Dev

.github/workflows/build_dist.yml

@@ -26,13 +26,16 @@ jobs:
       with:
         python-version: "3.10"
 
+    - name: Install uv
+      uses: astral-sh/setup-uv@v6
+      with:
+        version: "0.9.2"
+        python-version: "3.10"
+        enable-cache: true
     - name: Install dependencies
       run: |
-        python -m venv ./venv
-        . ./venv/bin/activate
-        python -m pip install --upgrade pip
-        pip install poetry
-        poetry install --with=dev --no-interaction --no-ansi
+        uv venv .venv
+        uv sync --locked --dev --extra cpu
 
     - name: Gather new package version
       id: version
@@ -45,14 +48,12 @@ jobs:
 
     - name: Bump package version in pyproject.toml
       run: |
-        . ./venv/bin/activate
-        poetry version ${{steps.version.outputs.new_tag}}
+        uv version ${{steps.version.outputs.new_tag}}
 
     - name: Build dist
       run: |
-        . ./venv/bin/activate
-        python -m build --sdist --wheel --no-isolation --outdir dist/ .
-        twine check dist/*
+        uv run python -m build --sdist --wheel --no-isolation --outdir dist/ .
+        uv run twine check dist/*
 
     - name: Commit updated pyproject.toml
       run: |

.github/workflows/docs.yml

@@ -17,21 +17,23 @@ jobs:
       with:
         python-version: "3.10"
 
+    - name: Install uv
+      uses: astral-sh/setup-uv@v6
+      with:
+        version: "0.9.2"
+        python-version: "3.10"
+        enable-cache: true
     - name: Install dependencies
       run: |
-        python -m venv ./venv
-        . ./venv/bin/activate
-        python -m pip install --upgrade pip
-        pip install poetry
-        poetry install --with=docs --no-interaction --no-ansi
+        uv venv .venv
+        uv sync --locked --dev --group docs --extra cpu
 
     - name: Build the sphinx docs
       run: |
-        . ./venv/bin/activate
-        sphinx-apidoc -f -o ./sphinx/source ./src/bolightningpipeline
-        make -C sphinx clean
-        python sphinx/clean_html_files.py
-        make -C sphinx html
+        uv run sphinx-apidoc -f -o ./sphinx/source ./src/bolightningpipeline
+        uv run make -C sphinx clean
+        uv run python sphinx/clean_html_files.py
+        uv run make -C sphinx html
         touch sphinx/build/html/.nojekyll
 
     - name: Checkout gh-pages branch
@@ -41,9 +43,8 @@ jobs:
 
     - name: Copy build files to docs folder
       run: |
-        . ./venv/bin/activate
         cp -a sphinx/build/html/. docs/
-        python sphinx/make_html_files_list.py
+        uv run python sphinx/make_html_files_list.py
         rm -rf sphinx/build
 
     - name: Commit to gh-pages branch

.github/workflows/tests.yml

@@ -27,31 +27,30 @@ jobs:
       uses: actions/setup-python@v3
       with:
         python-version: ${{ matrix.python-version }}
+    - name: Install uv
+      uses: astral-sh/setup-uv@v6
+      with:
+        version: "0.9.2"
+        python-version: ${{ matrix.python-version }}
+        enable-cache: true
     - name: Install dependencies
       run: |
-        python -m venv ./venv
-        . ./venv/bin/activate
-        python -m pip install --upgrade pip
-        pip install poetry
-        poetry install --with=dev --no-interaction --no-ansi
+        uv venv .venv
+        uv sync --locked --dev --extra cpu
     - name: Test Linting
       run: |
-        . ./venv/bin/activate
-        black src tests --check --diff
-        isort src tests --check-only --diff
+        uv run black src tests --check --diff
+        uv run isort src tests --check-only --diff
     - name: Test Typing
       run: |
-        . ./venv/bin/activate
-        mypy src tests
+        uv run mypy src tests
     - name: Test Notebooks
       run: |
-        . ./venv/bin/activate
-        pytest --nbmake notebooks -n=auto --nbmake-kernel=python3 --nbmake-timeout=600  # 10 minutes timeout
+        uv run pytest --nbmake notebooks -n=auto --nbmake-kernel=python3 --nbmake-timeout=600  # 10 minutes timeout
 
