(x, ...) { + CVN::visnetwork_cvn(x) } diff --git a/R/RcppExports.R b/R/RcppExports.R index 118747d..377c8dc 100644 --- a/R/RcppExports.R +++ b/R/RcppExports.R @@ -2,37 +2,69 @@ # Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393 #' Solving Generalized LASSO with fixed \eqn{\lambda = 1} -NULL - +#' +#' Solves efficiently the generalized LASSO problem of the form +#' \deqn{ +#' \hat{\beta} = \text{argmin } \frac{1}{2} || y - \beta ||_2^2 + ||D\beta||_1 +#' } +#' where \eqn{\beta} and \eqn{y} are \eqn{m}-dimensional vectors and +#' \eqn{D} is a \eqn{(c \times m)}-matrix where \eqn{c \geq m}. +#' We solve this optimization problem using an adaption of the ADMM +#' algorithm presented in Zhu (2017). +#' +#' @param y The \eqn{y} vector of length \eqn{m} +#' @param W The weight matrix \eqn{W} of dimensions \eqn{m x m} +#' @param m The number of graphs +#' @param eta1 Equals \eqn{\lambda_1 / rho} +#' @param eta2 Equals \eqn{\lambda_2 / rho} +#' @param a Value added to the diagonal of \eqn{-D'D} so that +#' the matrix is positive definite, see +#' \code{\link{matrix_A_inner_ADMM}} +#' @param rho The ADMM's parameter +#' @param max_iter Maximum number of iterations +#' @param eps Stopping criterion. If differences +#' are smaller than \eqn{\epsilon}, algorithm +#' is halted +#' @param truncate Values below \code{truncate} are +#' set to \code{0} +#' +#' @return The estimated vector \eqn{\hat{\beta}} +#' +#' @references +#' Zhu, Y. (2017). An Augmented ADMM Algorithm With Application to the +#' Generalized Lasso Problem. Journal of Computational and Graphical Statistics, +#' 26(1), 195–204. https://doi.org/10.1080/10618600.2015.1114491 +#' +#' @seealso \code{\link{genlasso_wrapper}} genlassoRcpp <- function(Y, W, m, eta1, eta2, a, rho, max_iter, eps, truncate) { .Call('_CVN_genlassoRcpp', PACKAGE = 'CVN', Y, W, m, eta1, eta2, a, rho, max_iter, eps, truncate) } #' The \eqn{Z}-update Step -#' -#' A \code{C} implementation of the \eqn{Z}-update step. We -#' solve a generalized LASSO problem repeatedly for each of the -#' individual edges -#' -#' @param m The number of graphs +#' +#' A \code{C} implementation of the \eqn{Z}-update step. We +#' solve a generalized LASSO problem repeatedly for each of the +#' individual edges +#' +#' @param m The number of graphs #' @param p The number of variables #' @param Theta A list of matrices with the \eqn{\Theta}-matrices #' @param Y A list of matrices with the \eqn{Y}-matrices -#' @param eta1 Equals \eqn{\lambda_1 / rho} -#' @param eta2 Equals \eqn{\lambda_2 / rho} +#' @param eta1 Equals \eqn{\lambda_1 / rho} +#' @param eta2 Equals \eqn{\lambda_2 / rho} #' @param a Value added to the diagonal of \eqn{-D'D} so that -#' the matrix is positive definite, see +#' the matrix is positive definite, see #' \code{\link{matrix_A_inner_ADMM}} #' @param rho The ADMM's parameter #' @param max_iter Maximum number of iterations -#' @param eps Stopping criterion. If differences +#' @param eps Stopping criterion. If differences #' are smaller than \eqn{\epsilon}, algorithm #' is halted -#' @param truncate Values below \code{truncate} are +#' @param truncate Values below \code{truncate} are #' set to \code{0} #' #' @return The estimated vector \eqn{\hat{\beta}} -#' +#' #' @seealso \code{\link{updateZ_wrapper}} updateZRcpp <- function(m, p, Theta, Y, W, eta1, eta2, a, rho, max_iter, eps, truncate) { .Call('_CVN_updateZRcpp', PACKAGE = 'CVN', m, p, Theta, Y, W, eta1, eta2, a, rho, max_iter, eps, truncate) diff --git a/R/data-grid.R b/R/data-grid.R new file mode 100644 index 0000000..3dabfe1 --- /dev/null +++ b/R/data-grid.R @@ -0,0 +1,14 @@ +#' Data for a grid of graphs (3 x 3) +#' +#' Data generated for 9 graphs in total, organized in a grid of +#' (3x3). See the package \code{CVNSim} for more information +#' on how the grid is constructed: \url{https://github.com/bips-hb/CVNSim} +#' +#' @name grid +#' @usage data(grid) +#' @docType data +#' @keywords datasets +#' @format List +#' @references \url{https://github.com/bips-hb/CVNSim} +grid + diff --git a/R/genlasso-wrapper.R b/R/genlasso-wrapper.R index a64f498..e757d99 100644 --- a/R/genlasso-wrapper.R +++ b/R/genlasso-wrapper.R @@ -1,9 +1,11 @@ #' Wrapper for \code{genlassoRcpp} -#' +#' #' See for details \code{\link{genlassoRcpp}} -#' +#' #' @seealso \code{\link{genlassoRcpp}} #' @export -genlasso_wrapper <- function(y, W, m, c, eta1, eta2, a, rho, max_iter, eps, truncate) { - genlassoRcpp(y, W, m, c, eta1, eta2, a, rho, max_iter, eps, truncate) +# FIXME: This passes an addiitonal argument 'c' that is not listed in the definition +# of genlassoRcpp in CVN.cpp +genlasso_wrapper <- function(y, W, m, c, eta1, eta2, a, rho, max_iter, eps, truncate) { + genlassoRcpp(y, W, m, c, eta1, eta2, a, rho, max_iter, eps, truncate) } diff --git a/R/glasso.R b/R/glasso.R index c084d9a..691b9e1 100644 --- a/R/glasso.R +++ b/R/glasso.R @@ -1,39 +1,39 @@ -#' Estimating Multiple Networks Separately -#' -#' A wrapper for the GLASSO in the context of CVNs. Each graph -#' is estimated individually. There is NO smoothing between the graphs. -#' This function relies completely on the \code{\link{glasso}} package. -#' The output is, therefore, slightly different than for the -#' \code{\link{CVN}} function. -#' -#' @param data A list with matrices, each entry associated with a single graph. -#' The number of columns should be the same for each matrix. +#' Estimating Multiple Networks Separately +#' +#' A wrapper for the GLASSO in the context of CVNs. Each graph +#' is estimated individually. There is NO smoothing between the graphs. +#' This function relies completely on the \code{\link{glasso}} package. +#' The output is, therefore, slightly different than for the +#' \code{\link{CVN}} function. +#' +#' @param data A list with matrices, each entry associated with a single graph. +#' The number of columns should be the same for each matrix. #' Number of observations can differ -#' @param lambda1 Vector with different \eqn{\lambda_1} LASSO penalty terms +#' @param lambda1 Vector with different \eqn{\lambda_1} LASSO penalty terms #' (Default: \code{1:2}) -#' @param eps Threshold for convergence (Default: \code{1e-4}; the same as in the +#' @param eps Threshold for convergence (Default: \code{1e-4}; the same as in the #' \code{glasso} package) #' @param maxiter Maximum number of iterations (Default: 10,000) -#' @param n_cores Number of cores used (Default: max. number of cores - 1, or +#' @param n_cores Number of cores used (Default: max. number of cores - 1, or #' the total number penalty term pairs if that is less) #' @param normalized Data is normalized if \code{TRUE}. Otherwise the data is only #' centered (Default: \code{FALSE}) -#' @param verbose Verbose (Default: \code{TRUE}) -#' -#' @return A \code{CVN} object containing the estimates for all the graphs -#' for different value of \eqn{\lambda_1}. General results for +#' @param verbose Verbose (Default: \code{TRUE}) +#' +#' @return A \code{CVN} object containing the estimates for all the graphs +#' for different value of \eqn{\lambda_1}. General results for #' the different value of \eqn{\lambda_1} can be found in the data frame -#' \code{results}. It consists of multiple columns, namely: +#' \code{results}. It consists of multiple columns, namely: #' \item{\code{lambda1}}{\eqn{\lambda_1} value} #' \item{\code{value}}{value of the negative log-likelihood function} #' \item{\code{aic}}{Aikake information criteration} #' \item{\code{id}}{The id. This corresponds to the indices of the lists} #' The estimates of the precision matrices and the corresponding adjacency matrices -#' for the different values of \eqn{\lambda_1} can be found +#' for the different values of \eqn{\lambda_1} can be found #' \item{\code{Theta}}{A list with the estimated precision matrices \eqn{\{ \hat{\Theta}_i(\lambda_1) \}_{i = 1}^m}} -#' \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the -#' estimated precision matrices in \code{Theta}. The entries -#' are \code{1} if there is an edge, \code{0} otherwise. +#' \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the +#' estimated precision matrices in \code{Theta}. The entries +#' are \code{1} if there is an edge, \code{0} otherwise. #' The matrices are sparse using package \code{\link[Matrix]{Matrix}}} #' In addition, the input given to this function is stored in the object as well: #' \item{\code{Sigma}}{Empirical covariance matrices \eqn{\{\hat{\Sigma}_i\}_{i = 1}^m}} @@ -42,123 +42,126 @@ #' \item{\code{n_obs}}{Vector of length \eqn{m} with number of observations for each graph} #' \item{\code{data}}{The \code{data}, but then normalized or centered} #' \item{\code{maxiter}}{Maximum number of iterations for the ADMM} -#' \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} +#' \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} #' \item{\code{n_lambda_values}}{Total number of \eqn{\lambda_1} values} #' \item{\code{normalized}}{If \code{TRUE}, \code{data} was normalized. Otherwise \code{data} was only centered} -#' @examples +#' @examples #' data(grid) #' m <- 9 # must be 9 for this example -#' -#' #' Choice of the weight matrix W. -#' #' (uniform random) +#' +#' #' Choice of the weight matrix W. +#' #' (uniform random) #' W <- matrix(runif(m*m), ncol = m) #' W <- W %*% t(W) #' W <- W / max(W) #' diag(W) <- 0 -#' +#' #' # lambdas: #' lambda1 = 1:4 -#' +#' #' (glasso_est <- CVN::glasso(grid, lambda1 = lambda1)) #' @export -glasso <- function(data, lambda1 = 1:2, +glasso <- function(data, lambda1 = 1:2, eps = 1e-4, - maxiter = 10000, - n_cores = min(length(lambda1), parallel::detectCores() - 1), - normalized = FALSE, - verbose = TRUE) { - + maxiter = 10000, + n_cores = min(length(lambda1), parallel::detectCores() - 1), + normalized = FALSE, + verbose = TRUE) { + # Extract variables ------------------------------------- - m <- length(data) # total number of graphs + m <- length(data) # total number of graphs p <- ncol(data[[1]]) # total number of variables - n_obs <- sapply(data, function(X) nrow(X)) # no. of observations per graph - + n_obs <- sapply(data, function(X) nrow(X)) # no. of observations per graph + # Set-up cluster --------------------------- cl <- makeCluster(n_cores) registerDoSNOW(cl) opts <- list(progress = function(i) {i}) # empty place holder for foreach - - if (verbose) { + + if (verbose) { cat(sprintf("Estimating %d individual graphs WITHOUT smoothing...\n\n", m)) cat(sprintf("Number of cores: %d\n", n_cores)) - + # Set-up a progress bar --------------------------------- pb <- progress::progress_bar$new( format = "estimating glasso [:bar] :percent eta: :eta", total = length(lambda1) + 1, clear = FALSE, width= 80, show_after = 0) pb$tick() - + #pb <- txtProgressBar(max = length(lambda1)*length(lambda2), style = 3) progress <- function(i) pb$tick() opts <- list(progress = progress) # used by doSNOW } - + # Center or normalize data ------------------------------- data <- lapply(data, function(X) scale(X, scale = normalized)) - + # Compute the empirical covariance matrices -------------- Sigma <- lapply(1:m, function(i) cov(data[[i]])*(n_obs[i] - 1) / n_obs[i]) - + # initialize results list ------------ global_res <- list( - Theta = lapply(1:length(lambda1), function(i) list()), - adj_matrices = lapply(1:length(lambda1), function(i) list()), + Theta = lapply(1:length(lambda1), function(i) list()), + adj_matrices = lapply(1:length(lambda1), function(i) list()), Sigma = Sigma, - m = m, - p = p, + m = m, + p = p, n_obs = n_obs, data = data, - maxiter = maxiter, + maxiter = maxiter, eps = eps, - n_lambda_values = length(lambda1), + n_lambda_values = length(lambda1), normalized = normalized ) - + # data frame with the results for each unique lambda1 value - res <- data.frame(expand.grid(lambda1 = lambda1, - value = NA, + res <- data.frame(expand.grid(lambda1 = lambda1, + value = NA, aic = NA)) res$id <- 1:nrow(res) - + # estimate the graphs for the different values of lambda1 -------- # go over each pair of penalty terms - est <- foreach(i = 1:(length(lambda1)), + est <- foreach(i = 1:(length(lambda1)), .options.snow = opts) %dopar% { - lapply(1:m, function(k) { + lapply(1:m, function(k) { glasso::glasso(s = Sigma[[k]], rho = res$lambda1[i]) }) } - + # Process results ----------------------------------------- - for (i in 1:(length(lambda1))) { - for (k in 1:m) { + for (i in 1:(length(lambda1))) { + for (k in 1:m) { global_res$Theta[[i]][[k]] <- est[[i]][[k]]$wi - + # create the adjacency matrix given the precision matrix - global_res$adj_matrices[[i]][[k]] <- Matrix( as.numeric( abs(est[[i]][[k]]$wi) >= 2*.Machine$double.eps), ncol = ncol(est[[i]][[k]]$wi) , sparse = TRUE) + global_res$adj_matrices[[i]][[k]] <- Matrix::Matrix( + as.numeric( abs(est[[i]][[k]]$wi) >= 2*.Machine$double.eps), + ncol = ncol(est[[i]][[k]]$wi), + sparse = TRUE) diag(global_res$adj_matrices[[i]][[k]]) <- 0 } - - res$aic[i] <- CVN::determine_information_criterion(Theta = global_res$Theta[[i]], - adj_matrices = global_res$adj_matrices[[i]], - Sigma = Sigma, + + res$aic[i] <- CVN::determine_information_criterion(Theta = global_res$Theta[[i]], + adj_matrices = global_res$adj_matrices[[i]], + Sigma = Sigma, n_obs = n_obs, - type = "AIC") + type = "AIC") } - - # stop the progress bar - if (verbose) { + + # stop the progress bar + if (verbose) { pb$terminate() } - + # stop the cluster - stopCluster(cl) - + stopCluster(cl) + # Collect all the results & input --------------------------- - global_res$results <- res - + global_res$results <- res + class(global_res) <- c("cvn", "cvn:glasso", "list") - - return(global_res) + + global_res } diff --git a/R/hits-end-lambda-intervals.R b/R/hits-end-lambda-intervals.R index cc94147..4325059 100644 --- a/R/hits-end-lambda-intervals.R +++ b/R/hits-end-lambda-intervals.R @@ -1,44 +1,44 @@ #' Hitting the end points of \eqn{(\lambda_1, \lambda_2)} interval -#' +#' #' One often selected the optimal model for the \eqn{(\lambda_1, \lambda_2)}-values -#' based on the AIC and BIC. -#' This function checks and warns when the optimal value lies on the border of the -#' values \eqn{(\lambda_1, \lambda_2)} takes. -#' +#' based on the AIC and BIC. +#' This function checks and warns when the optimal value lies on the border of the +#' values \eqn{(\lambda_1, \lambda_2)} takes. +#' #' @param results Results of the \code{\link{CVN}} function -#' -#' @return List with two values: +#' +#' @return List with two values: #' \item{\code{hits_border_aic}}{If \code{TRUE}, hits the border for the AIC} #' \item{\code{hits_border_bic}}{If \code{TRUE}, hits the border for the BIC} #' @export -hits_end_lambda_intervals <- function(results) { - +hits_end_lambda_intervals <- function(results) { + hits_border_aic <- FALSE hits_border_bic <- FALSE - + lambda1 <- unique(results$lambda1) lambda2 <- unique(results$lambda2) - + # if there are less than 3 entries for both lambda1 and lambda2, there is no point - # in checking whether the optimal estimate based on the AIC or the BIC lies on + # in checking whether the optimal estimate based on the AIC or the BIC lies on # the edge of the parameter search space - if(length(lambda1) >= 3 && length(lambda2) >= 3) { - + if(length(lambda1) >= 3 && length(lambda2) >= 3) { + # write the standard message and fill in the gaps later - message_warning <- function(criterion = c("BIC", "AIC"), - lambda = c("lambda1", "lambda2"), + message_warning <- function(criterion = c("BIC", "AIC"), + lambda = c("lambda1", "lambda2"), hits = c("smallest", "largest")) { - sprintf("In case you are selecting your model on the basis of the %s: Note that the maximum %s is achieved at the %s value of the %s interval. We suggest you increase the %s interval", - criterion[1], - criterion[1], - hits[1], + sprintf("In case you are selecting your model on the basis of the %s: Note that the maximum %s is achieved at the %s value of the %s interval. We suggest you increase the %s interval", + criterion[1], + criterion[1], + hits[1], lambda[1], lambda[1]) } - + # For the BIC score ---------------------------------------------------------- - temp <- results %>% filter(bic == max(bic)) - + temp <- results %>% dplyr::filter(bic == max(bic)) + # check whether it has a unique maximum if (nrow(temp) == 1) { # if lambda1 hits one of the left border @@ -46,29 +46,29 @@ hits_end_lambda_intervals <- function(results) { warning(message_warning("BIC", "lambda1", "smallest")) hits_border_bic <- TRUE } - + # if lambda1 hits one of the right border if (temp$lambda1 == max(lambda1)) { warning(message_warning("BIC", "lambda1", "largest")) hits_border_bic <- TRUE } - + # if lambda1 hits one of the left border if (temp$lambda2 == min(lambda2)) { warning(message_warning("BIC", "lambda2", "largest")) hits_border_bic <- TRUE } - + # if lambda1 hits one of the left border if (temp$lambda2 == max(lambda2)) { warning(message_warning("BIC", "lambda2", "largest")) hits_border_bic <- TRUE } } - + # For the AIC score ---------------------------------------------------------- - temp <- results %>% filter(aic == max(aic)) - + temp <- results %>% dplyr::filter(aic == max(aic)) + # check whether it has a unique maximum if (nrow(temp) == 1) { # if lambda1 hits one of the left border @@ -76,31 +76,31 @@ hits_end_lambda_intervals <- function(results) { warning(message_warning("AIC", "lambda1", "smallest")) hits_border_aic <- TRUE } - + # if lambda1 hits one of the right border if (temp$lambda1 == max(lambda1)) { warning(message_warning("AIC", "lambda1", "largest")) hits_border_aic <- TRUE } - + # if lambda1 hits one of the left border if (temp$lambda2 == min(lambda2)) { warning(message_warning("AIC", "lambda2", "largest")) hits_border_aic <- TRUE } - + # if lambda1 hits one of the left border if (temp$lambda2 == max(lambda2)) { warning(message_warning("AIC", "lambda2", "largest")) hits_border_aic <- TRUE } } - } - + } + return( list( hits_border_aic = hits_border_aic, hits_border_bic = hits_border_bic ) ) -} \ No newline at end of file +} diff --git a/R/plot-graph.R b/R/plot-graph.R index 578cf4b..442e516 100644 --- a/R/plot-graph.R +++ b/R/plot-graph.R @@ -1,280 +1,285 @@ #' Nodes for the \code{visNetwork} package -#' -#' Creates a data frame that can be used for the -#' \code{visNetwork} package. -#' +#' +#' Creates a data frame that can be used for the +#' \code{visNetwork} package. +#' #' @param n_nodes Number of nodes in the graph -#' @param labels The labels for the individual nodes +#' @param labels The labels for the individual nodes #' (Default: \code{1:n_nodes}) #' -#' @return Data frame with two columns: \code{id} and \code{title} -#' @export +#' @return Data frame with two columns: \code{id} and \code{title} +#' @export create_nodes_visnetwork <- function(n_nodes, labels = 1:n_nodes) { nodes <- data.frame(id = 1:n_nodes) nodes$title <- labels return(nodes) } -#' Create a \code{data.frame} for the Edges for \code{visNetwork} -#' -#' In order to visualize a graph, we need to create a -#' \code{data.frame} that can be used by the \code{visNetwork} package. +#' Create a \code{data.frame} for the Edges for \code{visNetwork} +#' +#' In order to visualize a graph, we need to create a +#' \code{data.frame} that can be used by the \code{visNetwork} package. #' This function returns the needed \code{data.frame} given a -#' adjacency matrix. -#' +#' adjacency matrix. +#' #' @param adj_matrix A symmetric adjacency matrix -#' -#' @return Data frame that be used as input for \code{visNetwork} +#' +#' @return Data frame that be used as input for \code{visNetwork} #' @export -create_edges_visnetwork <- function(adj_matrix) { - +create_edges_visnetwork <- function(adj_matrix) { + # needs to be of the type 'matrix' when using the function which adj_matrix <- as.matrix(adj_matrix) - + # set the lower diagonal to zero so that we do not count the same edge twice adj_matrix[lower.tri(adj_matrix)] <- 0 - + # find the indices of the non-zero elements in the adjacency matrix - edges <- as.data.frame(which(adj_matrix != 0, arr.ind = T)) - - # rename the columns of the data.frame to 'from' and 'to' which + edges <- as.data.frame(which(adj_matrix != 0, arr.ind = T)) + + # rename the columns of the data.frame to 'from' and 'to' which # is required for visNetwork - edges %>% + edges %>% dplyr::rename( from = row, to = col - ) %>% + ) %>% dplyr::arrange(from) } #' Add Attributes to Subset of Edges for \code{visNetwork} -#' -#' A subset of edges can be assign a different thickness -#' or color. -#' +#' +#' A subset of edges can be assign a different thickness +#' or color. +#' #' @param edges A data.frame create by \code{\link{create_edges_visnetwork}} -#' @param subset_edges A list with the elements \code{from} and \code{to}. Both -#' \code{from} and \code{to} are vectors of the same length -#' denoting the different edges -#' @param width Vector with two values. The first is assigned to the +#' @param subset_edges A list with the elements \code{from} and \code{to}. Both +#' \code{from} and \code{to} are vectors of the same length +#' denoting the different edges +#' @param width Vector with two values. The first is assigned to the #' edges in the subset given by \code{subset_edges}. The second -#' value is assigned to the rest. If \code{width = c(NA,NA)}, +#' value is assigned to the rest. If \code{width = c(NA,NA)}, #' no width is assigned -#' @param color Vector with two values. The first is assigned to the +#' @param color Vector with two values. The first is assigned to the #' edges in the subset given by \code{subset_edges}. The second -#' value is assigned to the rest. If \code{color = c(NULL,NULL)}, +#' value is assigned to the rest. If \code{color = c(NULL,NULL)}, #' no color is assigned -#' +#' #' @return A data frame that can be used by the \code{visNetwork} package #' @export -set_attributes_to_edges_visnetwork <- function(edges, - subset_edges, - width = c(NA, NA), - color = c(NULL, NULL)) { - +set_attributes_to_edges_visnetwork <- function(edges, + subset_edges, + width = c(NA, NA), + color = c(NULL, NULL)) { + # check whether the subset list is not empty - if (length(subset_edges$from) == 0) { - if (!is.na(width[1])) { + if (length(subset_edges$from) == 0) { + if (!is.na(width[1])) { edges$width <- width[2] } - - if (!is.null(color[1])) { + + if (!is.null(color[1])) { edges$color <- color[2] } return(edges) } - - # add IDs to the edges data.frame in order to keep track of the + + # add IDs to the edges data.frame in order to keep track of the # edges. The 'id' column is removed later on edges$id <- 1:nrow(edges) - + # filter out the edges that are in the subset_edges list ids <- edges %>% dplyr::filter( - from %in% subset_edges$from, + from %in% subset_edges$from, to %in% subset_edges$to ) - - # find the id codes of the edges in the 'in_group', i.e., + + # find the id codes of the edges in the 'in_group', i.e., # edges in the subset_edges. The 'out_group' is the rest in_group <- ids$id out_group <- edges$id[-in_group] - - # Setting the width of the in- and out group differently + + # Setting the width of the in- and out group differently # if width is given (not NA) - if (!is.na(width[1])) { + if (!is.na(width[1])) { edges$width[1:nrow(edges)] <- NA edges$width[in_group] <- width[1] edges$width[out_group] <- width[2] } - - # Setting the color of the in- and out group differently + + # Setting the color of the in- and out group differently # if color is given (not NULL) - if (!is.null(color[1])) { + if (!is.null(color[1])) { edges$color[1:nrow(edges)] <- NA edges$color[in_group] <- color[1] edges$color[out_group] <- color[2] } - + # remove the initially added 'id' column edges %>% dplyr::select(-id) } #' A \code{visNetwork} plot -#' -#' Creates a \code{visNetwork} plot given a list of -#' nodes and edges. The nodes data frame can be -#' created with \code{\link{create_nodes_visnetwork}}; -#' the edges with \code{create_edges_visnetwork}. -#' In order to highlight edges, you can use -#' \code{\link{set_attributes_to_edges_visnetwork}}. -#' +#' +#' Creates a \code{visNetwork} plot given a list of +#' nodes and edges. The nodes data frame can be +#' created with \code{\link{create_nodes_visnetwork}}; +#' the edges with \code{create_edges_visnetwork}. +#' In order to highlight edges, you can use +#' \code{\link{set_attributes_to_edges_visnetwork}}. +#' #' @return A \code{visNetwork} plot -#' @examples +#' @examples #' nodes <- CVN::create_nodes_visnetwork(n_nodes = 5, labels = LETTERS[1:5]) #' -#' adj_matrix <- matrix(c(0, 1, 0, 1, 0, -#' 1, 0, 1, 0, 0, -#' 0, 1, 0, 0, 0, +#' adj_matrix <- matrix(c(0, 1, 0, 1, 0, +#' 1, 0, 1, 0, 0, +#' 0, 1, 0, 0, 0, #' 1, 0, 0, 0, 1, #' 0, 0, 0, 1, 0), ncol = 5) -#' +#' #' edges <- CVN::create_edges_visnetwork(adj_matrix) -#' -#' edges <- set_attributes_to_edges_visnetwork(edges, +#' +#' edges <- set_attributes_to_edges_visnetwork(edges, #' subset_edges = list(from = c(1, 2), to = c(4, 3)), -#' width = c(3, .5), +#' width = c(3, .5), #' color = c("red", "blue")) -#' +#' #' CVN::visnetwork(nodes, edges) #' @export -visnetwork <- function(nodes, - edges, - node_titles = nodes$id, - title = "", - igraph_layout = "layout_in_circle") { - visNetwork(nodes, edges, width = "100%", main = list(text = title)) %>% +visnetwork <- function(nodes, + edges, + node_titles = nodes$id, + title = "", + igraph_layout = "layout_in_circle") { + + if (!(requireNamespace("igraph", quietly = TRUE))) { + stop("The igraph package is required for visNetwork, please install it first.") + } + + visNetwork(nodes, edges, width = "100%", main = list(text = title)) %>% visIgraphLayout(layout = igraph_layout) %>% visOptions(highlightNearest = list(enabled = T, hover = T)) } #' All \code{visNetwork} plots for a CVN object -#' -#' Creates all \code{visNetwork} plots, see \code{\link{CVN::visnetwork}}, +#' +#' Creates all \code{visNetwork} plots, see \code{\link{CVN::visnetwork}}, #' for all graphs in a \code{cvn} object -#' -#' @param cvn A \code{cvn} object, see \code{\link{CVN::CVN}} +#' +#' @param cvn A \code{cvn} object, see \code{\link{CVN::CVN}} #' or \code{\link{CVN::glasso}} #' @param node_titles Vector with title of the nodes (Default: \code{1:p}) -#' @param titles A list with \code{n_lambda_values} vectors. Each vector is of the +#' @param titles A list with \code{n_lambda_values} vectors. Each vector is of the #' lenght \code{m}. Regulates the titles of the graphs (Default: no title) #' @param show_core_graph,width,color Show the core graph using the width and colors -#' -#' @return List +#' +#' @return List #' @export -visnetwork_cvn <- function(cvn, - node_titles = 1:cvn$p, - titles = lapply(1:cvn$n_lambda_values, function(i) sapply(1:cvn$m, function(j) "")), - show_core_graph = TRUE, - width = c(3,1), - color = c("red", "blue"), - igraph_layout = "layout_in_circle", +visnetwork_cvn <- function(cvn, + node_titles = 1:cvn$p, + titles = lapply(1:cvn$n_lambda_values, function(i) sapply(1:cvn$m, function(j) "")), + show_core_graph = TRUE, + width = c(3,1), + color = c("red", "blue"), + igraph_layout = "layout_in_circle", verbose = TRUE) { - if (!(length(node_titles) == cvn$p)) { - stop("number of node labels does not correspond to the number of nodes") + if (!(length(node_titles) == cvn$p)) { + stop("number of node labels does not correspond to the number of nodes") } - - if (verbose) { + + if (verbose) { cat(sprintf("Creating visNetwork plots for the CVN...\n\n")) cat(sprintf("Number of graphs: %d\n", cvn$m)) cat(sprintf("Number of different lambda values: %d\n", cvn$n_lambda_values)) cat(sprintf("Creating nodes...\n")) } - + res <- list( - m = cvn$m, - p = cvn$p, - W = cvn$W, + m = cvn$m, + p = cvn$p, + W = cvn$W, results = cvn$results ) - - nodes <- CVN::create_nodes_visnetwork(n_nodes = cvn$p, labels = node_titles) - - # the edges that are constant in the different graphs are - # displayed differently - if (show_core_graph) { - + + nodes <- CVN::create_nodes_visnetwork(n_nodes = cvn$p, labels = node_titles) + + # the edges that are constant in the different graphs are + # displayed differently + if (show_core_graph) { + # get the core graphs for the different values of (lambda1, lamdba2) - if (verbose) { + if (verbose) { cat(sprintf("Determining the 'core graphs'...\n")) } - + core_graphs <- CVN::find_core_graph(cvn) - - if (verbose) { + + if (verbose) { cat(sprintf("Create the subset of edges in the core graphs...\n\n")) } - - subset_edges <- lapply(core_graphs, function(adj_matrix) { + + subset_edges <- lapply(core_graphs, function(adj_matrix) { as.list(create_edges_visnetwork(adj_matrix)) }) } - + # Set-up a progress bars --------------------------------- - if (verbose) { + if (verbose) { # progress bar for setting up the edges for the individual graphs pb_edges <- progress::progress_bar$new( format = "Creating edge lists [:bar] :percent eta: :eta", total = cvn$m * cvn$n_lambda_values + 1, clear = FALSE, width= 80, show_after = 0) pb_edges$tick() - + pb_plots <- progress::progress_bar$new( format = "Creating plots [:bar] :percent eta: :eta", total = cvn$m * cvn$n_lambda_values + 1, clear = FALSE, width= 80, show_after = 0) pb_plots$tick() } - + # create the edge dataframes for all the graphs all_edges <- lapply(1:cvn$n_lambda_values, function(i) { - lapply(1:cvn$m, function(k) { + lapply(1:cvn$m, function(k) { # cat(sprintf("%d\t%d\n", i,k)) edges <- CVN::create_edges_visnetwork(cvn$adj_matrices[[i]][[k]]) - # check whether there are core edges, since sometimes graphs are + # check whether there are core edges, since sometimes graphs are # completely empty - if (show_core_graph && length(subset_edges[[i]]$from) != 0) { + if (show_core_graph && length(subset_edges[[i]]$from) != 0) { edges <- CVN::set_attributes_to_edges_visnetwork(edges, subset_edges = subset_edges[[i]], width = width, color = color) } - - if (verbose) { + + if (verbose) { pb_edges$tick() } - + return(edges) }) }) - - if (verbose) { + + if (verbose) { pb_edges$terminate() - cat(sprintf("\nCreate plots given the determined edges...\n\n")) + cat(sprintf("\nCreate plots given the determined edges...