Merge pull request #49 from ChitambarLab/pennylane-update-29

updating pennylane to v0.29

README.md

@@ -32,7 +32,7 @@ $ pip install qnetvo
 Install PennyLane:
 
 ```
-$ pip install pennylane==0.28
+$ pip install pennylane==0.29
 ```
 
 Import packages:
@@ -48,7 +48,7 @@ Note
 </p>
 <p>
 For optimal use, qNetVO should be used with PennyLane.
-QNetVO is currently compatible with PennyLane v0.28.
+QNetVO is currently compatible with PennyLane v0.29.
 </p>
 </div>
 

docs/source/utilities.rst

@@ -22,8 +22,4 @@ File I/O
 Data Conversions
 ----------------
 
-.. autofunction:: settings_to_list
-
-.. autofunction:: settings_to_np
-
 .. autofunction:: mixed_base_num

requirements.txt

@@ -1,4 +1,4 @@
-pennylane==0.28
+pennylane~=0.29
 qiskit==0.34
 pennylane-qiskit==0.20
 tensorflow>=2.0,<3.0

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = qNetVO
-version = 0.4.1
+version = 0.4.2
 author = Brian Doolittle and Tom Bromley
 author_email = [email protected]
 description = The Quantum Network Variational Optimizer
@@ -20,7 +20,7 @@ package_dir =
 packages = find:
 python_requires = >=3.8
 install_requires =
-    pennylane==0.28
+    pennylane~=0.29
 
 [options.packages.find]
 where = src

src/qnetvo/cost/I_3322_bell_inequality.py

@@ -56,7 +56,7 @@ def I_3322_bell_inequality_cost_fn(network_ansatz, **qnode_kwargs):
 
     def cost(*network_settings):
         score = 0
-        for (x, y, mult) in [
+        for x, y, mult in [
             (0, 0, 1),
             (0, 1, 1),
             (0, 2, 1),

src/qnetvo/cost/chsh_inequality.py

@@ -40,7 +40,6 @@ def chsh_inequality_cost_fn(network_ansatz, parallel=False, **qnode_kwargs):
         chsh_qnode = global_parity_expval_qnode(network_ansatz, **qnode_kwargs)
 
     def chsh_cost(*network_settings):
-
         xy_settings = [
             network_ansatz.qnode_settings(network_settings, network_input)
             for network_input in network_inputs

src/qnetvo/cost/linear_inequalities.py

@@ -77,7 +77,7 @@ def linear_probs_cost_fn(network_ansatz, game, postmap=np.array([]), qnode_kwarg
 
     def cost(*network_settings):
         score = 0
-        for (i, input_id_set) in enumerate(node_input_ids):
+        for i, input_id_set in enumerate(node_input_ids):
             settings = network_ansatz.qnode_settings(network_settings, input_id_set)
 
             raw_probs = probs_qnode(settings)

src/qnetvo/cost/mermin_klyshko_inequality.py

@@ -79,13 +79,11 @@ def mermin_klyshko_cost_fn(ansatz, **qnode_kwargs):
     static_prep_inputs = [[0] * len(layer_nodes) for layer_nodes in ansatz.layers[0:-1]]
 
     def cost(*network_settings):
-
         score = 0
 
         num_correlators = len(meas_inputs_list)
 
         for i in range(num_correlators):
-
             meas_inputs = meas_inputs_list[i]
             scalar = scalars_list[i]
 

src/qnetvo/cost/mutual_info.py

@@ -66,7 +66,7 @@ def mutual_info_cost_fn(ansatz, priors, postmap=np.array([]), **qnode_kwargs):
     def cost(*network_settings):
         Hxy = 0
         py_vec = np.zeros(net_num_out)
-        for (i, input_id_set) in enumerate(node_input_ids):
+        for i, input_id_set in enumerate(node_input_ids):
             settings = ansatz.qnode_settings(network_settings, input_id_set)
             p_net = postmap @ probs_qnode(settings)
 