     - name: Test Unittests with pytest
       run: |
-        . ./venv/bin/activate
-        pytest tests -n=auto --cov=src --cov-report="xml:tests/.tmp/coverage.xml" --cov-report=term-missing --durations=10
+        uv run pytest tests -n=auto --cov=src --cov-report="xml:tests/.tmp/coverage.xml" --cov-report=term-missing --durations=10
 
     - name: Code Coverage
       uses: orgoro/coverage@v3.2
@@ -65,9 +64,8 @@ jobs:
     - name: Test Build
       if: ${{ matrix.python-version == '3.10' }}
       run: |
-        . ./venv/bin/activate
-        python -m build --sdist --wheel --no-isolation --outdir dist/ .
-        twine check dist/*
+        uv run python -m build --sdist --wheel --no-isolation --outdir dist/ .
+        uv run twine check dist/*
 
   Run-tests-on-Windows:
     name: Run tests on Windows-latest
@@ -82,14 +80,16 @@ jobs:
       uses: actions/setup-python@v3
       with:
         python-version: ${{ matrix.python-version }}
+    - name: Install uv
+      uses: astral-sh/setup-uv@v6
+      with:
+        version: "0.9.2"
+        python-version: ${{ matrix.python-version }}
+        enable-cache: true
     - name: Install dependencies
       run: |
-        python -m venv ./venv
-        . ./venv/Scripts/activate
-        python -m pip install --upgrade pip
-        pip install poetry
-        poetry install --with=dev --no-interaction --no-ansi
+        uv venv .venv
+        uv sync --locked --dev --extra cpu
     - name: Test Unittests with pytest
       run: |
-        . ./venv/Scripts/activate
-        pytest tests -n=auto
+        uv run pytest tests -n=auto

.gitignore

@@ -303,3 +303,5 @@ tmp_report_dir
 .tmp_report.json
 coverage.json
 /notebooks/Digits2D
+/notebooks/Cifar10
+/notebooks/data

README.md

@@ -30,29 +30,32 @@ With `python` and `pip` installed,
 pip install git+https://github.com/MatchCake/MatchCake-Opt
 ```
 
-With `poetry` installed,
+With `uv` installed,
 ```bash
-poetry add "git+https://github.com/MatchCake/MatchCake-Opt"
+uv add "git+https://github.com/MatchCake/MatchCake-Opt"
 ```
 
+To install the package with cu128 (CUDA), add `--extra cu128` to the installation commands above.
+
 
 ### For developers
 
 With `python` and `pip` installed, run the following commands to install the dependencies:
 ```bash
 python -m venv .venv
 source .venv/bin/activate
-pip install poetry
-python -m poetry install
+pip install uv
+uv sync --dev
 ```
 
-With `poetry` installed, run the following commands to install the dependencies:
+With `uv` installed, run the following commands to install the dependencies:
 ```bash
-python -m venv .venv
-source .venv/bin/activate
-poetry install
+uv venv .venv
+uv sync --dev
 ```
 
+If you'd like to contribute to this repository, please do so by submitting pull requests to the `dev` branch. Thank you!
+
 ## Quick Usage Exemple
 
 ```python
@@ -78,7 +81,7 @@ class LinearNN(ClassificationModel):
         RangeParameterConfig(
             name="n_neurons",
             parameter_type="int",
-            bounds=(4, 16),
+            bounds=(4, 2048),
         ),
     ]
 
@@ -92,7 +95,7 @@ class LinearNN(ClassificationModel):
     ):
         super().__init__(input_shape=input_shape, output_shape=output_shape, learning_rate=learning_rate, **kwargs)
         self.save_hyperparameters("learning_rate", "n_neurons")
-        self.nn = torch.Sequential(
+        self.nn = torch.nn.Sequential(
             torch.nn.Flatten(),
             torch.nn.LazyLinear(n_neurons), 
             torch.nn.ReLU(), 
@@ -106,7 +109,7 @@ class LinearNN(ClassificationModel):
     def output_size(self):
         return int(np.prod(self.output_shape))
 
