PyMaxEnt/generate_gold_standards.py at master · saadgroup/PyMaxEnt · 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
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
#!/usr/bin/env python
"""
generate_gold_standards.py
--------------------------
Run this script to generate (or rebless) the gold standard lambda values
used by the regression suite.  Results are saved to gold_standards.npz.

Usage:
    python generate_gold_standards.py

When to re-run:
    - First time setup
    - After intentionally changing the algorithm and verifying the new
      output is correct (rebless)
"""

import warnings
import numpy as np
from scipy.integrate import quad
from scipy.stats import lognorm

try:
    from pymaxent import reconstruct, moments_c
except ImportError:
    from src.pymaxent import reconstruct, moments_c

warnings.filterwarnings("ignore")

# ---------------------------------------------------------------------------
# Distribution definitions (mirror of examples.ipynb)
# ---------------------------------------------------------------------------

def gauss(x):
    sigma, mu = 0.2, 1.0
    A = 1.0 / (sigma * np.sqrt(2 * np.pi))
    return A * np.exp(-(x - mu) ** 2 / (2 * sigma ** 2))

def gauss2(x):
    s0, m0 = 1.0/14.0, 1.0/4.0;  A0 = 1.0/(2.0*s0*np.sqrt(2*np.pi))
    s1, m1 = 1.0/20.0, 2.0/4.0;  A1 = 1.0/(2.0*s1*np.sqrt(2*np.pi))
    return (A0*np.exp(-(x-m0)**2/(2*s0**2))
          + A1*np.exp(-(x-m1)**2/(2*s1**2)))

def gauss3(x):
    s0, m0 = 1.0/14.0, 1.0/4.0;  A0 = 1.0/(2.0*s0*np.sqrt(2*np.pi))
    s1, m1 = 1.0/20.0, 2.0/4.0;  A1 = 1.0/(2.0*s1*np.sqrt(2*np.pi))
    s2, m2 = 1.0/20.0, 3.0/4.0;  A2 = 1.0/(2.0*s1*np.sqrt(2*np.pi))
    return (A0*np.exp(-(x-m0)**2/(2*s0**2))
          + A1*np.exp(-(x-m1)**2/(2*s1**2))
          + A2*np.exp(-(x-m2)**2/(2*s2**2)))

def beta_dist(x):
    a, b = 3, 9
    btm = quad(lambda X: X**(a-1)*(1-X)**(b-1), 0, 1)[0]
    return x**(a-1) * (1-x)**(b-1) / btm

def beta2_dist(x):
    a = b = 0.5
    btm = quad(lambda X: X**(a-1)*(1-X)**(b-1), 0, 1)[0]
    return x**(a-1) * (1-x)**(b-1) / btm

_ld = lognorm([0.25], loc=0.2)
def lognorm_pdf(x):
    return np.squeeze(_ld.pdf(x))

# ---------------------------------------------------------------------------
# Run all 13 cases and collect lambdas
# ---------------------------------------------------------------------------

gs = {}

print("Generating gold standards...")

# --- Discrete ---
mu = np.array([1]);  x = np.array([1,2,3,4,5,6])
sol, lam = reconstruct(mu=mu, rndvar=x)
gs["D1_lambdas"] = lam;  gs["D1_sol"] = sol
print(f"  D1  lambdas={lam}")

mu = [1, 3.5];  x = [1,2,3,4,5,6]
sol, lam = reconstruct(mu=mu, rndvar=x)
gs["D2_lambdas"] = lam;  gs["D2_sol"] = sol
print(f"  D2  lambdas={lam}")

pi = np.array([1,1,1,2,3,4]) / 12.0;  x = [1,2,3,4,5,6]
mu = [1, np.sum(pi * np.array(x))]
sol, lam = reconstruct(mu=mu, rndvar=x)
gs["D3_lambdas"] = lam;  gs["D3_sol"] = sol
print(f"  D3  lambdas={lam}")

# --- Continuous ---
mu = moments_c(gauss, 3, bnds=[0,2])
_, lam = reconstruct(mu=mu, bnds=[0,2])
gs["C1_lambdas"] = lam
print(f"  C1  lambdas={lam}")

mu = moments_c(gauss2, 5, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C2_lambdas"] = lam
print(f"  C2  lambdas={lam}")

mu = moments_c(gauss2, 10, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C3_lambdas"] = lam
print(f"  C3  lambdas={lam}")

mu = moments_c(gauss3, 5, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C4_lambdas"] = lam
print(f"  C4  lambdas={lam}")

mu = moments_c(gauss3, 13, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C5_lambdas"] = lam
print(f"  C5  lambdas={lam}")

mu = moments_c(beta_dist, 3, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C6a_lambdas"] = lam
print(f"  C6a lambdas={lam}")

mu = moments_c(beta_dist, 5, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C6b_lambdas"] = lam
print(f"  C6b lambdas={lam}")

mu = moments_c(beta2_dist, 3, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C7a_lambdas"] = lam
print(f"  C7a lambdas={lam}")

mu = moments_c(beta2_dist, 5, bnds=[0,1])
_, lam = reconstruct(mu=mu, bnds=[0,1])
gs["C7b_lambdas"] = lam
print(f"  C7b lambdas={lam}")

mu = moments_c(lognorm_pdf, 3, bnds=[0,5])
_, lam = reconstruct(mu=mu, bnds=[0,5])
gs["C8a_lambdas"] = lam
print(f"  C8a lambdas={lam}")

mu = moments_c(lognorm_pdf, 5, bnds=[0,5])
_, lam = reconstruct(mu=mu, bnds=[0,5])
gs["C8b_lambdas"] = lam
print(f"  C8b lambdas={lam}")

# ---------------------------------------------------------------------------
# Save
# ---------------------------------------------------------------------------

np.savez("gold_standards.npz", **gs)
print("\nSaved to gold_standards.npz")