MPF_Potts/Run_Me.py at master · poonehm/MPF_Potts · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
"""
==================
MPF on Potts Model
==================

There are 4 steps:

   1- Generate the Original coupling matrix randomly

   2- Generate a sample from the above matrix

   3- Fit MPF to the sample and find the Fit coupling matrix

   4- Compare the Original and Fit coupling matrices

"""


from Gibbs_Potts import sampler
from MPF_Fitter import MPF
from Plots import coupling, covariance

import numpy as np


def main():
    """
    Parameters
    ----------


    q       : Potts model parameter -- q equals 2 converges to Ising model

    dim     : size of the coupling matrices

    nsample : sample size -- for bigger q should be bigger

    seed    : for consistancy

    """

    q = 10

    dim = 40

    nsamples = 50000

    seed = 13


    # 1- Generate the original coupling matrix randomly

    rng = np.random.RandomState(seed)

    J_orig = rng.randn(dim, dim)/np.sqrt(dim)
    np.fill_diagonal(J_orig, 0.)

    J_orig = (J_orig + J_orig.T) / 2.


    # 2- Generate a sample from the above matrix

    o = sampler(q, dim, rng)
    samples_orig = o.sampler(J_orig, nsamples)


    # 3- Fit MPF to the sample and find the Fit coupling matrix

    J_fit = o.minimizer(samples_orig)


    # 4- Compare the Original and Fit coupling matrices

    coupling(J_orig, J_fit)


    # Also compare the covariance of the samples from Original and Fit coupling matrices

    samples_fit = o.sampler(J_fit, nsamples)
    covariance(J_orig, J_fit, samples_orig, samples_fit)

    print (J_orig - J_fit).min(), (J_orig - J_fit).max()


if __name__ == '__main__':

    main()