-datamodule = DataModule.from_dataset_name("Digits2D")
+datamodule = DataModule.from_dataset_name("Digits2D", fold_id=0)
 automl_pipeline = AutoMLPipeline(model_cls=LinearNN, datamodule=datamodule)
 automl_pipeline.run()
 lt_pipeline, metrics = automl_pipeline.run_best_pipeline()

notebooks/automl_pipeline_tutorial.ipynb

@@ -24,6 +24,8 @@
     "from pathlib import Path\n",
     "from typing import Optional\n",
     "import json\n",
+    "import warnings\n",
+    "warnings.filterwarnings(\"ignore\")\n",
     "\n",
     "import numpy as np\n",
     "import pennylane as qml\n",
@@ -53,6 +55,18 @@
    "cell_type": "code",
    "source": [
     "class NIFCNN(ClassificationModel):\n",
+    "    \"\"\"\n",
+    "    Non-Interacting fermions classifier through CNN hamiltonian encoding. The flow of information\n",
+    "    in the model goes as follows.\n",
+    "\n",
+    "    1. CNN(X_{bchw}) -> X_{be}: Build the embeddings with the CNN encoder\n",
+    "    2. Linear(X_{be}) -> W_{bn}: Build the local fields weights from the embeddings\n",
+    "    3. Linear(X_{be}) -> W_{bt}: Build the ZZ couplings weights from the embeddings\n",
+    "    - We now have the hamiltonian: H(X_{be}) = W_{bn} Z_{n} + W_{bt} ZZ_{t}\n",
+    "    4. P_{knu}, P_{ktv}: We compute the probabilities to get the eigenstates of Z and ZZ\n",
+    "    5. E_{k} = W_{bn} P_{knu} Lambda_{u} + W_{bt} P_{ktv} Lambda_{v}: We now compute the expected values with Lambda_{u} and Lambda_{v} the eigenvalues of Z and ZZ respectively.\n",
+    "    6. Softmax(E_{k}) -> Y_{k}: Finally, we apply a softmax to get the probabilities of each class.\n",
+    "    \"\"\"\n",
     "    MODEL_NAME = \"NIFCNN\"\n",
     "    DEFAULT_N_QUBITS = 16\n",
     "    DEFAULT_LEARNING_RATE = 2e-4\n",
@@ -68,7 +82,7 @@
     "        RangeParameterConfig(\n",
     "            name=\"n_qubits\",\n",
     "            parameter_type=\"int\",\n",
-    "            bounds=(4, 16),\n",
+    "            bounds=(4, 92),\n",
     "        ),\n",
     "        ChoiceParameterConfig(\n",
     "            name=\"encoder_output_activation\",\n",
@@ -118,21 +132,13 @@
     "            torch.nn.LazyLinear(self._hamiltonian_n_params),\n",
     "            getattr(torch.nn, encoder_output_activation)()\n",
     "        )\n",
-    "        self.xx_body_couplings_head = torch.nn.Sequential(\n",
-    "            torch.nn.Flatten(),\n",
-    "            torch.nn.LazyLinear(self._hamiltonian_n_params),\n",
-    "            getattr(torch.nn, encoder_output_activation)()\n",
-    "        )\n",
     "\n",
     "        self.zz_body_coupling_weights = torch.nn.Parameter(torch.randn(self._hamiltonian_n_params), requires_grad=True)\n",
-    "        self.xx_body_coupling_weights = torch.nn.Parameter(torch.randn(self._hamiltonian_n_params), requires_grad=True)\n",
     "\n",
     "        self.local_fields_op_eigvals = torch.nn.Parameter(torch.from_numpy(np.array([1.0, -1.0])).float(),\n",
     "                                                          requires_grad=False)  # eigvals(Z)\n",
     "        self.zz_eigvals = torch.nn.Parameter(torch.from_numpy(np.array([1.0, -1.0, -1.0, 1.0])).float(),\n",
     "                                             requires_grad=False)  # eigvals(ZZ)\n",
-    "        self.xx_eigvals = torch.nn.Parameter(torch.from_numpy(np.array([1.0, -1.0, 1.0, -1.0])).float(),\n",
-    "                                             requires_grad=False)  # eigvals(XX)\n",
     "\n",
     "        self.local_fields_wires = [[i] for i in range(self.n_qubits)]\n",
