Homec4science
Diffusion PySONIC (old) (master) Browse

R4670/deprecated/update_lookups.py
043172721a44master

Last Change
Raw File

update_lookups.py
View Options

import os
import glob
import re
import pickle
import numpy as np
from sonic.utils import getNeuronsDict
# Get list of implemented neurons names
neurons = list(getNeuronsDict().keys())
# Define root directory and filename regular expression
root = 'C:/Users/admin/Google Drive/PhD/NICE model/sonic/sonic/lookups'
rgxp = re.compile('([A-Za-z]*)_lookups_a([0-9.]*)nm_f([0-9.]*)kHz.pkl')
# For each neuron
for neuron in neurons:
print('--------- Aggregating lookups for {} neuron -------'.format(neuron))
# Get filepaths from all lookups file in directory
neuron_root = '{}/{}'.format(root, neuron)
filepaths = glob.glob('{}/*.pkl'.format(neuron_root))
# Create empty frequencies list and empty directory to hold aggregated coefficients
freqs = []
agg_coeffs = {}
ifile = 0
# Loop through each lookup file
for fp in filepaths:
# Get information from filename (a, f)
filedir, filename = os.path.split(fp)
mo = rgxp.fullmatch(filename)
if mo:
name = mo.group(1)
a = float(mo.group(2)) * 1e-9 # nm
f = float(mo.group(3)) * 1e3 # Hz
else:
print('error: lookup file does not match regular expression pattern')
quit()
# Add lookup frequency to list
freqs.append(f)
print('f =', f * 1e-3, 'kHz')
# Open file and get coefficients dictionary
with open(fp, 'rb') as fh:
coeffs = pickle.load(fh)
# If first file: initialization steps
if ifile == 0:
# Get names of output coefficients
coeffs_keys = [ck for ck in coeffs.keys() if ck not in ['A', 'Q']]
# Save input coefficients vectors separately
A = coeffs['A']
Q = coeffs['Q']
print('nQ = ', len(Q))
# Initialize aggregating dictionary of output coefficients with empty lists
agg_coeffs = {ck: [] for ck in coeffs_keys}
# Append current coefficients to corresponding list in aggregating dictionary
for ck in coeffs_keys:
agg_coeffs[ck].append(coeffs[ck])
# Increment file index
ifile += 1
# Transform lists of 2D arrays into 3D arrays inside aggregating dictionary
for ck in agg_coeffs.keys():
# shape = agg_coeffs[ck][0].shape
# for tmp in agg_coeffs[ck]:
# if tmp.shape != shape:
# print('dimensions error:', shape, tmp.shape)
# quit()
print(ck, len(agg_coeffs[ck]), [tmp.shape for tmp in agg_coeffs[ck]])
agg_coeffs[ck] = np.array(agg_coeffs[ck])
# Add the 3 input vectors to the dictionary
agg_coeffs['f'] = np.array(freqs)
agg_coeffs['A'] = A
agg_coeffs['Q'] = Q
# Print out all coefficients names and dimensions
for ck in agg_coeffs.keys():
print(ck, agg_coeffs[ck].shape)
# Save aggregated lookups in file
filepath_out = '{}/{}_lookups_a{:.1f}nm.pkl'.format(root, neuron, a * 1e9)
print(filepath_out)
# with open(filepath_out, 'wb') as fh:
# pickle.dump(agg_coeffs, fh)
c4science · Help