\n\n")) } - + #return(all_edges) - + res$plots <- lapply(1:cvn$n_lambda_values, function(i) { - lapply(1:cvn$m, function(k) { - if (verbose) { + lapply(1:cvn$m, function(k) { + if (verbose) { pb_plots$tick() } return(CVN::visnetwork(nodes, all_edges[[i]][[k]], title = titles[[i]][[k]], igraph_layout = igraph_layout)) }) }) - - # stop the progress bar - if (verbose) { + + # stop the progress bar + if (verbose) { pb_plots$terminate() } - + return(res) } diff --git a/R/update-Z-wrapper.R b/R/update-Z-wrapper.R index 2b42422..0d78f42 100644 --- a/R/update-Z-wrapper.R +++ b/R/update-Z-wrapper.R @@ -1,38 +1,37 @@ #' Wrapper for the \eqn{Z}-update Step for the ADMM -#' -#' A wrapper for the \code{C} function that returns the -#' updated value of \eqn{Z} for the ADMM given the previously +#' +#' A wrapper for the \code{C} function that returns the +#' updated value of \eqn{Z} for the ADMM given the previously #' updated values of \eqn{\Theta} and \eqn{Y} -#' +#' #' @param m Number of graphs #' @param p Number of variables #' @param nrow_D Number of rows of the \eqn{D}-matrix #' @param Theta A list with matrices with the current values of \eqn{\Theta} -#' @param Y A list with matrices with the current values of \eqn{Y} -#' @param W Weight matrix -#' @param eta1 -#' @param rho_genlasso The \eqn{\rho} penalty parameter for the ADMM algorithm -#' @param eps_genlasso If the relative difference between two update steps is +#' @param Y A list with matrices with the current values of \eqn{Y} +#' @param W Weight matrix +#' @param eta1 TODO +#' @param rho_genlasso The \eqn{\rho} penalty parameter for the ADMM algorithm +#' @param eps_genlasso If the relative difference between two update steps is #' smaller than \eqn{\epsilon}, the algorithm stops -#' @param maxiter_genlasso Maximum number of iterations for solving -#' the generalized LASSO problem -#' @param truncate_genlasso All values of the final \eqn{\hat{\beta}} below -#' \code{truncate_genlasso} will be set to \code{0}. -#' +#' @param maxiter_genlasso Maximum number of iterations for solving +#' the generalized LASSO problem +#' @param truncate_genlasso All values of the final \eqn{\hat{\beta}} below +#' \code{truncate_genlasso} will be set to \code{0}. +#' #' @return A list with matrices with the new values of \eqn{Z} #' -#' @seealso \code{\link{create_matrix_D}} -#' +#' @seealso \code{\link{create_matrix_D}} +#' #' @export -updateZ_wrapper <- function(m, p, nrow_D, - Theta, Y, W, eta1, eta2, a, - rho_genlasso, maxiter_genlasso, eps_genlasso, - truncate_genlasso) { - - updateZRcpp(m, p, nrow_D, - Theta, Y, W, eta1, eta2, a, - rho_genlasso, maxiter_genlasso, eps_genlasso, - truncate_genlasso) - +updateZ_wrapper <- function(m, p, nrow_D, + Theta, Y, W, eta1, eta2, a, + rho_genlasso, maxiter_genlasso, eps_genlasso, + truncate_genlasso) { + + updateZRcpp(m, p, nrow_D, + Theta, Y, W, eta1, eta2, a, + rho_genlasso, maxiter_genlasso, eps_genlasso, + truncate_genlasso) + } - \ No newline at end of file diff --git a/R/utils-pipe.R b/R/utils-pipe.R new file mode 100644 index 0000000..fd0b1d1 --- /dev/null +++ b/R/utils-pipe.R @@ -0,0 +1,14 @@ +#' Pipe operator +#' +#' See \code{magrittr::\link[magrittr:pipe]{\%>\%}} for details. +#' +#' @name %>% +#' @rdname pipe +#' @keywords internal +#' @export +#' @importFrom magrittr %>% +#' @usage lhs \%>\% rhs +#' @param lhs A value or the magrittr placeholder. +#' @param rhs A function call using the magrittr semantics. +#' @return The result of calling `rhs(lhs)`. +NULL diff --git a/R/zzz.R b/R/zzz.R deleted file mode 100644 index d706f34..0000000 --- a/R/zzz.R +++ /dev/null @@ -1,26 +0,0 @@ -#' CVN -#' -#' An estimator for graphical models changing with multiple discrete -#' external covariates -#' -#' @docType package -#' @author Louis Dijkstra -#' @import Rcpp -#' @useDynLib CVN -#' @name CVN -NULL - -#' Data for a grid of graphs (3 x 3) -#' -#' Data generated for 9 graphs in total, organized in a grid of -#' (3x3). See the package \code{CVNSim} for more information -#' on how the grid is constructed: \url{https://github.com/bips-hb/CVNSim} -#' -#' @name grid -#' @usage data(grid) -#' @docType data -#' @keywords datasets -#' @format List -#' @references \url{https://github.com/bips-hb/CVNSim} -grid - diff --git a/README.md b/README.md index 528a1f8..9463bee 100644 --- a/README.md +++ b/README.md @@ -1,5 +1,9 @@ # Covariate-Varying Networks (CVN) + +[![R-CMD-check](https://github.com/bips-hb/CVN/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/bips-hb/CVN/actions/workflows/R-CMD-check.yaml) + + `CVN` is an `R` package for estimating high-dimensional Gaussian graphical models that change with multiple external covariates. The model is flexible, in the sense that complex smoothing patterns between the individual graphs can be used. See the commentary for more information. @@ -19,4 +23,4 @@ devtools::install_github("bips-hb/CVN") Louis Dijkstra Leibniz Institute for Prevention Research & Epidemiology -E-mail: dijkstra (at) leibniz-bips.de \ No newline at end of file +E-mail: dijkstra (at) leibniz-bips.de diff --git a/_pkgdown.yml b/_pkgdown.yml new file mode 100644 index 0000000..d71acfb --- /dev/null +++ b/_pkgdown.yml @@ -0,0 +1,4 @@ +url: ~ +template: + bootstrap: 5 + diff --git a/man/CVN-package.Rd b/man/CVN-package.Rd new file mode 100644 index 0000000..cd44194 --- /dev/null +++ b/man/CVN-package.Rd @@ -0,0 +1,18 @@ +% Generated by roxygen2: do not edit by hand +% Please edit documentation in R/CVN-package.R +\docType{package} +\name{CVN-package} +\alias{CVN-package} +\title{CVN: Covariate-varying Networks} +\description{ +A package for inferring high-dimensional Gaussian graphical networks that change with multiple discrete covariates +} +\seealso{ +Useful links: +\itemize{ + \item \url{https://github.com/bips-hb/CVN} + \item Report bugs at \url{https://github.com/bips-hb/CVN/issues} +} + +} +\keyword{internal} diff --git a/man/CVN.Rd b/man/CVN.Rd index 7de29b0..62e0ea0 100644 --- a/man/CVN.Rd +++ b/man/CVN.Rd @@ -1,6 +1,5 @@ % Generated by roxygen2: do not edit by hand -% Please edit documentation in R/CVN.R, R/zzz.R -\docType{package} +% Please edit documentation in R/CVN.R \name{CVN} \alias{CVN} \title{Estimating a Covariate-Varying Network (CVN)} @@ -28,54 +27,54 @@ CVN( ) } \arguments{ -\item{data}{A list with matrices, each entry associated with a single graph. -The number of columns should be the same for each matrix. +\item{data}{A list with matrices, each entry associated with a single graph. +The number of columns should be the same for each matrix. Number of observations can differ} -\item{W}{The \eqn{(m \times m)}-dimensional symmetric +\item{W}{The \eqn{(m \times m)}-dimensional symmetric weight matrix \eqn{W}} -\item{lambda1}{Vector with different \eqn{\lambda_1} LASSO penalty terms +\item{lambda1}{Vector with different \eqn{\lambda_1} LASSO penalty terms (Default: \code{1:2})} -\item{lambda2}{Vector with different \eqn{\lambda_2} global smoothing parameter values +\item{lambda2}{Vector with different \eqn{\lambda_2} global smoothing parameter values (Default: \code{1:2})} -\item{gamma1}{A vector of \eqn{\gamma_1}'s LASSO penalty terms, where -\eqn{\gamma_1 = \frac{2 \lambda_1}{m p (1 - p)}}. If \code{gamma1} +\item{gamma1}{A vector of \eqn{\gamma_1}'s LASSO penalty terms, where +\eqn{\gamma_1 = \frac{2 \lambda_1}{m p (1 - p)}}. If \code{gamma1} is set, the value of \code{lambda1} is ignored. (Default: \code{NULL}).} \item{gamma2}{A vector of \eqn{\gamma_2}'s global smoothing parameters, where -that \eqn{\gamma_2 = \frac{4 \lambda_2}{m(m-1)p(p-1)}}. If \code{gamma2} +that \eqn{\gamma_2 = \frac{4 \lambda_2}{m(m-1)p(p-1)}}. If \code{gamma2} is set, the value of \code{lambda2} is ignored.(Default: \code{NULL}).} \item{rho}{The \eqn{\rho} penalty parameter for the global ADMM algorithm (Default: \code{1})} -\item{eps}{If the relative difference between two update steps is -smaller than \eqn{\epsilon}, the algorithm stops. +\item{eps}{If the relative difference between two update steps is +smaller than \eqn{\epsilon}, the algorithm stops. See \code{\link{relative_difference_precision_matrices}} (Default: \code{1e-4})} \item{maxiter}{Maximum number of iterations (Default: \code{100})} -\item{truncate}{All values of the final \eqn{\hat{\Theta}_i}'s below \code{truncate} will be +\item{truncate}{All values of the final \eqn{\hat{\Theta}_i}'s below \code{truncate} will be set to \code{0}. (Default: \code{1e-5})} -\item{rho_genlasso}{The \eqn{\rho} penalty parameter for the ADMM algorithm +\item{rho_genlasso}{The \eqn{\rho} penalty parameter for the ADMM algorithm used to solve the generalized LASSO (Default: \code{1})} -\item{eps_genlasso}{If the relative difference between two update steps is -smaller than \eqn{\epsilon}, the algorithm stops. +\item{eps_genlasso}{If the relative difference between two update steps is +smaller than \eqn{\epsilon}, the algorithm stops. (Default: \code{1e-10})} -\item{maxiter_genlasso}{Maximum number of iterations for solving +\item{maxiter_genlasso}{Maximum number of iterations for solving the generalized LASSO problem (Default: \code{100})} -\item{truncate_genlasso}{All values of the final \eqn{\hat{\beta}} below -\code{truncate_genlasso} will be set to \code{0}. +\item{truncate_genlasso}{All values of the final \eqn{\hat{\beta}} below +\code{truncate_genlasso} will be set to \code{0}. (Default: \code{1e-4})} -\item{n_cores}{Number of cores used (Default: max. number of cores - 1, or +\item{n_cores}{Number of cores used (Default: max. number of cores - 1, or the total number penalty term pairs if that is less)} \item{normalized}{Data is normalized if \code{TRUE}. Otherwise the data is only @@ -84,17 +83,17 @@ centered (Default: \code{FALSE})} \item{warmstart}{If \code{TRUE}, use the \code{\link[huge]{huge}} package for estimating the individual graphs first (Default: \code{TRUE})} -\item{minimal}{If \code{TRUE}, the returned \code{cvn} is minimal in terms of -memory, i.e., \code{Theta}, \code{data} and \code{Sigma} are not +\item{minimal}{If \code{TRUE}, the returned \code{cvn} is minimal in terms of +memory, i.e., \code{Theta}, \code{data} and \code{Sigma} are not returned (Default: \code{FALSE})} \item{verbose}{Verbose (Default: \code{TRUE})} } \value{ -A \code{CVN} object containing the estimates for all the graphs - for each different value of \eqn{(\lambda_1, \lambda_2)}. General results for +A \code{CVN} object containing the