src/qnetvo/cost/nlocal_chain_bell_inequality.py

@@ -37,7 +37,6 @@ def chain_I22_fn(network_ansatz, parallel=False, **qnode_kwargs):
     I22_xy_inputs = [[x_a] + [0 for i in range(num_interior_nodes)] + [x_b] for x_a, x_b in xy_vals]
 
     def I22(*network_settings):
-
         I22_xy_settings = [
             network_ansatz.qnode_settings(network_settings, static_prep_inputs + [meas_inputs])
             for meas_inputs in I22_xy_inputs
@@ -91,7 +90,6 @@ def chain_J22_fn(network_ansatz, parallel=False, **qnode_kwargs):
     J22_xy_inputs = [[x_a] + [1 for i in range(num_interior_nodes)] + [x_b] for x_a, x_b in xy_vals]
 
     def J22(*network_settings):
-
         J22_xy_settings = [
             network_ansatz.qnode_settings(network_settings, static_prep_inputs + [meas_inputs])
             for meas_inputs in J22_xy_inputs
@@ -158,7 +156,6 @@ def nlocal_chain_22_cost_fn(network_ansatz, parallel=False, **qnode_kwargs):
     J22 = chain_J22_fn(network_ansatz, parallel=parallel, **qnode_kwargs)
 
     def cost(*network_settings):
-
         I22_score = I22(*network_settings)
         J22_score = J22(*network_settings)
 
@@ -218,7 +215,6 @@ def _nat_grad_fn(settings, qnode):
     grad_fn = _nat_grad_fn if natural_grad else _grad_fn
 
     def nlocal_chain_grad_fn(*network_settings):
-
         I22_score = I22(*network_settings)
         J22_score = J22(*network_settings)
 
@@ -249,7 +245,6 @@ def nlocal_chain_grad_fn(*network_settings):
         J22_scalar = -(1 / 4) * math.sign(J22_score) / math.sqrt(math.abs(J22_score))
 
         for i in range(4):
-
             x = I22_xy_meas_inputs[i][0]
             y = I22_xy_meas_inputs[i][-1]
             J22_sign = (-1) ** (x + y)

src/qnetvo/cost/nlocal_star_bell_inequality.py

@@ -53,7 +53,6 @@ def star_I22_fn(network_ansatz, parallel=False, nthreads=4, **qnode_kwargs):
         star_qnode = global_parity_expval_qnode(network_ansatz, **qnode_kwargs)
 
     def I22(*network_settings):
-
         I22_x_settings = [
             network_ansatz.qnode_settings(network_settings, network_inputs)
             for network_inputs in network_input_x_vals
@@ -132,7 +131,6 @@ def star_J22_fn(network_ansatz, parallel=False, nthreads=4, **qnode_kwargs):
         star_qnode = global_parity_expval_qnode(network_ansatz, **qnode_kwargs)
 
     def J22(*network_settings):
-
         J22_x_settings = [
             network_ansatz.qnode_settings(network_settings, network_inputs)
             for network_inputs in network_input_x_vals
@@ -207,7 +205,6 @@ def nlocal_star_22_cost_fn(network_ansatz, parallel=False, nthreads=4, **qnode_k
     J22 = star_J22_fn(network_ansatz, parallel=parallel, nthreads=nthreads, **qnode_kwargs)
 
     def cost(*network_settings):
-
         I22_score = I22(*network_settings)
         J22_score = J22(*network_settings)
 
