Created
March 24, 2020 21:26
-
-
Save xylar/ffe3317daa4203b7fe58397381969770 to your computer and use it in GitHub Desktop.
Multivariate power-law fit to the smoothing width based on number of iterations and nearest-neighbor weighting coefficient
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/env python | |
import numpy | |
import matplotlib.pyplot as plt | |
from scipy.ndimage.filters import convolve1d | |
from pandas import DataFrame | |
from sklearn import linear_model | |
nx = 2001 | |
x = numpy.linspace(-10., 10., nx) | |
dx = x[1] - x[0] | |
f = numpy.zeros((nx,)) | |
half = (nx-1)//2 | |
f[half] = 1. | |
outers = numpy.linspace(-2., -0.5, 40) | |
iterations = 1000 | |
iters = numpy.arange(1, iterations + 1) | |
Outers, Iters = numpy.meshgrid(outers, iters) | |
maxes = numpy.zeros(Outers.shape) | |
widths = numpy.zeros(Outers.shape) | |
for outerIndex, outer in enumerate(outers): | |
weights = numpy.array([10**outer, 1., 10**outer]) | |
weights = weights/weights.sum() | |
fsmooth = f | |
for iter in range(iterations): | |
fsmooth = convolve1d(fsmooth, weights, mode='wrap') | |
fmax = numpy.amax(fsmooth) | |
width = 2.*numpy.interp(0.5*fmax, fsmooth[-1:half:-1], x[-1:half:-1]) | |
maxes[iter, outerIndex] = fmax | |
widths[iter, outerIndex] = width/dx | |
mask = widths > 3.0 | |
data_dict = {'weight': Outers[mask], | |
'iteration': numpy.log10(Iters[mask]), | |
'width': numpy.log10(widths[mask])} | |
df = DataFrame(data_dict, columns=['weight', 'iteration', 'width']) | |
X = df[['weight', 'iteration']] | |
Y = df['width'] | |
# with sklearn | |
regr = linear_model.LinearRegression() | |
regr.fit(X, Y) | |
print('Intercept: \n', regr.intercept_, 10**regr.intercept_) | |
print('Coefficients: \n', regr.coef_) | |
#A = 10**regr.intercept_ | |
# p_weight = regr.coef_[0] | |
# p_iter = regr.coef_[1] | |
# width_fit = A*10**(Outers*p_weight)*Iters**p_iter | |
p_weight = 0.45 | |
p_iter = 0.5 | |
intercept = numpy.mean(data_dict['width'] - (p_weight*data_dict['weight'] + | |
p_iter*data_dict['iteration'])) | |
A = 10**intercept | |
width_fit = A*10**(Outers*p_weight)*Iters**p_iter | |
print('A: {}'.format(A)) | |
# A = 2.65 | |
# p_iter = 0.5 | |
# p_width = 2.25 | |
# weight = (width/(A*sqrt(iterations)))**p_width | |
vmin = numpy.amin(widths) | |
vmax = numpy.amax(widths) | |
plt.figure(1) | |
plt.pcolormesh(Outers, Iters, widths, vmin=vmin, vmax=vmax) | |
plt.colorbar() | |
plt.figure(2) | |
plt.pcolormesh(Outers, Iters, width_fit, vmin=vmin, vmax=vmax) | |
plt.colorbar() | |
plt.figure(3) | |
plt.pcolormesh(Outers, Iters, widths - width_fit) | |
plt.colorbar() | |
plt.figure(4) | |
plt.pcolormesh(Outers, Iters, mask) | |
plt.show() |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment