minor changes

Author: skushnir123

examples/benchmark_experiment.ipynb

@@ -11,7 +11,7 @@
     "import numpy as np\n",
     "import matplotlib.pyplot as plt\n",
     "from qhdopt import QHD\n",
-    "from qhdopt.utils.benchmark_utils import run_test, compare_coarse_and_refine, make_violin_plot"
+    "from benchmark_utils import run_test, compare_coarse_and_refine, make_violin_plot"
    ]
   },
   {

qhdopt/utils/benchmark_utils.py examples/benchmark_utils.py

@@ -30,26 +30,26 @@ def calc_h_and_J(qhd):
 def run_test(model, tol=1e-3):
     data_vector = np.zeros(16)
     # # Run QHD with post-processor = Ipopt
-    # response = model.optimize(verbose=1)
-    # qhd_ipopt_success_prob = response.get_success_probability()
-    # data_vector[0] = model.info["average_qpu_time"]
-    # data_vector[1] = model.info["post_processing_time"] / len(model.info["sample_times"])
-    # data_vector[12] = statistics.stdev(model.info["sample_times"][1:])
-    # data_vector[2] = qhd_ipopt_success_prob
-    # data_vector[3] = TTS((data_vector[0] + data_vector[1]), data_vector[2])
-    # best = response.minimum
-    # # Same QHD samples post-processed by TNC
-    # qhd_samples = [sample for sample in response.coarse_samples if sample is not None]
-    # response = model.classically_optimize(solver="TNC", initial_guess=qhd_samples)
-    #
-    # qhd_tnc_success_prob = response.get_success_probability(tol=tol, minimum=best)
-    #
-    #
-    # data_vector[4] = model.info["average_qpu_time"]
-    # data_vector[5] = model.info["refining_time"] / len(model.info["sample_times"])
-    # data_vector[13] = statistics.stdev(model.info["sample_times"][1:])
-    # data_vector[6] = qhd_tnc_success_prob
-    # data_vector[7] = TTS((data_vector[4] + data_vector[5]), data_vector[6])
+    response = model.optimize(verbose=1)
+    qhd_ipopt_success_prob = response.get_success_probability()
+    data_vector[0] = model.info["average_qpu_time"]
+    data_vector[1] = model.info["post_processing_time"] / len(model.info["sample_times"])
+    data_vector[12] = statistics.stdev(model.info["sample_times"][1:])
+    data_vector[2] = qhd_ipopt_success_prob
+    data_vector[3] = TTS((data_vector[0] + data_vector[1]), data_vector[2])
+    best = response.minimum
+    # Same QHD samples post-processed by TNC
+    qhd_samples = [sample for sample in response.coarse_samples if sample is not None]
+    response = model.classically_optimize(solver="TNC", initial_guess=qhd_samples)
+
+    qhd_tnc_success_prob = response.get_success_probability(tol=tol, minimum=best)
+
+
+    data_vector[4] = model.info["average_qpu_time"]
+    data_vector[5] = model.info["refining_time"] / len(model.info["sample_times"])
+    data_vector[13] = statistics.stdev(model.info["sample_times"][1:])
+    data_vector[6] = qhd_tnc_success_prob
+    data_vector[7] = TTS((data_vector[4] + data_vector[5]), data_vector[6])
 
     # Run Ipopt with random init
     response = model.classically_optimize(solver="IPOPT", num_shots=model.shots)

tests/test_backend.py

@@ -9,7 +9,7 @@ def test_backend_performance():
     bt = [3 / 4, -1 / 4]
     sol = -.75
     model = QHD.QP(Q, bt)
-    response = model.classically_optimize(solver="IPOPT", verbose=1)
+    response = model.classically_optimize(solver="TNC", verbose=1)
     assert abs(response.minimum - sol) < 1e-3
 
     model.qutip_setup(resolution=6, time_discretization=40)

tests/test_qhd.py

@@ -2,9 +2,8 @@
 
 from qhdopt import QHD
 import numpy as np
-from simuq.dwave import DWaveProvider
 
-from qhdopt.utils.benchmark_utils import calc_h_and_J
+from examples.benchmark_utils import calc_h_and_J
 
 
 def convert_key(tup: Tuple) -> str: