High-level Quantum Reservoir Computing Implementation (#290)

Author: David-Kreplin

.gitignore

@@ -169,7 +169,7 @@ examples/qnn/*_log.csv
 examples/**/*.log
 
 #Example caches
-examples/**/_cache
+**/_cache
 examples/integration/mlruns
 
 # Genereated doc images

docs/modules/classes.rst

@@ -20,6 +20,7 @@ QML Regressors
    kernel.ml.QKRR
    kernel.ml.QGPR
    qnn.QNNRegressor
+   qrc.QRCRegressor
 
 
 QML Classifiers
@@ -35,7 +36,7 @@ QML Classifiers
    kernel.ml.QSVC
    kernel.ml.QGPC
    qnn.QNNClassifier
-
+   qrc.QRCClassifier
 
 Circuit Design
 ====================================
@@ -283,6 +284,7 @@ Base Classes
     optimizers.optimizer_base.OptimizerBase
     qnn.base_qnn.BaseQNN
     qnn.loss.LossBase
+    qrc.base_qrc.BaseQRC
 
 
 

examples/qnn/example_minibatch.ipynb

@@ -295,7 +295,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.10"
+   "version": "3.9.13"
   },
   "orig_nbformat": 4
  },

examples/qrc/qrc_classification.ipynb

@@ -439,7 +439,7 @@ def __init__(
 
         # Set-up of the QNN
         self._qnn = LowLevelQNN(
-            self._encoding_circuit, self._measurement, executor, result_caching=self._caching
+            self._encoding_circuit, self._measurement, executor, caching=self._caching
         )
 
         # Set-up of the outer kernel

examples/qrc/qrc_regression.ipynb

@@ -324,7 +324,7 @@ def _fit(self, X, y, weights: np.ndarray = None) -> None:
 
     def _initialize_lowlevel_qnn(self):
         self._qnn = LowLevelQNN(
-            self.encoding_circuit, self.operator, self.executor, result_caching=self.caching
+            self.encoding_circuit, self.operator, self.executor, caching=self.caching
         )
 
     def _validate_input(self, X, y, incremental, reset):

examples/tutorials/quantum_bayesian_optimization.ipynb

@@ -84,7 +84,7 @@ class LowLevelQNNPennyLane(LowLevelQNNBase):
         operator (Union[ObservableBase,list]): Operator that is used in the expectation
             value of the QNN. Can be a list for multiple outputs.
         executor (Executor) : Executor that is used for the evaluation of the QNN
-        result_caching : Caching of the result for each `x`, `param`, `param_op` combination
+        caching : Caching of the result for each `x`, `param`, `param_op` combination
             (default = True)
 
     Attributes:
@@ -110,13 +110,13 @@ def __init__(
         parameterized_quantum_circuit: EncodingCircuitBase,
         observable: Union[ObservableBase, list],
         executor: Executor,
-        result_caching: bool = True,
+        caching: bool = True,
     ) -> None:
 
         super().__init__(parameterized_quantum_circuit, observable, executor)
 
         # Initialize result cache
-        self._result_caching = result_caching
+        self.caching = caching
         self.result_container = {}
 
         self._initialize_pennylane_circuit()
@@ -341,7 +341,7 @@ def evaluate(
 
         # return dictionary for input data, it will be empty
         # if the combination of x,param,param_op is touched the first time
-        if self._result_caching == True:
+        if self.caching == True:
             caching_tuple = (
                 to_tuple(x),
                 to_tuple(param),
@@ -460,7 +460,7 @@ def evaluate(
             value_dict[post.key] = post.evaluation_function(value_dict)
 
         # Store the updated dictionary for the theta value
-        if self._result_caching:
+        if self.caching:
             self.result_container[caching_tuple] = value_dict
 
         return value_dict

src/squlearn/kernel/matrix/projected_quantum_kernel.py

@@ -273,8 +273,7 @@ class LowLevelQNNQiskit(LowLevelQNNBase):
         operator (Union[ObservableBase,list]): Operator that is used in the expectation
             value of the QNN. Can be a list for multiple outputs.
         executor (Executor) : Executor that is used for the evaluation of the QNN
-        optree_caching : Caching of the optree expressions (default = True recommended)
-        result_caching : Caching of the result for each `x`, `param`, `param_op` combination
+        caching : Caching of the result for each `x`, `param`, `param_op` combination
             (default = True)
 
     Attributes:
@@ -300,17 +299,15 @@ def __init__(
         parameterized_quantum_circuit: EncodingCircuitBase,
         operator: Union[ObservableBase, list],
         executor: Executor,
-        optree_caching=True,
-        result_caching=True,
+        caching=True,
     ) -> None:
 
         parameterized_quantum_circuit = TranspiledEncodingCircuit(
             parameterized_quantum_circuit, executor.backend
         )
         super().__init__(parameterized_quantum_circuit, operator, executor)
 
-        self._optree_caching = optree_caching
-        self._result_caching = result_caching
+        self.caching = caching
 
         # Set-Up Executor
         if self._executor.optree_executor == "estimator":
@@ -405,8 +402,8 @@ def _initilize_derivative(self):
             self._observable.set_map(self._pqc.qubit_map, self._pqc.num_physical_qubits)
             num_qubits_operator = self._observable.num_qubits
 
-        self.operator_derivatives = ObservableDerivatives(self._observable, self._optree_caching)
-        self.pqc_derivatives = EncodingCircuitDerivatives(self._pqc, self._optree_caching)
+        self.operator_derivatives = ObservableDerivatives(self._observable)
+        self.pqc_derivatives = EncodingCircuitDerivatives(self._pqc)
 
         if self._pqc.num_virtual_qubits != num_qubits_operator:
             raise ValueError("Number of Qubits are not the same!")
@@ -937,7 +934,7 @@ def add_to_real_todo_dic(item: Expec, real_todo_dic, value_dict):
 
         # return dictionary for input data, it will be empty
         # if the combination of x,param,param_op is touched the first time
-        if self._result_caching == True:
+        if self.caching == True:
             caching_tuple = (
                 to_tuple(x),
                 to_tuple(param),
@@ -1120,7 +1117,7 @@ def add_to_real_todo_dic(item: Expec, real_todo_dic, value_dict):
         value_dict["param_op"] = param_op
 
         # Store the updated dictionary for the theta value
-        if self._result_caching:
+        if self.caching:
             self.result_container[caching_tuple] = value_dict
 
         return value_dict