Merge pull request tensorflow#480 from quantumlib/cuquantum-install-doc

Link to cuQuantum install guide

docs/tutorials/gcp_gpu.md

@@ -178,7 +178,9 @@ After a moment, you should see a result that looks similar to the following.
 ### Optional: Use the NVIDIA cuQuantum SDK
 
 If you have the [NVIDIA cuQuantum SDK](https://developer.nvidia.com/cuquantum-sdk)
-installed, you can use it with this tutorial. Before building qsim in step 5,
+installed (instructions are provided
+[here](https://docs.nvidia.com/cuda/cuquantum/custatevec/html/getting_started.html#installation-and-compilation)),
+you can use it with this tutorial. Before building qsim in step 5,
 set the `CUQUANTUM_DIR` environment variable from the command line:
 
 ```bash