fix/halving search

doc/installing.rst

@@ -3,44 +3,52 @@
 Installing pyRiemann-qiskit
 ===========================
 
-There is no yet stable version of pyRiemann-qiskit.
+_We recommend the use of [Anaconda](https://www.anaconda.com/) to manage python
+environements._
 
-Therefore, it is recommanded to clone the source code on `github <https://github.com/pyRiemann/pyRiemann-qiskit>`__ and install directly the package from source.
+`pyRiemann-qiskit` currently supports Windows, Mac and Linux OS with Python 3.9 - 3.11.
 
-``pip install -e .``
+You can install `pyRiemann-qiskit` release from PyPI:
 
-The install script will install the required dependencies. If you want also to build the documentation and to run the test locally, you could install all development dependencies with
+```
+pip install pyriemann-qiskit
+```
 
-``pip install -e .[docs,tests]``
+The development version can be installed by cloning this repository and installing the
+package on your local machine using the `setup.py` script:
 
-If you use a zsh shell, you need to write `pip install -e .\[docs,tests\]`. If you do not know what zsh is, you could use the above command.
+```
+python setup.py develop
+```
 
-You may check that the package was correctly installed by starting a python shell and writing:
+Or directly pip:
 
-``import pyriemann_qiskit``
+```
+pip install .
+```
+
+Note that the steps above need to be re-executed in your local environment after any
+changes inside your local copy of the `pyriemann_qiskit` folder, including pulling from
+remote.
+
+To check the installation, open a python shell and type:
+
+```
+import pyriemann_qiskit
+```
 
