improve docker instruction

Author: refraction-ray

docker/README.md

@@ -1,19 +1,25 @@
 Run the following command to build the docker for tensorcircuit at parent path:
 
 ```bash
-sudo docker build . -f docker/Dockerfile -t tc-dev
+sudo docker build . -f docker/Dockerfile -t tensorcircuit
 ```
 
-Run the docker container by the following command:
+One can also pull the [official image](https://hub.docker.com/repository/docker/tensorcircuit/tensorcircuit) from DockerHub as 
 
 ```bash
-sudo docker run -it --network host -e LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.0/targets/x86_64-linux/lib -e PYTHONPATH=/app -v "$(pwd)":/app  --gpus all tc-dev
+sudo docker pull tensorcircuit/tensorcircuit
 ```
 
-Or directly use the script provided in this dir:
+
+Run the docker container by the following command:
 
 ```bash
-./docker_run.sh
+sudo docker run -it --network host --gpus all tensorcircuit
+
+# if one also wants mount local source code, also add args `-v "$(pwd)":/app`
+
+# for old dockerfile with no runtime env setting
+# sudo docker run -it --network host -e LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.0/targets/x86_64-linux/lib -e PYTHONPATH=/app -v "$(pwd)":/app  --gpus all tensorcircuit
 ```
 
 `export TF_CPP_MIN_LOG_LEVEL=3` maybe necessary since jax suprisingly frequently complain about ptxas version problem. And `export CUDA_VISIBLE_DEVICES=-1` if you want to test only on CPU.