TCNSA_final_project/driver.py at master · odedwer/TCNSA_final_project · 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
import Network as net
import numpy as np
from time import time
import matplotlib.pylab as plt
import scipy.linalg as linalg
from importlib import reload
from email_by_demand import send_mail
import warnings
# warnings.filterwarnings("error")
# %%
reload(net)
# %%
# networkS = NetworkSlow(32, 4, 2, -1.2, 0.1, 9000e-1, np.tanh, 50, 100, 5, 2e5, 0.1118, seed=97)
networkF = net.Network(1024, 2, 2, -1.2, 0.1, 9000., 50, 100, 5, 2e3, 0.1118, seed=97)

# %%
# first_W_F = networkF.W.copy()

coefs_F, delta_u_F = networkF.run_first_phase(with_noise=False,LIMIT=1000)
#%%
plt.figure()
plt.plot(networkF.coef_history, label='F')
plt.legend()
#%%
coefs2, delta_u2 = networkF.run_second_phase(9.65, 9.65, True)

# %%
plt.figure()
plt.plot(coefs2, label='F')
plt.legend()

# %%
W = network.W
P = network.P
# %%
c1 = ((network.P[0, 0, :] @ network.W[:, 0]) / network.P[0, 0, 0])
c2 = ((network.P[1, 0, :] @ network.W[:, 0]) / network.P[1, 0, 0])
c3 = ((network.P[2, 0, :] @ network.W[:, 0]) / network.P[2, 0, 0])
c4 = ((network.P[3, 0, :] @ network.W[:, 0]) / network.P[3, 0, 0])
# %%
c = ((P[:, 0, :] @ W[:, 0]) / P[:, 0, 0])
# %%
t = 2
N = 5
steps = 10
delta_u = (np.ones((steps, N)) * np.arange(1, steps + 1, 1)[:, np.newaxis]).ravel()
np.random.shuffle(delta_u)
delta_u.shape = (steps, N)

# %%
Ksp = np.arange(steps)
Ksm = np.arange(steps) + 10
Ksp_sum = Ksp @ delta_u
Ksm_sum = Ksm @ delta_u
Ksp_outer = np.outer(delta_u[-1], Ksp_sum).T
Ksm_outer = np.outer(delta_u[-1], Ksm_sum)
# %%
W = np.zeros((N, N))
for i in range(N):
    for j in range(N):
        W[i, j] = delta_u[-1][j] * Ksp_sum[i] + delta_u[-1][i] * Ksm_sum[j]
# %%
c = np.vstack([Ksp, Ksm])
c_sum = c @ delta_u
c_outer0 = np.outer(delta_u[-1], c_sum[0, :]).T
c_outer1 = np.outer(delta_u[-1], c_sum[1, :])
# %%
c_outer[1].T == Ksm_outer
# %%
x = np.linspace(-100, 100, 10000)
kernel = network.stdp_kernel(x)
# %%
plt.figure()
plt.plot(x, kernel)
# %%
from scipy.integrate import quad

integ = quad(lambda t: network.stdp_kernel(t), -2500, 2500)[0]

# %%
g = 0.1
xi = 0.1118
gamma = 1 / 9000.
Am = -1.2
Ap = 2
tao_m = 100
tao_p = 50
tao0 = 2e5

# %%
c = np


# %%
def default_stdp_kernel(delta_t):
    A = np.full_like(delta_t, Am)
    tao_arr = np.full_like(delta_t, tao_m)
    negative = delta_t < 0
    A[negative] = Ap
    tao_arr[negative] = tao_p
    return A * np.exp(-np.abs(delta_t) / tao_arr)


a = np.arange(-100, 100, 0.1)
b = default_stdp_kernel(a)
plt.figure()
plt.plot(a, b)