Integration with qiskit-symb

.github/workflows/deploy_ghpages.yml

@@ -22,6 +22,12 @@ jobs:
         with:
           path: ~/.cache/pip
           key: deploy_ghpages.yml
+      - name: Copy feature maps
+        run: |
+          cp -a ./symb_statevectors/. ./examples/ERP/symb_statevectors
+          cp -a ./symb_statevectors/. ./examples/MI/symb_statevectors
+          cp -a ./symb_statevectors/. ./examples/other_datasets/symb_statevectors
+          cp -a ./symb_statevectors/. ./examples/toys_dataset/symb_statevectors
       - name: Generate HTML docs
         env:
           FIREBASE_CERTIFICATE: ${{ secrets.FIREBASE_CERTIFICATE }}
@@ -33,7 +39,6 @@ jobs:
             python -m pip install --upgrade pip
             apt-get -y install --fix-missing git-core
             apt-get -y install build-essential
-            pip install -r doc/requirements.txt
       - name:  Upload generated HTML as artifact
         uses: actions/upload-artifact@v4
         with:

README.md

@@ -123,12 +123,21 @@ To check the installation, open a python shell and type:
 import pyriemann_qiskit
 ```
 
-To enable Qiskit GPU optimization when using quantum simulation, run:
+To enable Qiskit GPU optimization (for Linux) when using quantum simulation, run:
+
+```
+pip install .[optim_linux]
+```
+
+To use symbolic quantum simulation, run:
 