estimates for all the graphs + for each different value of \eqn{(\lambda_1, \lambda_2)}. General results for the different values of \eqn{(\lambda_1, \lambda_2)} can be found in the data frame - \code{results}. It consists of multiple columns, namely: + \code{results}. It consists of multiple columns, namely: \item{\code{lambda1}}{\eqn{\lambda_1} value} \item{\code{lambda2}}{\eqn{\lambda_2} value} \item{\code{converged}}{whether algorithm converged or not} @@ -106,15 +105,15 @@ A \code{CVN} object containing the estimates for all the graphs \item{\code{id}}{The id. This corresponds to the indices of the lists} \item{\code{bic}}{Bayesian information criterion} The estimates of the precision matrices and the corresponding adjacency matrices - for the different values of \eqn{(\lambda_1, \lambda_2)} can be found - \item{\code{Theta}}{A list with the estimated precision matrices \eqn{\{ \hat{\Theta}_i(\lambda_1, \lambda_2) \}_{i = 1}^m}, + for the different values of \eqn{(\lambda_1, \lambda_2)} can be found + \item{\code{Theta}}{A list with the estimated precision matrices \eqn{\{ \hat{\Theta}_i(\lambda_1, \lambda_2) \}_{i = 1}^m}, (only if \code{minimal = FALSE})} - \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the - estimated precision matrices in \code{Theta}. The entries - are \code{1} if there is an edge, \code{0} otherwise. + \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the + estimated precision matrices in \code{Theta}. The entries + are \code{1} if there is an edge, \code{0} otherwise. The matrices are sparse using package \code{\link[Matrix]{Matrix}}} In addition, the input given to the CVN function is stored in the object as well: - \item{\code{Sigma}}{Empirical covariance matrices \eqn{\{\hat{\Sigma}_i\}_{i = 1}^m}, + \item{\code{Sigma}}{Empirical covariance matrices \eqn{\{\hat{\Sigma}_i\}_{i = 1}^m}, (only if \code{minimal = FALSE})} \item{\code{m}}{Number of graphs} \item{\code{p}}{Number of variables} @@ -122,12 +121,12 @@ A \code{CVN} object containing the estimates for all the graphs \item{\code{data}}{The \code{data}, but then normalized or centered (only if \code{minimal = FALSE})} \item{\code{W}}{The \eqn{(m \times m)}-dimensional weight matrix \eqn{W}} \item{\code{maxiter}}{Maximum number of iterations for the ADMM} - \item{\code{rho}}{The \eqn{\rho} ADMM's penalty parameter} - \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} + \item{\code{rho}}{The \eqn{\rho} ADMM's penalty parameter} + \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} \item{\code{truncate}}{Truncation value for \eqn{\{ \hat{\Theta}_i \}_{i = 1}^m}} \item{\code{maxiter_genlasso}}{Maximum number of iterations for the generarlzed LASSO} - \item{\code{rho_genlasso}}{The \eqn{\rho} generalized LASSO penalty parameter} - \item{\code{eps_genlasso}}{The stopping criterion \eqn{\epsilon} for the generalized LASSO} + \item{\code{rho_genlasso}}{The \eqn{\rho} generalized LASSO penalty parameter} + \item{\code{eps_genlasso}}{The stopping criterion \eqn{\epsilon} for the generalized LASSO} \item{\code{truncate_genlasso}}{Truncation value for \eqn{\beta} of the generalized LASSO} \item{\code{n_lambda_values}}{Total number of \eqn{(\lambda_1, \lambda_2)} value combinations} \item{\code{normalized}}{If \code{TRUE}, \code{data} was normalized. Otherwise \code{data} was only centered} @@ -137,18 +136,15 @@ A \code{CVN} object containing the estimates for all the graphs \item{\code{hits_border_bic}}{If \code{TRUE}, the optimal model based on the BIC hits the border of \eqn{(\lambda_1, \lambda_2)}} } \description{ -Estimates a covariate-varying network model (CVN), i.e., \eqn{m} -Gaussian graphical models that change with (multiple) external covariate(s). +Estimates a covariate-varying network model (CVN), i.e., \eqn{m} +Gaussian graphical models that change with (multiple) external covariate(s). The smoothing between the graphs is specified by the \eqn{(m \times m)}-dimensional -weight matrix \eqn{W}. The function returns the estimated precision matrices +weight matrix \eqn{W}. The function returns the estimated precision matrices for each graph. - -An estimator for graphical models changing with multiple discrete -external covariates } \section{Reusing Estimates}{ - When estimating the graph for different values of -\eqn{\lambda_1} and \eqn{\lambda_2}, we use the graph estimated (if available) + When estimating the graph for different values of +\eqn{\lambda_1} and \eqn{\lambda_2}, we use the graph estimated (if available) for other \eqn{\lambda_1} and \eqn{\lambda_2} values closest to them. } @@ -156,8 +152,8 @@ for other \eqn{\lambda_1} and \eqn{\lambda_2} values closest to them. data(grid) m <- 9 # must be 9 for this example -#' Choice of the weight matrix W. -#' (uniform random) +#' Choice of the weight matrix W. +#' (uniform random) W <- matrix(runif(m*m), ncol = m) W <- W \%*\% t(W) W <- W / max(W) @@ -167,8 +163,6 @@ diag(W) <- 0 lambda1 = 1:2 lambda2 = 1:2 -(cvn <- CVN::CVN(grid, W, lambda1 = lambda1, lambda2 = lambda2, eps = 1e-3, maxiter = 1000, verbose = TRUE)) -} -\author{ -Louis Dijkstra +(cvn <- CVN::CVN(grid, W, lambda1 = lambda1, lambda2 = lambda2, + eps = 1e-3, maxiter = 1000, verbose = TRUE)) } diff --git a/man/create_edges_visnetwork.Rd b/man/create_edges_visnetwork.Rd index 81eb6a9..33919bb 100644 --- a/man/create_edges_visnetwork.Rd +++ b/man/create_edges_visnetwork.Rd @@ -13,8 +13,8 @@ create_edges_visnetwork(adj_matrix) Data frame that be used as input for \code{visNetwork} } \description{ -In order to visualize a graph, we need to create a -\code{data.frame} that can be used by the \code{visNetwork} package. +In order to visualize a graph, we need to create a +\code{data.frame} that can be used by the \code{visNetwork} package. This function returns the needed \code{data.frame} given a adjacency matrix. } diff --git a/man/create_nodes_visnetwork.Rd b/man/create_nodes_visnetwork.Rd index f5f3cd1..9ce4baf 100644 --- a/man/create_nodes_visnetwork.Rd +++ b/man/create_nodes_visnetwork.Rd @@ -9,13 +9,13 @@ create_nodes_visnetwork(n_nodes, labels = 1:n_nodes) \arguments{ \item{n_nodes}{Number of nodes in the graph} -\item{labels}{The labels for the individual nodes +\item{labels}{The labels for the individual nodes (Default: \code{1:n_nodes})} } \value{ Data frame with two columns: \code{id} and \code{title} } \description{ -Creates a data frame that can be used for the +Creates a data frame that can be used for the \code{visNetwork} package. } diff --git a/man/genlassoRcpp.Rd b/man/genlassoRcpp.Rd new file mode 100644 index 0000000..feba40b --- /dev/null +++ b/man/genlassoRcpp.Rd @@ -0,0 +1,55 @@ +% Generated by roxygen2: do not edit by hand +% Please edit documentation in R/RcppExports.R +\name{genlassoRcpp} +\alias{genlassoRcpp} +\title{Solving Generalized LASSO with fixed \eqn{\lambda = 1}} +\usage{ +genlassoRcpp(Y, W, m, eta1, eta2, a, rho, max_iter, eps, truncate) +} +\arguments{ +\item{W}{The weight matrix \eqn{W} of dimensions \eqn{m x m}} + +\item{m}{The number of graphs} + +\item{eta1}{Equals \eqn{\lambda_1 / rho}} + +\item{eta2}{Equals \eqn{\lambda_2 / rho}} + +\item{a}{Value added to the diagonal of \eqn{-D'D} so that +the matrix is positive definite, see +\code{\link{matrix_A_inner_ADMM}}} + +\item{rho}{The ADMM's parameter} + +\item{max_iter}{Maximum number of iterations} + +\item{eps}{Stopping criterion. If differences +are smaller than \eqn{\epsilon}, algorithm +is halted} + +\item{truncate}{Values below \code{truncate} are +set to \code{0}} + +\item{y}{The \eqn{y} vector of length \eqn{m}} +} +\value{ +The estimated vector \eqn{\hat{\beta}} +} +\description{ +Solves efficiently the generalized LASSO problem of the form +\deqn{ + \hat{\beta} = \text{argmin } \frac{1}{2} || y - \beta ||_2^2 + ||D\beta||_1 +} +where \eqn{\beta} and \eqn{y} are \eqn{m}-dimensional vectors and +\eqn{D} is a \eqn{(c \times m)}-matrix where \eqn{c \geq m}. +We solve this optimization problem using an adaption of the ADMM +algorithm presented in Zhu (2017). +} +\references{ +Zhu, Y. (2017). An Augmented ADMM Algorithm With Application to the +Generalized Lasso Problem. Journal of Computational and Graphical Statistics, +26(1), 195–204. https://doi.org/10.1080/10618600.2015.1114491 +} +\seealso{ +\code{\link{genlasso_wrapper}} +} diff --git a/man/glasso.Rd b/man/glasso.Rd index 7f0c561..950748c 100644 --- a/man/glasso.Rd +++ b/man/glasso.Rd @@ -15,19 +15,19 @@ glasso( ) } \arguments{ -\item{data}{A list with matrices, each entry associated with a single graph. -The number of columns should be the same for each matrix. +\item{data}{A list with matrices, each entry associated with a single graph. +The number of columns should be the same for each matrix. Number of observations can differ} -\item{lambda1}{Vector with different \eqn{\lambda_1} LASSO penalty terms +\item{lambda1}{Vector with different \eqn{\lambda_1} LASSO penalty terms (Default: \code{1:2})} -\item{eps}{Threshold for convergence (Default: \code{1e-4}; the same as in the +\item{eps}{Threshold for convergence (Default: \code{1e-4}; the same as in the \code{glasso} package)} \item{maxiter}{Maximum number of iterations (Default: 10,000)} -\item{n_cores}{Number of cores used (Default: max. number of cores - 1, or +\item{n_cores}{Number of cores used (Default: max. number of cores - 1, or the total number penalty term pairs if that is less)} \item{normalized}{Data is normalized if \code{TRUE}. Otherwise the data is only @@ -36,20 +36,20 @@ centered (Default: \code{FALSE})} \item{verbose}{Verbose (Default: \code{TRUE})} } \value{ -A \code{CVN} object containing the estimates for all the graphs - for different value of \eqn{\lambda_1}. General results for +A \code{CVN} object containing the estimates for all the graphs + for different value of \eqn{\lambda_1}. General results for the different value of \eqn{\lambda_1} can be found in the data frame - \code{results}. It consists of multiple columns, namely: + \code{results}. It consists of multiple columns, namely: \item{\code{lambda1}}{\eqn{\lambda_1} value} \item{\code{value}}{value of the negative log-likelihood