@@ -280,7 +277,6 @@ def _ng(qnode, settings):
     _grad = _ng if natural_grad else _g
 
     def nlocal_star_grad(*network_settings):
-
         I22_score = I22(*network_settings)
         J22_score = J22(*network_settings)
 

src/qnetvo/information.py

@@ -62,7 +62,7 @@ def behavior_fn(network_ansatz, postmap=np.array([]), qnode_kwargs={}):
 
     def behavior(network_settings):
         raw_behavior = np.zeros((raw_net_num_out, net_num_in))
-        for (i, input_id_set) in enumerate(node_input_ids):
+        for i, input_id_set in enumerate(node_input_ids):
             settings = network_ansatz.qnode_settings(
                 network_settings,
                 [input_id_set[0 : len(num_in_prep_nodes)], input_id_set[len(num_in_prep_nodes) :]],

src/qnetvo/network_ansatz.py

@@ -155,7 +155,6 @@ def get_network_parameter_partitions(self):
         for i, layer in enumerate(self.layers):
             parameter_partitions += [[]]
             for j, node in enumerate(layer):
-
                 parameter_partitions[i] += [[]]
                 for _ in range(node.num_in):
                     stop_id = start_id + node.num_settings

src/qnetvo/utilities.py

@@ -108,9 +108,9 @@ def write_optimization_json(opt_dict, filename):
 
     opt_dict_json["opt_score"] = float(opt_dict_json["opt_score"])
     opt_dict_json["scores"] = [float(score) for score in opt_dict_json["scores"]]
-    opt_dict_json["opt_settings"] = settings_to_list(opt_dict_json["opt_settings"])
+    opt_dict_json["opt_settings"] = [float(setting) for setting in opt_dict_json["opt_settings"]]
     opt_dict_json["settings_history"] = [
-        settings_to_list(settings) for settings in opt_dict_json["settings_history"]
+        [float(setting) for setting in settings] for settings in opt_dict_json["settings_history"]
     ]
 
     with open(filename + ".json", "w") as file:
@@ -130,44 +130,9 @@ def read_optimization_json(filepath):
     with open(filepath) as file:
         opt_dict = json.load(file)
 
-    opt_dict["opt_settings"] = settings_to_np(opt_dict["opt_settings"])
-    opt_dict["settings_history"] = [
-        settings_to_np(settings) for settings in opt_dict["settings_history"]
-    ]
-
     return opt_dict
 
 
-def settings_to_list(np_network_settings):
-    """Converts the numpy array in scenario settings to lists.
-    This function is intended for printing purposes.
-
-    :param np_network_settings: The scenario settings structure for a `NetworkAnsatz`.
-
-    :returns: The same nested array elements and structure using lists.
-    """
-
-    list_prep_settings = [node_settings.tolist() for node_settings in np_network_settings[0]]
-    list_meas_settings = [node_settings.tolist() for node_settings in np_network_settings[1]]
-
-    return [list_prep_settings, list_meas_settings]
-
-
-def settings_to_np(list_network_settings):
-    """Converts the nested list elements in the scenario settings to numpy arrays.
-    This function is intended for printing purposes.
-
-    :param list_network_settings: The scenario settings structure for a `NetworkAnsatz`.
-
-    :returns: The same nested array elements and structure using numpy arrays.
-    """
-
-    np_prep_settings = [qnp.array(node_settings) for node_settings in list_network_settings[0]]
-    np_meas_settings = [qnp.array(node_settings) for node_settings in list_network_settings[1]]
-
-    return [np_prep_settings, np_meas_settings]
-
-
 def mixed_base_num(n, base_digits):
     """Converts a base-10 number ``n`` into a mixed base number with digit
     values described by the ``base_digits`` array.

test/ansatz_library_test.py

@@ -277,7 +277,6 @@ def match_expval(noise_param):
                 return qml.expval(obs(wires=[0]))
 
             for noise_param in np.arange(0, 1.001, 1 / 10):
-
                 assert np.isclose(test_expval(noise_param), match_expval(noise_param), atol=1e-7)
 
     @pytest.mark.parametrize(
@@ -340,12 +339,10 @@ def match_expval(noise_param):
                 return qml.expval(obs(wires=[0]))
 
             for noise_param in np.arange(0, 1.001, 1 / 10):
-
                 assert np.isclose(test_expval(noise_param), match_expval(noise_param), atol=1e-7)
 