     "        self.couplings_wires = [[i, j] for i, j in np.vstack(np.triu_indices(self.n_qubits, k=1)).T]\n",
@@ -153,7 +159,6 @@
     "\n",
     "        local_fields = self.local_fields_head(embeddings)\n",
     "        zz_couplings = self.zz_body_couplings_head(embeddings)\n",
-    "        xx_couplings = self.xx_body_couplings_head(embeddings)\n",
     "        self.q_device.execute_generator(self.circuit_gen(), reset=True)\n",
     "\n",
     "        local_fields_probs = (\n",
@@ -170,11 +175,8 @@
     "        weighted_zz_eigvals = torch.einsum(\n",
     "            \"bi,kij,j->bk\", zz_couplings, couplings_probs, self.zz_eigvals\n",
     "        )\n",
-    "        weighted_xx_eigvals = torch.einsum(\n",
-    "            \"bi,kij,j->bk\", xx_couplings, couplings_probs, self.xx_eigvals\n",
-    "        )\n",
     "\n",
-    "        expval = weighted_local_eigvals + weighted_zz_eigvals + weighted_xx_eigvals\n",
+    "        expval = weighted_local_eigvals + weighted_zz_eigvals\n",
     "        return expval\n",
     "\n",
     "    def circuit_gen(self):\n",
@@ -253,12 +255,11 @@
     "model_cls = NIFCNN\n",
     "\n",
     "# Pipeline\n",
-    "job_output_folder_root = Path(os.getcwd()) / \"data\" / \"automl\"\n",
-    "job_output_folder = Path(dataset_name) / model_cls.MODEL_NAME\n",
+    "job_output_folder = Path(os.getcwd()) / \"data\" / \"automl\" / dataset_name / model_cls.MODEL_NAME\n",
     "checkpoint_folder = Path(job_output_folder) / \"checkpoints\"\n",
     "pipeline_args = dict(\n",
-    "    max_epochs=100,\n",
-    "    max_time=\"00:00:03:00\",  # DD:HH:MM:SS\n",
+    "    max_epochs=100,  # increase at least to 256\n",
+    "    max_time=\"00:00:02:00\",  # DD:HH:MM:SS, increase at least to \"00:01:00:00\"\n",
     ")"
    ],
    "id": "d8db16a0825411",
@@ -292,9 +293,9 @@
     "    model_cls=model_cls,\n",
     "    datamodule=datamodule,\n",
     "    checkpoint_folder=checkpoint_folder,\n",
-    "    automl_iterations=5,\n",
-    "    inner_max_epochs=10,\n",
-    "    inner_max_time=\"00:00:01:00\",\n",
+    "    automl_iterations=5,  # increase at least to 32\n",
+    "    inner_max_epochs=10,  # increase at least to 128\n",
+    "    inner_max_time=\"00:00:01:00\",  # increase at least to \"00:00:10:00\"\n",
     "    automl_overwrite_fit=True,\n",
     "    **pipeline_args\n",
     ")"
@@ -310,19 +311,36 @@
    "id": "af11b75e95e08c0c"
   },
   {
-   "metadata": {},
+   "metadata": {
+    "jupyter": {
+     "is_executing": true
+    }
+   },
    "cell_type": "code",
    "source": [
     "start_time = time.perf_counter()\n",
     "automl_pipeline.run()\n",
     "end_time = time.perf_counter()\n",
-    "print(f\"Time taken: {end_time - start_time:.4f} seconds\")\n",
-    "print(f\"Best Hyperparameters:\\n{json.dumps(automl_pipeline.get_best_params(), indent=2, default=str)}\")"
+    "print(f\"Time taken: {end_time - start_time:.4f} seconds\")"
    ],
    "id": "cf7202cce5b169d9",
    "outputs": [],
    "execution_count": null
   },
+  {
+   "metadata": {},
+   "cell_type": "markdown",
+   "source": "We can then check the best hyperparameters found.",
+   "id": "cb990e91a45ed0b7"
+  },
+  {
+   "metadata": {},
+   "cell_type": "code",
+   "source": "print(f\"Best Hyperparameters:\\n{json.dumps(automl_pipeline.get_best_params(), indent=2, default=str)}\")",
+   "id": "dd640fc80cf084de",
+   "outputs": [],
+   "execution_count": null
+  },
   {
    "metadata": {},
    "cell_type": "markdown",