function} \item{\code{aic}}{Aikake information criteration} \item{\code{id}}{The id. This corresponds to the indices of the lists} The estimates of the precision matrices and the corresponding adjacency matrices - for the different values of \eqn{\lambda_1} can be found + for the different values of \eqn{\lambda_1} can be found \item{\code{Theta}}{A list with the estimated precision matrices \eqn{\{ \hat{\Theta}_i(\lambda_1) \}_{i = 1}^m}} - \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the - estimated precision matrices in \code{Theta}. The entries - are \code{1} if there is an edge, \code{0} otherwise. + \item{\code{adj_matrices}}{A list with the estimated adjacency matrices corresponding to the + estimated precision matrices in \code{Theta}. The entries + are \code{1} if there is an edge, \code{0} otherwise. The matrices are sparse using package \code{\link[Matrix]{Matrix}}} In addition, the input given to this function is stored in the object as well: \item{\code{Sigma}}{Empirical covariance matrices \eqn{\{\hat{\Sigma}_i\}_{i = 1}^m}} @@ -58,23 +58,23 @@ A \code{CVN} object containing the estimates for all the graphs \item{\code{n_obs}}{Vector of length \eqn{m} with number of observations for each graph} \item{\code{data}}{The \code{data}, but then normalized or centered} \item{\code{maxiter}}{Maximum number of iterations for the ADMM} - \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} + \item{\code{eps}}{The stopping criterion \eqn{\epsilon}} \item{\code{n_lambda_values}}{Total number of \eqn{\lambda_1} values} \item{\code{normalized}}{If \code{TRUE}, \code{data} was normalized. Otherwise \code{data} was only centered} } \description{ -A wrapper for the GLASSO in the context of CVNs. Each graph -is estimated individually. There is NO smoothing between the graphs. -This function relies completely on the \code{\link{glasso}} package. -The output is, therefore, slightly different than for the +A wrapper for the GLASSO in the context of CVNs. Each graph +is estimated individually. There is NO smoothing between the graphs. +This function relies completely on the \code{\link{glasso}} package. +The output is, therefore, slightly different than for the \code{\link{CVN}} function. } \examples{ data(grid) m <- 9 # must be 9 for this example -#' Choice of the weight matrix W. -#' (uniform random) +#' Choice of the weight matrix W. +#' (uniform random) W <- matrix(runif(m*m), ncol = m) W <- W \%*\% t(W) W <- W / max(W) diff --git a/man/grid.Rd b/man/grid.Rd index fc15cfe..62cc5cd 100644 --- a/man/grid.Rd +++ b/man/grid.Rd @@ -1,5 +1,5 @@ % Generated by roxygen2: do not edit by hand -% Please edit documentation in R/zzz.R +% Please edit documentation in R/data-grid.R \docType{data} \name{grid} \alias{grid} @@ -11,8 +11,8 @@ List data(grid) } \description{ -Data generated for 9 graphs in total, organized in a grid of -(3x3). See the package \code{CVNSim} for more information +Data generated for 9 graphs in total, organized in a grid of +(3x3). See the package \code{CVNSim} for more information on how the grid is constructed: \url{https://github.com/bips-hb/CVNSim} } \references{ diff --git a/man/hits_end_lambda_intervals.Rd b/man/hits_end_lambda_intervals.Rd index b9593c5..ee3a365 100644 --- a/man/hits_end_lambda_intervals.Rd +++ b/man/hits_end_lambda_intervals.Rd @@ -10,13 +10,13 @@ hits_end_lambda_intervals(results) \item{results}{Results of the \code{\link{CVN}} function} } \value{ -List with two values: +List with two values: \item{\code{hits_border_aic}}{If \code{TRUE}, hits the border for the AIC} \item{\code{hits_border_bic}}{If \code{TRUE}, hits the border for the BIC} } \description{ One often selected the optimal model for the \eqn{(\lambda_1, \lambda_2)}-values -based on the AIC and BIC. -This function checks and warns when the optimal value lies on the border of the +based on the AIC and BIC. +This function checks and warns when the optimal value lies on the border of the values \eqn{(\lambda_1, \lambda_2)} takes. } diff --git a/man/pipe.Rd b/man/pipe.Rd new file mode 100644 index 0000000..1f8f237 --- /dev/null +++ b/man/pipe.Rd @@ -0,0 +1,20 @@ +% Generated by roxygen2: do not edit by hand +% Please edit documentation in R/utils-pipe.R +\name{\%>\%} +\alias{\%>\%} +\title{Pipe operator} +\usage{ +lhs \%>\% rhs +} +\arguments{ +\item{lhs}{A value or the magrittr placeholder.} + +\item{rhs}{A function call using the magrittr semantics.} +} +\value{ +The result of calling `rhs(lhs)`. +} +\description{ +See \code{magrittr::\link[magrittr:pipe]{\%>\%}} for details. +} +\keyword{internal} diff --git a/man/plot.cvn.Rd b/man/plot.cvn.Rd index 9c06c83..d0cd1dc 100644 --- a/man/plot.cvn.Rd +++ b/man/plot.cvn.Rd @@ -4,7 +4,7 @@ \alias{plot.cvn} \title{Plot Function for CVN Object Class} \usage{ -\method{plot}{cvn}(cvn, ...) +\method{plot}{cvn}(x, ...) } \description{ Plot Function for CVN Object Class diff --git a/man/plot_information_criterion.Rd b/man/plot_information_criterion.Rd index 78b0a1d..4e7c60c 100644 --- a/man/plot_information_criterion.Rd +++ b/man/plot_information_criterion.Rd @@ -1,5 +1,5 @@ % Generated by roxygen2: do not edit by hand -% Please edit documentation in R/plot-aic.R +% Please edit documentation in R/plot-information-criterion.R \name{plot_information_criterion} \alias{plot_information_criterion} \title{Heat Map of an Information Criterion (AIC or BIC)} diff --git a/man/print.cvn.Rd b/man/print.cvn.Rd index d1d395f..fd66254 100644 --- a/man/print.cvn.Rd +++ b/man/print.cvn.Rd @@ -4,7 +4,7 @@ \alias{print.cvn} \title{Print Function for the CVN Object Class} \usage{ -\method{print}{cvn}(cvn, ...) +\method{print}{cvn}(x, ...) } \description{ Print Function for the CVN Object Class diff --git a/man/set_attributes_to_edges_visnetwork.Rd b/man/set_attributes_to_edges_visnetwork.Rd index 42b84df..0c612fd 100644 --- a/man/set_attributes_to_edges_visnetwork.Rd +++ b/man/set_attributes_to_edges_visnetwork.Rd @@ -14,24 +14,24 @@ set_attributes_to_edges_visnetwork( \arguments{ \item{edges}{A data.frame create by \code{\link{create_edges_visnetwork}}} -\item{subset_edges}{A list with the elements \code{from} and \code{to}. Both -\code{from} and \code{to} are vectors of the same length +\item{subset_edges}{A list with the elements \code{from} and \code{to}. Both +\code{from} and \code{to} are vectors of the same length denoting the different edges} -\item{width}{Vector with two values. The first is assigned to the +\item{width}{Vector with two values. The first is assigned to the edges in the subset given by \code{subset_edges}. The second -value is assigned to the rest. If \code{width = c(NA,NA)}, +value is assigned to the rest. If \code{width = c(NA,NA)}, no width is assigned} -\item{color}{Vector with two values. The first is assigned to the +\item{color}{Vector with two values. The first is assigned to the edges in the subset given by \code{subset_edges}. The second -value is assigned to the rest. If \code{color = c(NULL,NULL)}, +value is assigned to the rest. If \code{color = c(NULL,NULL)}, no color is assigned} } \value{ A data frame that can be used by the \code{visNetwork} package } \description{ -A subset of edges can be assign a different thickness +A subset of edges can be assign a different thickness or color. } diff --git a/man/strip_cvn.Rd b/man/strip_cvn.Rd index 5363758..aa49ddd 100644 --- a/man/strip_cvn.Rd +++ b/man/strip_cvn.Rd @@ -14,7 +14,7 @@ Reduced cvn where \code{Theta}, \code{data} and \code{Sigma} are removed } \description{ -Function that removes most of the items to make the CVN object -more memory sufficient. This is especially important when the +Function that removes most of the items to make the CVN object +more memory sufficient. This is especially important when the graphs are rather larger } diff --git a/man/updateZRcpp.Rd b/man/updateZRcpp.Rd index 1da6127..561a28e 100644 --- a/man/updateZRcpp.Rd +++ b/man/updateZRcpp.Rd @@ -20,26 +20,26 @@ updateZRcpp(m, p, Theta, Y, W, eta1, eta2, a, rho, max_iter, eps, truncate) \item{eta2}{Equals \eqn{\lambda_2 / rho}} \item{a}{Value added to the diagonal of \eqn{-D'D} so that -the matrix is positive definite, see +the matrix is positive definite, see \code{\link{matrix_A_inner_ADMM}}} \item{rho}{The ADMM's parameter} \item{max_iter}{Maximum number of iterations} -\item{eps}{Stopping criterion. If differences +\item{eps}{Stopping criterion. If differences are smaller than \eqn{\epsilon}, algorithm is halted} -\item{truncate}{Values below \code{truncate} are +\item{truncate}{Values below \code{truncate} are set to \code{0}} } \value{ The estimated vector \eqn{\hat{\beta}} } \description{ -A \code{C} implementation of the \eqn{Z}-update step. We -solve a generalized LASSO problem repeatedly for each of the +A \code{C} implementation of the \eqn{Z}-update step. We +solve a generalized LASSO problem repeatedly for each of the individual edges } \seealso{ diff --git a/man/updateZ_wrapper.Rd b/man/updateZ_wrapper.Rd index f8e210f..a11947f 100644 --- a/man/updateZ_wrapper.Rd +++ b/man/updateZ_wrapper.Rd @@ -33,25 +33,25 @@ updateZ_wrapper( \item{W}{Weight matrix} -\item{eta1}{} +\item{eta1}{TODO} \item{rho_genlasso}{The \eqn{\rho} penalty parameter for the ADMM algorithm} -\item{maxiter_genlasso}{Maximum number of iterations for solving +\item{maxiter_genlasso}{Maximum number of iterations for solving the generalized LASSO problem} -\item{eps_genlasso}{If the relative difference between two update steps is +\item{eps_genlasso}{If the relative difference between two update steps is smaller than \eqn{\epsilon}, the algorithm stops} -\item{truncate_genlasso}{All values of the final \eqn{\hat{\beta}} below +\item{truncate_genlasso}{All values of the final \eqn{\hat{\beta}} below \code{truncate_genlasso} will be set to \code{0}.