 Dependencies
 ~~~~~~~~~~~~
 
--  Python (>= 3.6)
+-  Python (>= 3.9)
 
 Mandatory dependencies
 ^^^^^^^^^^^^^^^^^^^^^^
 
--  `cython <https://cython.org/>`__
-
--  `pyriemann <https://github.com/pyRiemann/pyRiemann-qiskit>`__
-
--  `qiskit==0.20.0 <https://qiskit.org/>`__
-
--  `cvxpy=1.1.12 <https://www.cvxpy.org/>`__
-
-Recommended dependencies
-^^^^^^^^^^^^^^^^^^^^^^^^
-These dependencies are recommanded to use the plotting functions of pyriemann or to run examples and tutorials, but they are not mandatory:
+The project relies mainly on:
 
--  `matplotlib>=2.2 <https://matplotlib.org/>`__
+- `pyRiemann==0.5.0 <https://github.com/pyRiemann/pyRiemann-qiskit>`
 
--  `mne-python <http://mne-tools.github.io/>`__
+- `qiskit==0.45.0 <https://qiskit.org/>`__
 
--  `seaborn <https://seaborn.pydata.org>`__
+Other dependencies can be checked in the `setup.py` file.

doc/requirements.txt

@@ -13,7 +13,7 @@ cvxpy==1.4.1
 qiskit_machine_learning==0.6.1
 qiskit-ibm-provider==0.7.2
 qiskit-optimization==0.5.0
-qiskit-aer==0.13.0
+qiskit-aer==0.12.2
 scipy==1.11.3
 moabb>=1.0.0
 pyriemann==0.5

examples/other_datasets/plot_financial_data.py

@@ -24,7 +24,7 @@
 
 In this example, we will use RG to identify whether or no a fraud is a probable
 collusion.
-Because this method work on a small number of components, it is also compatible
+Because this method works on a small number of components, it is also compatible
 with quantum computing.
 """
 # Authors: Gregoire Cattan, Filipe Barroso
@@ -48,6 +48,7 @@
 import warnings
 import pandas as pd
 import numpy as np
+import os
 
 print(__doc__)
 
@@ -278,7 +279,7 @@ def transform(self, X):
 
 
 # Create a RandomForest for baseline comparison of direct classification:
-rf = RandomForestClassifier()
+rf = RandomForestClassifier(random_state=42)
 
 # Classical pipeline: puts together transformers,
 # then adds at the end a classical SVM
@@ -291,23 +292,37 @@ def transform(self, X):
     SVC(probability=True),
 )
 
+if os.getenv("CI") == "true":
+    print("Feeding a good estimator for CI (skipping grid search)")
+    param_grid: dict = {
+        "toepochs__n": [30],
+        "xdawncovariances__nfilter": [1],
+        "optionalwhitening__process": [True],
+        "optionalwhitening__n_components": [4],
+        "slimvector__keep_diagonal": [True],
+        "svc__C": [1],
+        "svc__gamma": ["auto"],
+    }
+else:
+    param_grid: dict = {
+        "toepochs__n": [20, 30],
+        "xdawncovariances__nfilter": [1, 2],
+        "optionalwhitening__process": [True, False],
+        "optionalwhitening__n_components": [2, 4],
+        "slimvector__keep_diagonal": [True, False],
+        "svc__C": [0.1, 1],
+        "svc__gamma": ["auto", "scale"],
+    }
+
 # Optimize the pipeline:
 # let's save some time and run the optimization with the classical SVM
 gs = HalvingGridSearchCV(
     pipe,
-    param_grid={
-        "toepochs__n": [10, 20, 30],
-        "xdawncovariances__nfilter": [1, 2, 3],
-        "optionalwhitening__process": [True, False],
-        "optionalwhitening__n_components": [2, 4, 6],
-        "slimvector__keep_diagonal": [True, False],
-        "svc__C": [0.1, 1, 10, 100],
-        "svc__gamma": ["auto", "scale", 1, 10],
-    },
+    param_grid=param_grid,
     scoring="balanced_accuracy",
     cv=4,
-    min_resources="smallest",
     verbose=1,
+    random_state=0,
 )
 
 
@@ -324,12 +339,14 @@ def transform(self, X):
 # So `NearMiss` we choose the closest non-fraud epochs to the fraud-epochs.
 # Here we will keep a ratio of 2 non-fraud epochs for 1 fraud epochs.
 # Note: at this stage `features` also contains the `index` column.
-X, y = NearMiss(sampling_strategy=0.5).fit_resample(
+
+# Possibly avoids tie-break situations
+np.random.seed(42)
+
+X, y = NearMiss(sampling_strategy=0.5, n_jobs=-1, n_neighbors=3).fit_resample(
     features.to_numpy(), target.to_numpy()
 )
-# X, y = EditedNearestNeighbours().fit_resample(features.to_numpy(), target.to_numpy())
-# X = features.to_numpy()
-# y = target.to_numpy()
+
 X_train, X_test, y_train, y_test = train_test_split(
     X, y, test_size=0.2, random_state=42, stratify=y
 )
@@ -377,7 +394,7 @@ 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),
+    QuanticSVM(quantum=True, C=best_C, gamma=best_gamma, seed=42),
 )
 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)
@@ -426,7 +443,7 @@ def transform(self, X):
 
 
 class ERP_CollusionClassifier(ClassifierMixin):
-    def __init__(self, row_clf, erp_clf, threshold=0.7):
+    def __init__(self, row_clf, erp_clf, threshold=0.5):
         self.row_clf = row_clf
         self.erp_clf = erp_clf
         self.threshold = threshold

pyriemann_qiskit/classification.py

@@ -15,7 +15,7 @@
     NaiveQAOAOptimizer,
     set_global_optimizer,
 )
-from qiskit.utils import QuantumInstance, algorithm_globals
+from qiskit.utils import QuantumInstance
 from qiskit.utils.quantum_instance import logger
 from qiskit_ibm_provider import IBMProvider, least_busy
 from qiskit_machine_learning.algorithms import QSVC, VQC, PegasosQSVC
@@ -62,12 +62,16 @@ class QuanticClassifierBase(BaseEstimator, ClassifierMixin):
     gen_feature_map : Callable[int, 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
 
     Notes
     -----
     .. versionadded:: 0.0.1
     .. versionchanged:: 0.1.0
         Added support for multi-class classification.
+    .. versionchanged:: 0.2.0
+        Add seed parameter
 
     Attributes
     ----------
@@ -95,20 +99,20 @@ def __init__(
         verbose=True,
         shots=1024,
         gen_feature_map=gen_zz_feature_map(),
+        seed=None,
     ):
         self.verbose = verbose
         self._log("Initializing Quantum Classifier")
         self.q_account_token = q_account_token
         self.quantum = quantum
         self.shots = shots
+        self.seed = datetime.now().microsecond if seed is None else seed
         self.gen_feature_map = gen_feature_map
         # protected field for child classes
         self._training_input = {}
 
     def _init_quantum(self):
         if self.quantum:
-            algorithm_globals.random_seed = datetime.now().microsecond
-            self._log("seed = ", algorithm_globals.random_seed)
             if self.q_account_token:
                 self._log("Real quantum computation will be performed")
                 if not self.q_account_token == "load_account":
@@ -186,14 +190,13 @@ def fit(self, X, y):
                 except Exception:
                     self._log("Devices are all busy. Getting the first one...")
                     self._backend = devices[0]
-                self._log("Quantum backend = ", self._backend)
-            seed_sim = algorithm_globals.random_seed
-            seed_trs = algorithm_globals.random_seed
+            self._log("Quantum backend = ", self._backend)
+            self._log("seed = ", self.seed)
             self._quantum_instance = QuantumInstance(
                 self._backend,
                 shots=self.shots,
-                seed_simulator=seed_sim,
-                seed_transpiler=seed_trs,
+                seed_simulator=self.seed,
+                seed_transpiler=self.seed,
             )
         self._classifier = self._init_algo(n_features)
         self._train(X, y)
@@ -248,9 +251,11 @@ class QuanticSVM(QuanticClassifierBase):
     -----
     .. versionadded:: 0.0.1
     .. versionchanged:: 0.0.2
-       Qiskit's Pegasos implementation [4]_, [5]_.
+        Qiskit's Pegasos implementation [4]_, [5]_.
     .. versionchanged:: 0.1.0
-       Fix: copy estimator not keeping base class parameters.
+        Fix: copy estimator not keeping base class parameters.
+    .. versionchanged:: 0.2.0
+        Add seed parameter
 
     Parameters
     ----------
@@ -285,6 +290,8 @@ class QuanticSVM(QuanticClassifierBase):
     gen_feature_map : Callable[int, 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
 
     See Also
     --------
@@ -324,9 +331,10 @@ def __init__(
         verbose=True,
         shots=1024,
         gen_feature_map=gen_zz_feature_map(),
+        seed=None,
     ):
         QuanticClassifierBase.__init__(
-            self, quantum, q_account_token, verbose, shots, gen_feature_map
+            self, quantum, q_account_token, verbose, shots, gen_feature_map, seed
         )
         self.gamma = gamma
         self.C = C
@@ -435,13 +443,17 @@ class QuanticVQC(QuanticClassifierBase):
     gen_feature_map : Callable[int, 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
 
     Notes
     -----
     .. versionadded:: 0.0.1
     .. versionchanged:: 0.1.0
-       Fix: copy estimator not keeping base class parameters.
-       Added support for multi-class classification.
+        Fix: copy estimator not keeping base class parameters.
+        Added support for multi-class classification.
+    .. versionchanged:: 0.2.0
+        Add seed parameter
 
     See Also
     --------
@@ -477,14 +489,15 @@ def __init__(
         verbose=True,
         shots=1024,
         gen_feature_map=gen_zz_feature_map(),
+        seed=None,
     ):
         if quantum is False:
             raise ValueError(
                 "VQC can only run on a quantum \
                               computer or simulator."
             )
         QuanticClassifierBase.__init__(
-            self, quantum, q_account_token, verbose, shots, gen_feature_map
+            self, quantum, q_account_token, verbose, shots, gen_feature_map, seed
         )
         self.optimizer = optimizer
         self.gen_var_form = gen_var_form
@@ -568,7 +581,9 @@ class QuanticMDM(QuanticClassifierBase):
     -----
     .. versionadded:: 0.0.4
     .. versionchanged:: 0.1.0
-       Fix: copy estimator not keeping base class parameters.
+        Fix: copy estimator not keeping base class parameters.
+    .. versionchanged:: 0.2.0
+        Add seed parameter
 
     Parameters
     ----------
@@ -598,6 +613,8 @@ class QuanticMDM(QuanticClassifierBase):
     gen_feature_map : Callable[int, 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
 
     See Also
     --------
@@ -626,9 +643,10 @@ def __init__(
         verbose=True,
         shots=1024,
         gen_feature_map=gen_zz_feature_map(),
+        seed=None,
     ):
         QuanticClassifierBase.__init__(
-            self, quantum, q_account_token, verbose, shots, gen_feature_map
+            self, quantum, q_account_token, verbose, shots, gen_feature_map, seed
         )
         self.metric = metric
 

requirements.txt

@@ -4,7 +4,7 @@ git+https://github.com/pyRiemann/pyRiemann#egg=pyriemann
 qiskit_machine_learning==0.6.1
 qiskit-ibm-provider==0.7.2
 qiskit-optimization==0.5.0
-qiskit-aer==0.13.0
+qiskit-aer==0.12.2
 cvxpy==1.4.1
 scipy==1.11.3
 docplex>=2.21.207

setup.py

@@ -37,10 +37,11 @@
                         'numpy<1.27',
                         'cython',
                         'pyriemann==0.5',
+                        'qiskit==0.45.0',
                         'qiskit_machine_learning==0.6.1',
                         'qiskit-ibm-provider==0.7.2',
                         'qiskit-optimization==0.5.0',
-                        'qiskit-aer==0.13.0',
+                        'qiskit-aer==0.12.2',
                         'cvxpy==1.4.1',
                         'scipy==1.11.3',
                         'docplex>=2.21.207',