harvest_data_gauss.py

from circuits_IC import *

from IC_aux import sample_parameters, sample_function
import numpy as np
from argparse import ArgumentParser


parser = ArgumentParser()

parser.add_argument('--nqubits', type=int, default=4)
parser.add_argument('--nlayers', type=int, default = 4)
parser.add_argument('--model', type=int, default = 2)
parser.add_argument('--seed', type=int, default=0)
parser.add_argument('--extra_dim', type=int, default=2)
parser.add_argument('--data_size', type=int, default=5)

def main(nqubits, nlayers, model, seed, extra_dim, data_size):
    '''
    nqubits: size of the circuits
    nlayers: depth of the ansatz
    model: choose the model, from the ones available in circuits_IC.py
    seed: random seed for reproducibility
    extra_dim: overhead in the number of dimensions for computing IC, see [48]
    data_size: size of data x to compute hypothesis functions
    '''
    nameGAUSS = f'data/q{nqubits}_l{nlayers}_gauss{model}_y_r{seed}.csv'
    try: 
        np.loadtxt(nameGAUSS)
        print(nqubits, nlayers, model, seed, extra_dim, data_size, 'already done')
    except: 
        if model == 1:
            circuit = single_qubit(nqubits, nlayers)

        elif model == 2:
            circuit = single_qubit_and_entangling(nqubits, nlayers)

        elif model == 3:
            circuit = permutation_simple(nqubits, nlayers)

        elif model == 4:
            circuit = permutation_double(nqubits, nlayers)

        elif model == 5:
            circuit = permutation_composed(nqubits, nlayers)

        # Create circuits
        circuit.create_PQC()
        circuit.create_QML()

        np.random.seed(seed)

        if model in [1, 2]:
            n_params = len(circuit.PQC.get_parameters())
        elif model == 3:
            n_params = circuit.nlayers
        elif model in [4, 5]:
            n_params = 2 * circuit.nlayers

        nameX = f'data/q{nqubits}_l{nlayers}_model{model}_X_r{seed}.csv'
        theta = np.loadtxt(nameX)  # load parameters fro m PQC

        y_gauss = sample_function(lambda x: circuit.function(x, np.sqrt(np.pi)* np.random.randn(data_size)), theta) # Execute functions
        nameGAUSS = f'data/q{nqubits}_l{nlayers}_gauss{model}_y_r{seed}.csv' # Save data
        np.savetxt(nameGAUSS, y_gauss)
     

if __name__ == '__main__':
    args = vars(parser.parse_args())
    main(**args)