qrisp.qaoa.QAOABenchmark.rank

QAOABenchmark.rank(metric='approx_ratio', print_res=False, average_repetitions=False)

Ranks the runs of the benchmark according to a given metric.

The default metric is approximation ratio. Similar to .evaluate, the metric can be user specified.

Parameters

metricstr or callable, optional

The metric according to which should be ranked. The default is “approx_ratio”.

Returns

list[dict]

List of dictionaries, where the first element has the highest rank.

Examples

We create a MaxCut instance and benchmark several parameters

from qrisp import *
from networkx import Graph
G = Graph()

G.add_edges_from([[0,3],[0,4],[1,3],[1,4],[2,3],[2,4]])

from qrisp.qaoa import maxcut_problem

max_cut_instance = maxcut_problem(G)

benchmark_data = max_cut_instance.benchmark(qarg = QuantumVariable(5),
                           depth_range = [3,4,5],
                           shot_range = [5000, 10000],
                           iter_range = [25, 50],
                           optimal_solution = "11100",
                           repetitions = 2
                           )

To rank the results, we call the according method:

print(benchmark_data.rank()[0])
#Yields: {'layer_depth': 5, 'circuit_depth': 44, 'qubit_amount': 5, 'shots': 10000, 'iterations': 50, 'counts': {'11100': 0.4909, '00011': 0.4909, '00010': 0.002, '11110': 0.002, '00001': 0.002, '11101': 0.002, '10000': 0.0015, '01000': 0.0015, '00100': 0.0015, '11011': 0.0015, '10111': 0.0015, '01111': 0.0015, '00000': 0.0001, '10010': 0.0001, '01010': 0.0001, '11010': 0.0001, '00110': 0.0001, '10110': 0.0001, '01110': 0.0001, '10001': 0.0001, '01001': 0.0001, '11001': 0.0001, '00101': 0.0001, '10101': 0.0001, '01101': 0.0001, '11111': 0.0001, '11000': 0.0, '10100': 0.0, '01100': 0.0, '10011': 0.0, '01011': 0.0, '00111': 0.0}, 'runtime': 1.4269020557403564, 'optimal_solution': '11100'}