 ```
 pip install .[optim]
 ```
 
+Which will enable [qiskit-symb](https://github.com/SimoneGasperini/qiskit-symb)
+integration.
+
 Note, Qiskit only provide binaries for Linux. For other platforms, or if you want to
 enable specific NVIDIA optimization for quantum, you need to build the binary
 [yourself](https://github.com/Qiskit/qiskit-aer/blob/main/CONTRIBUTING.md#building-with-gpu-support).

doc/requirements.txt

@@ -23,3 +23,5 @@ moabb==1.1.0
 pyriemann==0.6
 docplex>=2.21.207
 firebase_admin==6.5.0
+qiskit-symb
+symengine==0.11.0

examples/ERP/plot_classify_P300_bi.py

@@ -86,7 +86,10 @@
     # the non-qunatum SVM version used in qiskit
     # On a real Quantum computer (n_components = qubits)
     dim_red=PCA(n_components=5),
-    # params={'q_account_token': '<IBM Quantum TOKEN>'}
+    params={
+        "n_jobs": 1,  # Number of jobs for the simulator
+        # 'q_account_token': '<IBM Quantum TOKEN>'
+    },
 )
 
 # Here we provide a pipeline for comparison:

examples/other_datasets/plot_financial_data.py

@@ -50,6 +50,7 @@
 from sklearn.metrics import balanced_accuracy_score
 
 from pyriemann_qiskit.classification import QuanticSVM
+from pyriemann_qiskit.utils.hyper_params_factory import gen_zz_feature_map
 from pyriemann_qiskit.utils.preprocessing import NdRobustScaler
 
 print(__doc__)
@@ -370,7 +371,15 @@ def transform(self, X):
 # for the quantum SVM as for the classical one.
 gs.best_estimator_.steps[-1] = (
     "quanticsvm",
-    QuanticSVM(quantum=True, C=best_C, gamma=best_gamma, seed=42),
+    QuanticSVM(
+        quantum=True,
+        C=best_C,
+        gamma=best_gamma,
+        gen_feature_map=gen_zz_feature_map(),
+        seed=42,
+        n_jobs=1,
+        use_qiskit_symb=False,
+    ),
 )
 train_pred_qsvm = gs.best_estimator_.fit(X_train, y_train).predict(X_train)
 train_score_qsvm = balanced_accuracy_score(y_train, train_pred_qsvm)

pyriemann_qiskit/classification.py

@@ -81,7 +81,7 @@ class QuanticClassifierBase(BaseEstimator, ClassifierMixin):
         If true, will output all intermediate results and logs.
     shots : int, default=1024
         Number of repetitions of each circuit, for sampling.
-    gen_feature_map : Callable[int, QuantumCircuit | FeatureMap], \
+    gen_feature_map : Callable[[int, str], QuantumCircuit | FeatureMap], \
                       default=Callable[int, ZZFeatureMap]
         Function generating a feature map to encode data into a quantum state.
     seed : int | None, default=None
@@ -326,6 +326,9 @@ class QuanticSVM(QuanticClassifierBase):
         Predict is now using predict_proba with a softmax, when using QSVC.
     .. versionchanged:: 0.3.0
         Add use_fidelity_state_vector_kernel parameter
+    .. versionchanged:: 0.4.0
+        Add n_jobs and use_qiskit_symb parameter
+        for SymbFidelityStatevectorKernel
 
     Parameters
     ----------
@@ -357,17 +360,24 @@ class QuanticSVM(QuanticClassifierBase):
         If true, will output all intermediate results and logs.
     shots : int, default=1024
         Number of repetitions of each circuit, for sampling.
-    gen_feature_map : Callable[int, QuantumCircuit | FeatureMap], \
+    gen_feature_map : Callable[[int, str], QuantumCircuit | FeatureMap], \
                       default=Callable[int, ZZFeatureMap]
         Function generating a feature map to encode data into a quantum state.
     seed : int | None, default=None
         Random seed for the simulation
-    use_fidelity_state_vector_kernel: boolean (default=True)
+    use_fidelity_state_vector_kernel: boolean, default=True
         if True, use a FidelitystatevectorKernel for simulation.
+    use_qiskit_symb: boolean, default=True
+        This flag is used only if qiskit-symb is installed, and pegasos is False.
+        If True and the number of qubits < 9, then qiskit_symb is used.
+    n_jobs: boolean
+        The number of jobs for the qiskit-symb fidelity state vector
+        (if applicable)
 
     See Also
     --------
     QuanticClassifierBase
+    SymbFidelityStatevectorKernel
 
     References
     ----------
@@ -405,6 +415,8 @@ def __init__(
         gen_feature_map=gen_zz_feature_map(),
         seed=None,
         use_fidelity_state_vector_kernel=True,
+        use_qiskit_symb=True,
+        n_jobs=4,
     ):
         QuanticClassifierBase.__init__(
             self, quantum, q_account_token, verbose, shots, gen_feature_map, seed
@@ -414,14 +426,19 @@ def __init__(
         self.max_iter = max_iter
         self.pegasos = pegasos
         self.use_fidelity_state_vector_kernel = use_fidelity_state_vector_kernel
+        self.n_jobs = n_jobs
+        self.use_qiskit_symb = use_qiskit_symb
 
     def _init_algo(self, n_features):
         self._log("SVM initiating algorithm")