     @pytest.mark.parametrize("gamma", np.arange(-0.1, 1.11, 0.1))
     def test_two_qubit_depolarizing(self, gamma):
-
         dev = qml.device("default.mixed", wires=[0, 1])
 
         @qml.qnode(dev)
@@ -373,7 +370,6 @@ def test_noise(gamma):
 
     @pytest.mark.parametrize("gamma", np.arange(-0.1, 1.11, 0.1))
     def test_colored_noise(self, gamma):
-
         dev = qml.device("default.mixed", wires=[0, 1])
 
         @qml.qnode(dev)

test/cost/I_3322_bell_inequality_test.py

@@ -7,7 +7,6 @@
 
 class TestCostI3322BellInequality:
     def test_post_process_I_3322_joint_probs(self):
-
         probs_vec = np.array([0.1, 0.2, 0.3, 0.4])
         P00xy = qnet.post_process_I_3322_joint_probs(probs_vec)
         assert P00xy == 0.1

test/cost/chsh_inequality_test.py

@@ -7,7 +7,6 @@
 class TestCHSHInequalityCost:
     @pytest.mark.parametrize("parallel_flag", [False, True])
     def test_chsh_inequality_cost_fn(self, parallel_flag):
-
         prep_nodes = [qnet.PrepareNode(1, [0, 1], qnet.ghz_state, 0)]
         meas_nodes = [
             qnet.MeasureNode(2, 2, [0], qnet.local_RY, 1),

test/cost/magic_squares_game_test.py

@@ -7,7 +7,6 @@
 
 class TestCostMagicSquaresGame:
     def test_magic_squares_game_cost_fn(self):
-
         prep_nodes = [qnet.PrepareNode(1, range(4), qnet.bell_state_copies, 0)]
         meas_nodes = [
             qnet.MeasureNode(3, 4, [0, 1], qml.templates.subroutines.ArbitraryUnitary, 15),

test/cost/mermin_klyshko_inequality_test.py

@@ -37,7 +37,6 @@ class TestMerminKlyshkoInequality:
         ],
     )
     def test_mermin_klyshko_inputs_scalars(self, n, match_inputs, match_scalars):
-
         meas_inputs_list, scalars_list = qnet.mermin_klyshko_inputs_scalars(n)
 
         assert scalars_list == match_scalars
@@ -55,7 +54,6 @@ def test_num_terms_mermin_klyshk_inputs_scalars(self, n):
         assert len(scalars_list) == num_terms
 
     def test_mermin_klyshko_inequality_fn_CHSH_scenario(self):
-
         prep_nodes = [qnet.PrepareNode(1, [0, 1], qnet.ghz_state, 0)]
 
         meas_nodes = [

test/cost/mutual_info_test.py

@@ -30,7 +30,6 @@ class TestMutualInfoCostFn:
         ],
     )
     def test_mutual_info_cost_qubit_33(self, network_settings, priors, postmap, match):
-
         ansatz = qnet.NetworkAnsatz(
             [qnet.PrepareNode(3, [0], qnet.local_RY, 1)],
             [qnet.MeasureNode(1, 3, [0], qnet.local_RY, 1)],
@@ -41,7 +40,6 @@ def test_mutual_info_cost_qubit_33(self, network_settings, priors, postmap, matc
         assert np.isclose(mutual_info(*network_settings), match)
 
     def test_mutual_info_2_senders(self):
-
         ansatz = qnet.NetworkAnsatz(
             [
                 qnet.PrepareNode(2, [0], qnet.local_RY, 1),
@@ -62,7 +60,6 @@ class TestMutualInfoOptimimzation:
         "priors,match", [([np.ones(3) / 3], 0.9182), ([np.array([0.5, 0.5, 0])], 1)]
     )
     def test_mutual_info_opt_qubit_33(self, priors, match):
-
         ansatz = qnet.NetworkAnsatz(
             [qnet.PrepareNode(3, [0], qnet.local_RY, 1)],
             [qnet.MeasureNode(1, 3, [0], qnet.local_RY, 1)],

test/cost/nlocal_chain_bell_inequality_test.py

@@ -65,7 +65,6 @@ def test_nlocal_chain_22_cost_fn(self, parallel_flag):
     )
     @flaky(max_runs=5, min_passes=1)
     def test_parallel_nlocal_chain_grad_fn(self, natural_grad, prep_nodes, meas_nodes):
-
         chain_ansatz = qnet.NetworkAnsatz(prep_nodes, meas_nodes)
 
         np.random.seed(45)
@@ -82,8 +81,8 @@ def test_parallel_nlocal_chain_grad_fn(self, natural_grad, prep_nodes, meas_node
             assert np.allclose(grad, grad_match)
 
     @pytest.mark.parametrize("parallel_flag", [True, False])
+    @flaky(max_runs=5, min_passes=1)
     def test_chain_J22_fn(self, parallel_flag):
-
         prep_nodes = [
             qnet.PrepareNode(1, [0, 1], qnet.local_RY, 2),
             qnet.PrepareNode(1, [2, 3], qnet.local_RY, 2),
@@ -133,8 +132,8 @@ def test_chain_J22_fn(self, parallel_flag):
         assert np.isclose(4, J22(*settings))
 
     @pytest.mark.parametrize("parallel_flag", [True, False])
+    @flaky(max_runs=5, min_passes=1)
     def test_chain_I22_fn(self, parallel_flag):
-
         prep_nodes = [
             qnet.PrepareNode(1, [0, 1], qnet.local_RY, 2),
             qnet.PrepareNode(1, [2, 3], qnet.local_RY, 2),

test/cost/nlocal_star_bell_inequality_test.py

@@ -39,6 +39,7 @@ def trilocal_star_ry_ansatz(self):
         return qnet.NetworkAnsatz(prep_nodes, meas_nodes)
 
     @pytest.mark.parametrize("parallel_flag, nthreads", [(False, 4), (True, 4), (True, 3)])
+    @flaky(max_runs=5, min_passes=1)
     def test_bilocal_star_22_cost(self, parallel_flag, nthreads):
         bilocal_star_ansatz = self.bilocal_star_ry_ansatz()
         bilocal_22_cost = qnet.nlocal_star_22_cost_fn(
@@ -54,6 +55,7 @@ def test_bilocal_star_22_cost(self, parallel_flag, nthreads):
         assert np.isclose(bilocal_22_cost(*ideal_settings), -(np.sqrt(2)))
 
     @pytest.mark.parametrize("parallel_flag, nthreads", [(False, 4), (True, 4), (True, 5)])
+    @flaky(max_runs=5, min_passes=1)
     def test_trilocal_star_cost(self, parallel_flag, nthreads):
         trilocal_star_ansatz = self.trilocal_star_ry_ansatz()
         trilocal_22_cost = qnet.nlocal_star_22_cost_fn(

test/cost/qnodes_test.py

@@ -22,7 +22,6 @@ def test_parity_observable(self):
         assert qubit3_obs.wires == qml.wires.Wires([0, 1, 2])
 
     def test_local_parity_observables(self):
-
         measure_nodes = [
             qnet.MeasureNode(2, 2, range(3 * i, 3 * i + 3), lambda settings, wires: None, 3)
             for i in range(3)
@@ -60,7 +59,6 @@ def test_local_parity_expval_qnode(self):
         assert np.all(qnode([np.pi, np.pi, np.pi, 0]) == [1, -1])
 
     def test_global_parity_expval_qnode(self):
-
         prep_nodes = [
             qnet.PrepareNode(
                 1, [0, 1, 2, 3], lambda settings, wires: qml.BasisState(settings, wires=wires), 4