playground/random_survival_prob.py at master · x75/playground · 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
"""random_survival_prob.py

Compute survival probability of point agent with budget B and easyness goal area/total area
"""

import numpy as np
import matplotlib.pyplot as plt

# from matplotlib import rc
# rc('font',**{'family':'sans-serif','sans-serif':['Helvetica']})
# ## for Palatino and other serif fonts use:
# #rc('font',**{'family':'serif','serif':['Palatino']})
# rc('text', usetex=True)

# loop over
# goal area / total area ratios
# budget
# space dimensions
Budget = 1000
size_space = 1.0
size_goal = 0.35

binom_n = Budget + 0
binom_k = 1
binom_p = size_goal/size_space

ps = []
# dims = [1, 2, 3, 4, 5, 6, 10, 20, 50, 100]
# dims = np.exp(np.linspace(0, 4, 20)).astype(int)
# dims = np.power(2, np.linspace(0, 5, 12)).astype(int)
dims = np.linspace(1, 7, 7).astype(int)
dim_colors = np.linspace(1, 7, 7).astype(int)
cmap = plt.get_cmap('tab20c')

print(dims)

for dim in dims:
    vol_space = size_space**dim
    vol_goal = (size_goal**dim) * vol_space # np.pi *
    p = np.log(vol_goal/vol_space)
    print("p_%d = %f" % (dim, np.exp(p)))
    ps.append(p)

ps = np.array(ps)

# p_success = ps
# exp(p) is used a generalization of p(x^d) = goal-size / length-of-dimension^d
p_fail_dim = []
Budget_reqs = []
for i, dim in enumerate(dims):
    p = ps[i]
    Budget_req = Budget * (1-np.exp(p))
    print("Budget_req", Budget_req)
    Budget_reqs.append(Budget_req)
    # p_fail = np.random.binomial(n = binom_n, p = 1 - np.exp(p), size = 1000)
    p_fail = np.random.binomial(n = binom_n, p = np.exp(p), size = 1000)
    p_fail_dim.append(p_fail)

p_fail_dim = np.array(p_fail_dim)
Budget_reqs = np.array(Budget_reqs)

# print p_fail_dim.shape

# p_nohit

# fig = plt.figure()

# plt.subplot(211)

row_num = 2
row_logp = 0
row_logp_span = 1
row_hists = 1

ax = plt.subplot2grid((row_num, len(dims)), (row_logp, 0), colspan=len(dims), rowspan=row_logp_span)
# ax.plot(dims, ps, "ko")
for i, dim in enumerate(dims):
    # ax.plot([dim], ps[i], color=cmap.colors[i], marker="o", markersize=10, alpha=0.7)
    ax.plot([dim], np.exp(ps[i]), color=cmap.colors[i], marker="o", markersize=10, alpha=0.7)
    ax.text(dim + 0.1, ps[i]-0.25, round(np.exp(ps[i]), 2))
# ax.set_title('Number of failures in episodes of budget-size = 1000')

# ax.plot(dims, (Budget_reqs - np.min(Budget_reqs))/np.max(Budget_reqs), "ro")
ax.set_xlabel('Number of state space dimensions d')
ax.set_ylabel('p')
# ax.set_ylabel('log(p)')
# ax.set_yscale('log')
ylim = ax.get_ylim()
ax.set_ylim((ylim[0] - 0.04, ylim[1] + 0.04))

# plt.subplot(212)
ax = plt.subplot2grid((row_num, len(dims)), (row_hists, 0), colspan=len(dims))
for i, dim in enumerate(dims):
    # ax = plt.subplot2grid((row_num, len(dims)), (row_hists, i), colspan=1)
    h, bins = np.histogram(p_fail_dim[i], bins = 'auto', density = False)
    # print "h", h, "bins", bins
    # h_ = h / np.sum(h)
    h_ = h.copy()
    bins_ = bins[:-1] + np.diff(bins)
    # plt.bar(left = bins_, height = h_, bottom = 0, width = 1.0)
    # ax.bar(x = bins_, height = h_, bottom = 0, width = 0.125 * max(5.0, (bins_[-1] - bins_[0])), color='black')
    ax.bar(x = bins_, height = h_, bottom = 0, width = 10.0, color=cmap.colors[i], alpha=0.5)
    # ax.plot([Budget_reqs[i]] * 2, [0, 1.0], "r-")
    # print ax.get_xlim()

    # xlim = np.array(ax.get_xlim())
    # if xlim[1] - xlim[0] < 10.0:
    #     xlim[1] += 1.
    #     xlim[0] = xlim[1] - 10
    # ax.set_xlim(xlim)
    # ax.set_xticks(xlim)
    # ax.set_xticklabels([int(xlim[0]), min(10000, int(xlim[1]))], fontsize=6)

    # ax.set_xlim((500 - 10, binom_n + 10))
    # ax.set_ylim((0, 0.4))
    # ax.set_yscale('log')

    # if i > (len(dims)-2):
    #     ax.set_yticklabels([])
    # plt.hist(p_fail_dim[i], bins = 'auto', normed = True)
    # plt.gca().set_yscale('log')
# plt.ylabel('log(p)')
ax.set_xlabel('Number of successes over 1000 episodes of 1000 steps each')
ax.set_ylabel('Success count')

hspace = 0.5
wspace = 0.3

fig = ax.figure
# fig.suptitle("Probability of success for random strategy with $ \\rho = {0} $".format(binom_p))
fig.suptitle("Probability of success for random strategy for p = {0}".format(binom_p))
fig.subplots_adjust(bottom=0.12, wspace=wspace, hspace=hspace)
fig.set_size_inches((12, 2.5*row_num))
fig.savefig('random_survival_prob_v4.pdf', dpi = 300, bbox_inches = None) # 'tight'
# fig.savefig('random_survival_prob.pdf', dpi = 300, bbox_inches = None)

plt.show()