} } \value{ A list with matrices with the new values of \eqn{Z} } \description{ -A wrapper for the \code{C} function that returns the -updated value of \eqn{Z} for the ADMM given the previously +A wrapper for the \code{C} function that returns the +updated value of \eqn{Z} for the ADMM given the previously updated values of \eqn{\Theta} and \eqn{Y} } \seealso{ diff --git a/man/visnetwork.Rd b/man/visnetwork.Rd index 83bf8f6..21b217d 100644 --- a/man/visnetwork.Rd +++ b/man/visnetwork.Rd @@ -16,27 +16,27 @@ visnetwork( A \code{visNetwork} plot } \description{ -Creates a \code{visNetwork} plot given a list of -nodes and edges. The nodes data frame can be -created with \code{\link{create_nodes_visnetwork}}; -the edges with \code{create_edges_visnetwork}. -In order to highlight edges, you can use +Creates a \code{visNetwork} plot given a list of +nodes and edges. The nodes data frame can be +created with \code{\link{create_nodes_visnetwork}}; +the edges with \code{create_edges_visnetwork}. +In order to highlight edges, you can use \code{\link{set_attributes_to_edges_visnetwork}}. } \examples{ nodes <- CVN::create_nodes_visnetwork(n_nodes = 5, labels = LETTERS[1:5]) -adj_matrix <- matrix(c(0, 1, 0, 1, 0, - 1, 0, 1, 0, 0, - 0, 1, 0, 0, 0, +adj_matrix <- matrix(c(0, 1, 0, 1, 0, + 1, 0, 1, 0, 0, + 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0), ncol = 5) edges <- CVN::create_edges_visnetwork(adj_matrix) -edges <- set_attributes_to_edges_visnetwork(edges, +edges <- set_attributes_to_edges_visnetwork(edges, subset_edges = list(from = c(1, 2), to = c(4, 3)), - width = c(3, .5), + width = c(3, .5), color = c("red", "blue")) CVN::visnetwork(nodes, edges) diff --git a/man/visnetwork_cvn.Rd b/man/visnetwork_cvn.Rd index 57c40f2..c31aa89 100644 --- a/man/visnetwork_cvn.Rd +++ b/man/visnetwork_cvn.Rd @@ -16,12 +16,12 @@ visnetwork_cvn( ) } \arguments{ -\item{cvn}{A \code{cvn} object, see \code{\link{CVN::CVN}} +\item{cvn}{A \code{cvn} object, see \code{\link{CVN::CVN}} or \code{\link{CVN::glasso}}} \item{node_titles}{Vector with title of the nodes (Default: \code{1:p})} -\item{titles}{A list with \code{n_lambda_values} vectors. Each vector is of the +\item{titles}{A list with \code{n_lambda_values} vectors. Each vector is of the lenght \code{m}. Regulates the titles of the graphs (Default: no title)} \item{show_core_graph, width, color}{Show the core graph using the width and colors} @@ -30,6 +30,6 @@ lenght \code{m}. Regulates the titles of the graphs (Default: no title)} List } \description{ -Creates all \code{visNetwork} plots, see \code{\link{CVN::visnetwork}}, +Creates all \code{visNetwork} plots, see \code{\link{CVN::visnetwork}}, for all graphs in a \code{cvn} object } diff --git a/src/CVN.cpp b/src/CVN.cpp index 0a3f6d3..b3e2aa3 100644 --- a/src/CVN.cpp +++ b/src/CVN.cpp @@ -2,332 +2,332 @@ using namespace Rcpp; //' Solving Generalized LASSO with fixed \eqn{\lambda = 1} - //' - //' Solves efficiently the generalized LASSO problem of the form - //' \deqn{ - //' \hat{\beta} = \text{argmin } \frac{1}{2} || y - \beta ||_2^2 + ||D\beta||_1 - //' } - //' where \eqn{\beta} and \eqn{y} are \eqn{m}-dimensional vectors and - //' \eqn{D} is a \eqn{(c \times m)}-matrix where \eqn{c \geq m}. - //' We solve this optimization problem using an adaption of the ADMM - //' algorithm presented in Zhu (2017). - //' - //' @param y The \eqn{y} vector of length \eqn{m} - //' @param W The weight matrix \eqn{W} of dimensions \eqn{m x m} - //' @param m The number of graphs - //' @param eta1 Equals \eqn{\lambda_1 / rho} - //' @param eta2 Equals \eqn{\lambda_2 / rho} - //' @param a Value added to the diagonal of \eqn{-D'D} so that - //' the matrix is positive definite, see - //' \code{\link{matrix_A_inner_ADMM}} - //' @param rho The ADMM's parameter - //' @param max_iter Maximum number of iterations - //' @param eps Stopping criterion. If differences - //' are smaller than \eqn{\epsilon}, algorithm - //' is halted - //' @param truncate Values below \code{truncate} are - //' set to \code{0} - //' - //' @return The estimated vector \eqn{\hat{\beta}} - //' - //' @references - //' Zhu, Y. (2017). An Augmented ADMM Algorithm With Application to the - //' Generalized Lasso Problem. Journal of Computational and Graphical Statistics, - //' 26(1), 195–204. https://doi.org/10.1080/10618600.2015.1114491 - //' - //' @seealso \code{\link{genlasso_wrapper}} - // [[Rcpp::export]] - NumericVector genlassoRcpp(const NumericVector Y, - const NumericMatrix W, - const int m, - const double eta1, - const double eta2, - double a, - const double rho, - const int max_iter, - const double eps, - const double truncate) { - - - // some frequently used constants - a = rho*a ; - const double C = 1 / (1 + a) ; - - // number of rows in matrix D - const int c = (m*m + m) / 2 ; - - /* Compute y to a double array. I did this to make sure that the C compilation - * environment does not have an effect */ - double* y = new double[m + 1] ; - for (int i = 0; i < m; i ++) { - *(y + i) = (double)(Y[i]) ; - } - - /* initialize vectors for beta-update step in the ADMM */ - double *beta_new = new double[m+1]; - double *beta_old = new double[m+1] ; - double *delta = new double[m+1] ; - - for (int i = 0; i < m; i ++) { - *(beta_new + i) = 0; - *(beta_old + i) = 0; - *(delta + i) = 0; - } - - /* initialize vectors for alpha-update step in the ADMM */ - double *alpha_new = new double[c+1] ; - double *alpha_old1 = new double[c+1] ; - double *alpha_old2 = new double[c+1] ; - double *alpha = new double[c+1]; - - for (int i = 0; i < c; i ++) { - *(alpha_new + i) = 0 ; - *(alpha_old1 + i) = 0 ; - *(alpha_old2 + i) = 0 ; - *(alpha + i) = 0 ; - } - - /* Indices used for the alpha-update step */ - int steps[m-1] ; - - *steps = 0 ; - - for (int i = 1; i < (m-1); i ++) { - *(steps + i) = m - i + *(steps + i - 1) ; - } - - int iter = 0 ; // number of iterations - double diff = 0 ; // absolute difference between beta^(k+1) and beta^k - - /* loop until either the max. no. of iterations are reached - * or the difference (diff) is smaller then eps - */ - while (iter < max_iter) { - - /* ------- beta-update step ---------*/ - // alpha = 2*alpha_old1 - alpha_old2 ; - for (int i = 0; i < c; i++) { - *(alpha + i) = 2*(*(alpha_old1 + i)) - *(alpha_old2 + i) ; - } - - // go over all possible pairs (i,j), same as D^T %*% (2 alpha^(k) - alpha^(k-1)) - for (int i = 0; i < m; i++) { - *(delta + i) = eta1 * (*(alpha + i)) ; - - for (int j = i+1; j < m; j++) { - *(delta + i) += eta2* W(i,j)*(*(alpha + m + steps[i] + (j - i) - 1)) ; - } - - for (int j = 0; j < i; j++) { - *(delta + i) -= eta2*W(i,j)*(*(alpha + m + steps[j] - (j - i) - 1)) ; - } - } - - diff = 0 ; - for (int i = 0; i < m; i ++) { - *(beta_new + i) = C*(a*(*(beta_old + i)) + y[i] - *(delta + i)) ; - diff += fabs(*(beta_new + i) - *(beta_old + i)) ; //abs(beta_new[i] - beta_old[i]) ; - } - - // determine whether converged or not - if (diff < eps) { - /* Turn to zero when really close */ - for (int i = 0; i < m; i ++) { - if (fabs(*(beta_new + i)) < truncate) { - *(beta_new + i) = 0 ; - } - } - - // convert results to NumericVector for R - NumericVector res = NumericVector(beta_new, beta_new + m) ; - - // Clean-up ------------ - delete[] beta_old; - delete[] alpha_old1; - delete[] alpha_old2; - delete[] alpha_new; - delete[] alpha; - delete[] y; - delete[] beta_new ; - - return(res); - } - - /* --------- alpha update step ----------- */ - for (int i = 0; i < m; i++) { - *(alpha_new + i) = *(alpha_old1 + i) + rho * eta1 * *(beta_new + i) ; - } - - int k = m; - // go over all unique pairs (i,j) - for (int i = 0; i < m-1; i++) { - for (int j = i+1; j < m; j++) { - *(alpha_new + k) = *(alpha_old1 + k) + rho * eta2 * W(i,j) * (*(beta_new + i) - *(beta_new + j)) ; - k ++; - } - } - - /* Threshold alpha. Must lie in [-1, 1] range */ - for (int i = 0; i < c; i ++) { - if (alpha_new[i] > 1) { - alpha_new[i] = 1 ; - } - if (alpha_new[i] < -1) { - alpha_new[i] = -1 ; - } - } - - /* update beta and alpha for the next iteration step */ - for (int i = 0; i < m; i ++) { - *(beta_old + i) = *(beta_new + i) ; - } - - for (int i = 0; i < c; i ++) { - *(alpha_old2 + i) = *(alpha_old1 + i) ; - *(alpha_old1 + i) = *(alpha_new + i) ; - } - - iter ++; - } - - /* Turn to zero when really close */ - for (int i = 0; i < m; i ++) { - if (fabs(*(beta_new + i)) < truncate) { - *(beta_new + i) = 0 ; - } - } - - // convert results to NumericVector for R - NumericVector res = NumericVector(beta_new, beta_new + m) ; - - // Clean-up ------------ - delete[] beta_old; - delete[] alpha_old1; - delete[] alpha_old2; - delete[] alpha_new; - delete[] alpha; - delete[] y; - delete[] beta_new ; - - return(res); - } +//' +//' Solves efficiently the generalized LASSO problem of the form +//' \deqn{ +//' \hat{\beta} = \text{argmin } \frac{1}{2} || y - \beta ||_2^2 + ||D\beta||_1 +//' } +//' where \eqn{\beta} and \eqn{y} are \eqn{m}-dimensional vectors and +//' \eqn{D} is a \eqn{(c \times m)}-matrix where \eqn{c \geq m}. +//' We solve this optimization problem using an adaption of the ADMM +//' algorithm presented in Zhu (2017). +//' +//' @param y The \eqn{y} vector of length \eqn{m} +//' @param W The weight matrix \eqn{W} of dimensions \eqn{m x m} +//' @param m The number of graphs +//' @param eta1 Equals \eqn{\lambda_1 / rho} +//' @param eta2 Equals \eqn{\lambda_2 / rho} +//' @param a Value added to the diagonal of \eqn{-D'D} so that +//' the matrix is positive definite, see +//' \code{\link{matrix_A_inner_ADMM}} +//' @param rho The ADMM's parameter +//' @param max_iter Maximum number of iterations +//' @param eps Stopping criterion. If differences +//' are smaller than \eqn{\epsilon}, algorithm +//' is halted +//' @param truncate Values below \code{truncate} are +//' set to \code{0} +//' +//' @return The estimated vector \eqn{\hat{\beta}} +//' +//' @references +//' Zhu, Y. (2017). An Augmented ADMM Algorithm With Application to the +//' Generalized Lasso Problem. Journal of Computational and Graphical Statistics, +//' 26(1), 195–204. https://doi.org/10.1080/10618600.2015.1114491 +//' +//' @seealso \code{\link{genlasso_wrapper}} +// [[Rcpp::export]] +NumericVector genlassoRcpp(const NumericVector Y, + const NumericMatrix W, + const int m, + const double eta1, + const double eta2, + double a, + const double rho, + const int max_iter, + const double eps, + const double truncate) { + + +// some frequently used constants +a = rho*a ; +const double C = 1 / (1 + a) ; + +// number of rows in matrix D +const int c = (m*m + m) / 2 ; + +/* Compute y to a double array. I did this to make sure that the C compilation + * environment does not have an effect */ +double* y = new double[m + 1] ; +for (int i = 0; i < m; i ++) { + *(y + i) = (double)(Y[i]) ; +} + +/* initialize vectors for beta-update step in the ADMM */ +double *beta_new = new double[m+1]; +double *beta_old = new double[m+1] ; +double *delta = new double[m+1] ; + +for (int i = 0; i < m; i ++) { + *(beta_new + i) = 0; + *(beta_old + i) = 0; + *(delta + i) = 0; +} + +/* initialize vectors for alpha-update step in the ADMM */ +double *alpha_new = new double[c+1] ; +double *alpha_old1 = new double[c+1] ; +double *alpha_old2 = new double[c+1] ; +double *alpha = new double[c+1]; + +for (int i = 0; i < c; i ++) { + *(alpha_new + i) = 0 ; + *(alpha_old1 + i) = 0 ; + *(alpha_old2 + i) = 0 ; + *(alpha + i) = 0 ; +} + +/* Indices used for the alpha-update step */ +int steps[m-1] ; + +*steps = 0 ; + +for (int i = 1; i < (m-1); i ++) { + *(steps + i) = m - i + *(steps + i - 1) ; +} + +int iter = 0 ; // number of iterations +double diff = 0 ; // absolute difference between beta^(k+1) and beta^k + +/* loop until either the max. no. of iterations are reached + * or the difference (diff) is smaller then eps + */ +while (iter < max_iter) { + + /* ------- beta-update step ---------*/ + // alpha = 2*alpha_old1 - alpha_old2 ; + for (int i = 0; i < c; i++) { + *(alpha + i) = 2*(*(alpha_old1 + i)) - *(alpha_old2 + i) ; + } + + // go over all possible pairs (i,j), same as D^T %*% (2 alpha^(k) - alpha^(k-1)) + for (int i = 0; i < m; i++) { + *(delta + i) = eta1 * (*(alpha + i)) ; + + for (int j = i+1; j < m; j++) { + *(delta + i) += eta2* W(i,j)*(*(alpha + m + steps[i] + (j - i) - 1)) ; + } + + for (int j = 0; j < i; j++) { + *(delta + i) -= eta2*W(i,j)*(*(alpha + m + steps[j] - (j - i) - 1)) ; + } + } + + diff = 0 ; + for (int i = 0; i < m; i ++) { + *(beta_new + i) = C*(a*(*(beta_old + i)) + y[i] - *(delta + i)) ; + diff += fabs(*(beta_new + i) - *(beta_old + i)) ; //abs(beta_new[i] - beta_old[i]) ; + } + + // determine whether converged or not + if (diff < eps) { + /* Turn to zero when really close */ + for (int i = 0; i < m; i ++) { + if (fabs(*(beta_new + i)) < truncate) { + *(beta_new + i) = 0 ; + } + } + + // convert results to NumericVector for R + NumericVector res = NumericVector(beta_new, beta_new + m) ; + + // Clean-up ------------ + delete[] beta_old; + delete[] alpha_old1; + delete[] alpha_old2; + delete[] alpha_new; + delete[] alpha; + delete[] y; + delete[] beta_new ; + + return(res); + } + + /* --------- alpha update step ----------- */ + for (int i = 0; i < m; i++) { + *(alpha_new + i) = *(alpha_old1 + i) + rho * eta1 * *(beta_new + i) ; + } + + int k = m; + // go over all unique pairs (i,j) + for (int i = 0; i < m-1; i++) { + for (int j = i+1; j < m; j++) { + *(alpha_new + k) = *(alpha_old1 + k) + rho * eta2 * W(i,j) * (*(beta_new + i) - *(beta_new + j)) ; + k ++; + } + } + + /* Threshold alpha. Must lie in [-1, 1] range */ + for (int i = 0; i < c; i ++) { + if (alpha_new[i] > 1) { + alpha_new[i] = 1 ; + } + if (alpha_new[i] < -1) { + alpha_new[i] = -1 ; + } + } + + /* update beta and alpha for the next iteration step */ + for (int i = 0; i < m; i ++) { + *(beta_old + i) = *(beta_new + i) ; + } + + for (int i = 0; i < c; i ++) { + *(alpha_old2 + i) = *(alpha_old1 + i) ; + *(alpha_old1 + i) = *(alpha_new + i) ; + } + + iter ++; +} + +/* Turn to zero when really close */ +for (int i = 0; i < m; i ++) { + if (fabs(*(beta_new + i)) < truncate) { + *(beta_new + i) = 0 ; + } +} + +// convert results to NumericVector for R +NumericVector res = NumericVector(beta_new, beta_new + m) ; + +// Clean-up ------------ +delete[] beta_old; +delete[] alpha_old1; +delete[] alpha_old2; +delete[] alpha_new; +delete[] alpha; +delete[] y; +delete[] beta_new ; + +return(res); +} //' The \eqn{Z}-update Step -//' -//' A \code{C} implementation of the \eqn{Z}-update step. We -//' solve a generalized LASSO problem repeatedly for each of the -//' individual edges -//' -//' @param m The number of graphs +//' +//' A \code{C} implementation of the \eqn{Z}-update step. We +//' solve a generalized LASSO problem repeatedly for each of the +//' individual edges +//' +//' @param m The number of graphs //' @param p The number of variables //' @param Theta A list of matrices with the \eqn{\Theta}-matrices //' @param Y A list of matrices with the \eqn{Y}-matrices -//' @param eta1 Equals \eqn{\lambda_1 / rho} -//' @param eta2 Equals \eqn{\lambda_2 / rho} +//' @param eta1 Equals \eqn{\lambda_1 / rho} +//' @param eta2 Equals \eqn{\lambda_2 / rho} //' @param a Value added to the diagonal of \eqn{-D'D} so that -//' the matrix is positive definite, see +//' the matrix is positive definite, see //' \code{\link{matrix_A_inner_ADMM}} //' @param rho The ADMM's parameter //' @param max_iter Maximum number of iterations -//' @param eps Stopping criterion. If differences +//' @param eps Stopping criterion. If differences //' are smaller than \eqn{\epsilon}, algorithm //' is halted -//' @param truncate Values below \code{truncate} are +//' @param truncate Values below \code{truncate} are //' set to \code{0} //' //' @return The estimated vector \eqn{\hat{\beta}} -//' +//' //' @seealso \code{\link{updateZ_wrapper}} // [[Rcpp::export]] -Rcpp::ListMatrix updateZRcpp(const int m, - const int p, - Rcpp::ListMatrix Theta, +Rcpp::ListMatrix updateZRcpp(const int m, + const int p, + Rcpp::ListMatrix Theta, Rcpp::ListMatrix Y, - const Rcpp::NumericMatrix& W, - const double eta1, - const double eta2, - const double a, - const double rho, + const Rcpp::NumericMatrix& W, + const double eta1, + const double eta2, + const double a, + const double rho, const int max_iter, - const double eps, - const double truncate) { - + const double eps, + const double truncate) { + // indices - int i,j,k,l ; - + int i,j,k,l ; + /* Initialize variables ----------- */ - - /* The y-vector and the resulting beta values are - stored in a vector and a matrix. The rows of beta + + /* The y-vector and the resulting beta values are + stored in a vector and a matrix. The rows of beta represent the edges. */ - Rcpp::NumericVector y (m) ; - Rcpp::NumericMatrix beta (p*(p-1)/2, m) ; - - // Final results will be stored here - Rcpp::ListMatrix Z (m); - + Rcpp::NumericVector y (m) ; + Rcpp::NumericMatrix beta (p*(p-1)/2, m) ; + + // Final results will be stored here + Rcpp::ListMatrix Z (m); + // set the diagonal of Z - for (k = 0; k < m; k ++) { - Rcpp::NumericMatrix A = Theta(k,0) ; - Rcpp::NumericMatrix B = Y(k,0) ; - Rcpp::NumericMatrix C (p,p) ; - - for (i = 0; i < p; i ++) { - for (j = 0; j < p; j ++) { - C(i,j) = 0; - } + for (k = 0; k < m; k ++) { + Rcpp::NumericMatrix A = Theta(k,0) ; + Rcpp::NumericMatrix B = Y(k,0) ; + Rcpp::NumericMatrix C (p,p) ; + + for (i = 0; i < p; i ++) { + for (j = 0; j < p; j ++) { + C(i,j) = 0; + } } - - for (i = 0; i < p; i ++) { + + for (i = 0; i < p; i ++) { C(i,i) = A(i,i) + B(i,i) ; } - Z(k,0) = clone(C) ; + Z(k,0) = clone(C) ; } - + /* Compute betas -------------- */ l = 0 ; // the index for the edge - - // go over all edges - for (i = 0; i < (p-1); i ++) { + + // go over all edges + for (i = 0; i < (p-1); i ++) { for (j = i+1; j < p; j ++) { - + // go over the different graphs - for (k = 0; k < m; k ++) { + for (k = 0; k < m; k ++) { // get the matrix Theta and Y and store them in A and B - Rcpp::NumericMatrix A = Theta(k,0) ; - Rcpp::NumericMatrix B = Y(k,0) ; - y[k] = A(i,j) + B(i,j) ; + Rcpp::NumericMatrix A = Theta(k,0) ; + Rcpp::NumericMatrix B = Y(k,0) ; + y[k] = A(i,j) + B(i,j) ; } - + // determine the beta-vector - Rcpp::DoubleVector b = genlassoRcpp(y, W, m, eta1, eta2, a, rho, max_iter, eps, truncate) ; - + Rcpp::DoubleVector b = genlassoRcpp(y, W, m, eta1, eta2, a, rho, max_iter, eps, truncate) ; + // store the results in the beta matrix - for (k = 0; k < m; k ++) { - beta(l, k) = b[k] ; + for (k = 0; k < m; k ++) { + beta(l, k) = b[k] ; } - + l ++; // update the edge index } } - + // set the entries of Z - for (k = 0; k < m; k ++) { - Rcpp::NumericMatrix A = Z(k,0) ; - - l = 0; - for (i = 0; i < (p-1); i ++) { - for (j = i+1; j < p; j ++) { - A(i,j) = beta(l, k) ; - A(j,i) = beta(l, k) ; - - l ++; + for (k = 0; k < m; k ++) { + Rcpp::NumericMatrix A = Z(k,0) ; + + l = 0; + for (i = 0; i < (p-1); i ++) { + for (j = i+1; j < p; j ++) { + A(i,j) = beta(l, k) ; + A(j,i) = beta(l, k) ; + + l ++; } } - - Z(k,0) = clone(A) ; + + Z(k,0) = clone(A) ; } - - return(Z) ; + + return(Z) ; } diff --git a/src/Makevars b/src/Makevars deleted file mode 100644 index 12d7d66..0000000 --- a/src/Makevars +++ /dev/null @@ -1,14 +0,0 @@ - -## With R 3.1.0 or later, you can uncomment the following line to tell R to -## enable compilation with C++11 (where available) -## -## Also, OpenMP support in Armadillo prefers C++11 support. However, for wider -## availability of the package we do not yet enforce this here. It is however -## recommended for client packages to set it. -## -## And with R 3.4.0, and RcppArmadillo 0.7.960.*, we turn C++11 on as OpenMP -## support within Armadillo prefers / requires it -CXX_STD = CXX11 - -#PKG_CXXFLAGS = $(SHLIB_OPENMP_CXXFLAGS) -#PKG_LIBS = $(SHLIB_OPENMP_CFLAGS) $(LAPACK_LIBS) $(BLAS_LIBS) $(FLIBS) diff --git a/src/Makevars.win b/src/Makevars.win deleted file mode 100644 index 12d7d66..0000000 --- a/src/Makevars.win +++ /dev/null @@ -1,14 +0,0 @@ - -## With R 3.1.0 or later, you can uncomment the following line to tell R to -## enable compilation with C++11 (where available) -## -## Also, OpenMP support in Armadillo prefers C++11 support. However, for wider -## availability of the package we do not yet enforce this here. It is however -## recommended for client packages to set it. -## -## And with R 3.4.0, and RcppArmadillo 0.7.960.*, we turn C++11 on as OpenMP -## support within Armadillo prefers / requires it -CXX_STD = CXX11 - -#PKG_CXXFLAGS = $(SHLIB_OPENMP_CXXFLAGS) -#PKG_LIBS = $(SHLIB_OPENMP_CFLAGS) $(LAPACK_LIBS) $(BLAS_LIBS) $(FLIBS)