bga_elo/sim100.py at master · ffallrain/bga_elo · 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
#!/usr/bin/python
import sys,os
import random as rd
import numpy as np
import math
import matplotlib.pyplot as plt

N_p = 100
K = 20

Strength_Scale = 120

players = np.array( range( N_p ) )

if False:
    istrength = np.zeros( (N_p,) )
    for i in players:
        istrength[i] = rd.gauss(0,1.5)
    avg = np.mean(istrength)
    istrength = (istrength - avg )*Strength_Scale + 1500

    ## log data
    ofp = open("players_elo.data",'w')
    for i in players:
        ofp.write("%g\n"%istrength[i])
    ofp.close()

else:
    istrength = np.array ( open("players_elo.data").readlines(), dtype=np.float32 )

# istrength = [ -1,1 ]

elo = np.zeros( (N_p,) ) + 1500

def run(i,j):

    ## Expectation
    p_iwin = 1.0/( 1+10**((istrength[j]-istrength[i])/400) )

    ## Simulation
    dice = rd.random()
    if dice < p_iwin : # i wins
        resulti,resultj = 1,0
    elif dice > p_iwin:               # j wins
        resulti,resultj = 0,1
    else:
        resulti,resultj = 0.5, 0.5   # Draw

    ## Analysis result
    pelo_iwin = 1.0/( 1+10**( (elo[j]-elo[i])/400 ) )
    diff = K*(resulti-pelo_iwin)
    elo[i] = elo[i] + diff
    elo[j] = elo[j] - diff

    return (diff,  )


def one_game():
    ## count
    global total_count

    ## random pick P
    i,j = rd.sample(players,2)

    ## would the g played?
    if True:
        if True: # exponent
            dice = rd.random()
            diff_strength = abs(elo[i]-elo[j])
            criteria = math.exp( - diff_strength / 200. )
            if dice <= criteria :
                flag = True
            else:
                flag = False
        else:
            diff_strength = abs(elo[i]-elo[j])
            if diff_strength > 200 :
                flag = False
            else:
                flag = True

    else:
        flag = True

    result = run(i,j)

    ## output
    if flag:
        total_count += 1
        # print ">>>>> The %d game, winner get %8.3f elo"%(total_count,result[0])

    return

def simulate( N ):
    for i in range(N):
        one_game()


if True: ## main

    plt.figure(figsize=(9,9) )

    colors = ['g','r','b','black','purple']
    for i in (200000, ):
        color = colors.pop(0)
        total_count = 0
        simulate( i )

        ## log simulated elo
        ofp = open("simulated_elo.data",'w')
        for i in players:
            ofp.write("%g\n"%elo[i])
        ofp.close()

        ## plot elo
        plt.scatter( istrength,elo, color = color , alpha = 0.5, label = "%d games played "%total_count )

        ## plot linear fit
        A = np.vstack( [istrength, np.ones(len(istrength))] ).T
        m,c = np.linalg.lstsq(A,elo)[0]
        x = np.linspace(1000,2200,1000)
        y = m*x+c
        plt.plot(x,y,color = color ,label="y = %g*x%+g"%(m,c) ,linewidth=2)

    plt.legend(loc='best')
    plt.xlabel("Intrinsic Elo")
    plt.ylabel("Simulated Elo")
    plt.xlim( (1000,2100) )
    plt.ylim( (1000,2100) )

    plt.show()