         if self.quantum:
             quantum_kernel = get_quantum_kernel(
                 self._feature_map,
+                self.gen_feature_map,
                 self._quantum_instance,
                 self.use_fidelity_state_vector_kernel,
+                self.use_qiskit_symb and not self.pegasos,
+                self.n_jobs,
             )
             if self.pegasos:
                 self._log("[Warning] `gamma` is not supported by PegasosQSVC")
@@ -496,7 +513,7 @@ class QuanticVQC(QuanticClassifierBase):
         If true, will output all intermediate results and logs
     shots : int, default=1024
         Number of repetitions of each circuit, for sampling
-    gen_feature_map : Callable[int, QuantumCircuit | FeatureMap], \
+    gen_feature_map : Callable[[int, str], QuantumCircuit | FeatureMap], \
                       default=Callable[int, ZZFeatureMap]
         Function generating a feature map to encode data into a quantum state.
     seed : int | None, default=None

pyriemann_qiskit/pipelines.py

@@ -458,7 +458,7 @@ class QuantumMDMVotingClassifier(BasePipeline):
         If true, will output all intermediate results and logs.
     shots : int (default:1024)
         Number of repetitions of each circuit, for sampling.
-    gen_feature_map : Callable[int, QuantumCircuit | FeatureMap] \
+    gen_feature_map : Callable[[int, str], QuantumCircuit | FeatureMap] \
                       (default : Callable[int, ZZFeatureMap])
         Function generating a feature map to encode data into a quantum state.
 

pyriemann_qiskit/utils/hyper_params_factory.py

@@ -25,8 +25,12 @@ def gen_x_feature_map(reps=2):
 
     Returns
     -------
-    ret : XFeatureMap
-        An instance of XFeatureMap.
+    ret : Callable[[int, str], XFeatureMap]
+        A callable that takes into arguments:
+
+        - the number of features
+        - the prefix string for the parameters
+        And returns an instance of XFeatureMap.
 
     Raises
     ------
@@ -41,16 +45,18 @@ def gen_x_feature_map(reps=2):
     Notes
     -----
     .. versionadded:: 0.2.0
+    .. versionchanged:: 0.4.0
+        Possibility to specify parameter prefix.
     """
     if reps < 1:
         raise ValueError(f"Parameter reps must be superior or equal to 1 (Got {reps})")
 
-    return lambda n_features: PauliFeatureMap(
+    return lambda n_features, param_prefix="x": PauliFeatureMap(
         feature_dimension=n_features,
         paulis=["X"],
         reps=reps,
         data_map_func=None,
-        parameter_prefix="x",
+        parameter_prefix=param_prefix,
         insert_barriers=False,
         name="XFeatureMap",
     )
@@ -70,8 +76,12 @@ def gen_z_feature_map(reps=2):
 
     Returns
     -------
-    ret : ZFeatureMap
-        An instance of ZFeatureMap.
+    ret : Callable[[int, str], ZFeatureMap]
+        A callable that takes into arguments:
+
+        - the number of features
+        - the prefix string for the parameters
+        And returns an instance of ZFeatureMap.
 
     Raises
     ------
@@ -86,11 +96,17 @@ def gen_z_feature_map(reps=2):
     Notes
     -----
     .. versionadded:: 0.2.0
+    .. versionchanged:: 0.4.0
+        Possibility to specify parameter prefix.
     """
     if reps < 1:
         raise ValueError(f"Parameter reps must be superior or equal to 1 (Got {reps})")
 
-    return lambda n_features: ZFeatureMap(feature_dimension=n_features, reps=reps)
+    return lambda n_features, param_prefix="x": ZFeatureMap(
+        feature_dimension=n_features,
+        reps=reps,
+        parameter_prefix=param_prefix,
+    )
 
 
 def gen_zz_feature_map(reps=2, entanglement="linear"):
@@ -114,8 +130,12 @@ def gen_zz_feature_map(reps=2, entanglement="linear"):
 
     Returns
     -------
-    ret : ZZFeatureMap
-        An instance of ZZFeatureMap.
+    ret : Callable[[int, str], ZZFeatureMap]
+        A callable that takes into arguments:
+
+        - the number of features
+        - the prefix string for the parameters
+        And returns an instance of ZZFeatureMap.
 
     Raises
     ------
@@ -132,12 +152,17 @@ def gen_zz_feature_map(reps=2, entanglement="linear"):
     Notes
     -----
     .. versionadded:: 0.0.1
+    .. versionchanged:: 0.4.0
+        Possibility to specify parameter prefix.
     """
     if reps < 1:
         raise ValueError(f"Parameter reps must be superior or equal to 1 (Got {reps})")
 
-    return lambda n_features: ZZFeatureMap(
-        feature_dimension=n_features, reps=reps, entanglement=entanglement
+    return lambda n_features, param_prefix="x": ZZFeatureMap(
+        feature_dimension=n_features,
+        reps=reps,
+        entanglement=entanglement,
+        parameter_prefix=param_prefix,
     )