From c0b35d656304afe0f1c36c2150fe4a438a0f4343 Mon Sep 17 00:00:00 2001
From: Jintao Li <j.t.li@i4ai.org>
Date: Thu, 12 Sep 2024 15:13:38 +0800
Subject: [PATCH] Create readme.md

---
 RGNN_RL_Docker/readme.md | 58 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 58 insertions(+)
 create mode 100644 RGNN_RL_Docker/readme.md

diff --git a/RGNN_RL_Docker/readme.md b/RGNN_RL_Docker/readme.md
new file mode 100644
index 0000000..c2d0d42
--- /dev/null
+++ b/RGNN_RL_Docker/readme.md
@@ -0,0 +1,58 @@
+## Running the Simulation
+
+There are three ways to run the simulation:
+
+1. **Using Docker with Local Setup**  
+   The `Dockerfile` and environment configuration file `environment.yml` are included in the project. You can build your own Docker image and container with the following steps:
+
+   - Open a terminal in the `RGNN_RL` directory and build the Docker image (ensure the `.` at the end is included):
+     ```bash
+     docker build -t analoggym_rgcn .
+     ```
+
+   - After the image is built, create a Docker container with the following command. The `-v` flag links the local path (`/xxx/rgcn_rl`) to the container's internal path (`/app`), allowing real-time file modification. The `-it` flag opens an interactive terminal:
+     ```bash
+     docker run -v /xxx/rgnn_rl:/app -it analoggym_rgcn /bin/bash
+     ```
+
+   - Inside the container, check if the `analoggym-env` is activated. If not, activate it with:
+     ```bash
+     conda activate analoggym-env
+     ```
+
+2. **Using Pre-Built Docker Images from Docker Hub**  
+   A quicker method is to pull the pre-built Docker image from Docker Hub and run the simulation directly. Use the following command to pull the image:
+   ```bash
+   docker pull chenzhenxin/analoggym_rgcn:latest
+   ```
+   After pulling, refer to the steps in Method 1 to create a container and run the simulation.
+
+3. **Running Locally for Precise Simulation Results**  
+   Docker might not resolve ngspice version issues, so for precise results, you can run the simulation locally. First, create a conda virtual environment and install the required packages:
+   ```bash
+   conda install -c conda-forge ngspice
+   conda install pytorch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 cpuonly -c pytorch
+   pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.13.1+cpu.html
+   pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-1.13.1+cpu.html
+   pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-1.13.1+cpu.html
+   pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.13.1+cpu.html
+   pip install torch-geometric
+   pip install gymnasium
+   pip install tabulate
+   pip install numpy
+   pip install matplotlib
+   pip install lPython
+   ```
+
+   For details on resolving ngspice version issues, refer to the Supplementary Information. Once the environment is set up, you can run:
+   ```bash
+   python main_AMP.py
+   ```
+   or
+   ```bash
+   python main_LDO.py
+   ```
+   to start the simulation.
+```
+
+This version is more concise, polished, and suitable for a README file. It maintains a professional tone while clearly explaining the steps.