Reduced the number of docker images by one pybamm:latest

.github/workflows/docker.yml

@@ -7,15 +7,11 @@ on:
     - develop
 
 jobs:
-  build_docker_images:
+  build_docker_image:
     # This workflow is only of value to PyBaMM and would always be skipped in forks
     if: github.repository_owner == 'pybamm-team'
-    name: Image (${{ matrix.build-args }})
+    name: Build image
     runs-on: ubuntu-latest
-    strategy:
-      matrix:
-        build-args: ["No solvers", "JAX", "IDAKLU", "ALL"]
-      fail-fast: true
 
     steps:
       - name: Checkout
@@ -33,27 +29,12 @@ jobs:
             username: ${{ secrets.DOCKERHUB_USERNAME }}
             password: ${{ secrets.DOCKERHUB_TOKEN }}
 
-      - name: Create tags for Docker images based on build-time arguments
-        id: tags
-        run: |
-          if [ "${{ matrix.build-args }}" = "No solvers" ]; then
-            echo "tag=latest" >> "$GITHUB_OUTPUT"
-          elif [ "${{ matrix.build-args }}" = "JAX" ]; then
-            echo "tag=jax" >> "$GITHUB_OUTPUT"
-          elif [ "${{ matrix.build-args }}" = "ODES" ]; then
-            echo "tag=odes" >> "$GITHUB_OUTPUT"
-          elif [ "${{ matrix.build-args }}" = "IDAKLU" ]; then
-            echo "tag=idaklu" >> "$GITHUB_OUTPUT"
-          elif [ "${{ matrix.build-args }}" = "ALL" ]; then
-            echo "tag=all" >> "$GITHUB_OUTPUT"
-          fi
-
-      - name: Build and push Docker image to Docker Hub (${{ matrix.build-args }})
+      - name: Build and push Docker image to Docker Hub
         uses: docker/build-push-action@v5
         with:
           context: .
           file: scripts/Dockerfile
-          tags: pybamm/pybamm:${{ steps.tags.outputs.tag }}
+          tags: pybamm/pybamm:latest
           push: true
           platforms: linux/amd64, linux/arm64
 

CHANGELOG.md

@@ -32,6 +32,7 @@
 
 ## Breaking changes
 
+- Removed multiple Docker images. Here on, a single Docker image tagged `pybamm/pybamm:latest` will be provided with both solvers (`IDAKLU` and `JAX`) pre-installed. ([#3992](https://github.com/pybamm-team/PyBaMM/pull/3992))
 - Removed support for Python 3.8 ([#3961](https://github.com/pybamm-team/PyBaMM/pull/3961))
 - Renamed "ocp_soc_0_dimensional" to "ocp_soc_0" and "ocp_soc_100_dimensional" to "ocp_soc_100" ([#3942](https://github.com/pybamm-team/PyBaMM/pull/3942))
 - The ODES solver was removed due to compatibility issues. Users should use IDAKLU, Casadi, or JAX instead. ([#3932](https://github.com/pybamm-team/PyBaMM/pull/3932))

docs/source/user_guide/installation/install-from-docker.rst

@@ -20,29 +20,11 @@ Pulling the Docker image
 
 Use the following command to pull the PyBaMM Docker image from Docker Hub:
 
-.. tab:: No optional solver
 
-      .. code:: bash
-
-            docker pull pybamm/pybamm:latest
-
-.. tab:: JAX solver
-
-      .. code:: bash
-
-            docker pull pybamm/pybamm:jax
-
-.. tab:: IDAKLU solver
-
-      .. code:: bash
-
-            docker pull pybamm/pybamm:idaklu
-
-.. tab:: All solvers
+.. code:: bash
 
-      .. code:: bash
+    docker pull pybamm/pybamm
 
-            docker pull pybamm/pybamm:all
 
 Running the Docker container
 ----------------------------
@@ -51,35 +33,11 @@ Once you have pulled the Docker image, you can run a Docker container with the P
 
 1. In your terminal, use the following command to start a Docker container from the pulled image:
 
-.. tab:: Basic
-
-      .. code:: bash
-
-            docker run -it pybamm/pybamm:latest
-
-.. tab:: ODES Solver
 
-      .. code:: bash
-
-            docker run -it pybamm/pybamm:odes
-
-.. tab:: JAX Solver
-
-      .. code:: bash
-
-            docker run -it pybamm/pybamm:jax
-
-.. tab:: IDAKLU Solver
-
-      .. code:: bash
-
-            docker run -it pybamm/pybamm:idaklu
-
-.. tab:: All Solver
+.. code:: bash
 
-      .. code:: bash
+    docker run -it pybamm/pybamm
 
-            docker run -it pybamm/pybamm:all
 
 2. You will now be inside the Docker container's shell. You can use PyBaMM and its dependencies as if you were in a virtual environment.
 
@@ -138,52 +96,9 @@ If you want to build the PyBaMM Docker image locally from the PyBaMM source code
 
       conda activate pybamm
 
-Building Docker images with optional arguments
-----------------------------------------------
-
-When building the PyBaMM Docker images locally, you have the option to include specific solvers by using optional arguments. These solvers include:
-
-- ``IDAKLU``: For IDA solver provided by the SUNDIALS plus KLU.
-- ``JAX``: For Jax solver.
-- ``ALL``: For all the above solvers.
-
-To build the Docker images with optional arguments, you can follow these steps for each solver:
-
-.. tab:: JAX solver
-
-      .. code-block:: bash
-
-            docker build -t pybamm:jax -f scripts/Dockerfile --build-arg JAX=true .
-
-.. tab:: IDAKLU solver
-
-      .. code-block:: bash
-
-            docker build -t pybamm:idaklu -f scripts/Dockerfile --build-arg IDAKLU=true .
-
-.. tab:: All solvers
-
-      .. code-block:: bash
-
-            docker build -t pybamm:all -f scripts/Dockerfile --build-arg ALL=true .
-
-After building the Docker images with the desired solvers, use the ``docker run`` command followed by the desired image name. For example, to run a container from the image built with all optional solvers:
-
-.. code-block:: bash
-
-      docker run -it pybamm:all
-
-Activate PyBaMM development environment inside docker container using:
-
-.. code-block:: bash
-
-      conda activate pybamm
-
-If you want to exit the Docker container's shell, you can simply type:
-
-.. code-block:: bash
+.. note::
 
-      exit
+  PyBaMM's Docker image comes with all available solvers by default. These solvers include ``IDAKLU`` IDAS solver provided by the SUNDIALS linked with SuiteSparse's KLU and the ``JAX`` solver.
 
 
 Using Git inside a running Docker container

scripts/Dockerfile

@@ -26,35 +26,12 @@ RUN conda init --all
 SHELL ["conda", "run", "-n", "pybamm", "/bin/bash", "-c"]
 RUN conda install -y pip
 
-ARG IDAKLU
-ARG JAX
-ARG ALL
-
 RUN pip install --upgrade --user pip setuptools wheel wget
 RUN pip install cmake
 
-RUN if [ "$IDAKLU" = "true" ]; then \
-    python scripts/install_KLU_Sundials.py && \
-    rm -rf pybind11 && \
-    git clone https://github.com/pybind/pybind11.git && \
-    pip install --user -e ".[all,dev,docs]"; \
-    fi
-
-RUN if [ "$JAX" = "true" ]; then \
-    pip install --user -e ".[all,dev,docs,jax]"; \
-    fi
-
-RUN if [ "$ALL" = "true" ]; then \
-    python scripts/install_KLU_Sundials.py && \
+RUN python scripts/install_KLU_Sundials.py && \
     rm -rf pybind11 && \
     git clone https://github.com/pybind/pybind11.git && \
-    pip install --user -e ".[all,dev,docs,jax]"; \
-    fi
-
-RUN if [ -z "$IDAKLU" ] \
-    && [ -z "$JAX" ] \
-    && [ -z "$ALL" ]; then \
-    pip install --user -e ".[all,dev,docs]"; \
-    fi
+    pip install --user -e ".[all,dev,docs,jax]";
 
 ENTRYPOINT ["/bin/bash"]