Modification to testing

convergence_functions.py

@@ -128,5 +128,7 @@ def calculate_convergence_ratios(nn, N, time, patterns):
     # Let's calculate how many of the patterns ended up in the one that they started closer too
     fraction_of_well_behaviour = [start - end for start, end in zip(closest_pattern_start, closest_pattern_end)].count(0)
     fraction_of_well_behaviour = fraction_of_well_behaviour * 1.0 / N
+    # import IPython
+    # IPython.embed()
 
     return fraction_of_convergence, fraction_of_well_behaviour

network.py

@@ -75,7 +75,6 @@ def __init__(self, hypercolumns, minicolumns, beta=None, w=None, o=None, s=None,
             self.w = np.zeros((self.n_units, self.n_units))
 
 
-
         # Set the adaptation to zeros by default
         self.a = np.zeros_like(self.o)
         # Set the clamping to zero by defalut
@@ -97,7 +96,7 @@ def empty_history(self):
                         'p_post': empty_array, 'p_co': empty_array_square, 'w': empty_array_square,
                         'beta': empty_array}
 
-    def reset_values(self):
+    def reset_values(self, keep_connectivity=False):
         self.o = np.ones(self.n_units) * (1.0 / self.minicolumns)
         self.s = np.log(np.ones(self.n_units) * (1.0 / self.minicolumns))
         self.z_pre = np.ones_like(self.o) * (1.0 / self.minicolumns)
@@ -108,8 +107,9 @@ def reset_values(self):
 
         self.a = np.zeros_like(self.o)
 
-        self.beta = np.log(np.ones_like(self.o) * (1.0 / self.minicolumns))
-        self.w = np.zeros((self.n_units, self.n_units))
+        if not keep_connectivity:
+            self.beta = np.log(np.ones_like(self.o) * (1.0 / self.minicolumns))
+            self.w = np.zeros((self.n_units, self.n_units))
 
     def randomize_pattern(self):
         self.o = self.prng.rand(self.n_units)

play.py

@@ -5,170 +5,18 @@
 import matplotlib.gridspec as gridspec
 from mpl_toolkits.axes_grid1 import make_axes_locatable
 
-from connectivity_functions import get_beta, get_w
-from connectivity_functions import calculate_probability, calculate_coactivations
-from data_transformer import build_ortogonal_patterns
-from network import BCPNN
+from connectivity_functions import softmax
 
 np.set_printoptions(suppress=True)
 
-hypercolumns = 3
-minicolumns = 3
+x = 0
+points = np.logspace(0, 4, num=1000)
+distance = np.zeros_like(points)
 
-patterns_dic = build_ortogonal_patterns(hypercolumns, minicolumns)
-patterns = list(patterns_dic.values())
+for index, y in enumerate(points):
+    print(y)
+    array = np.array((x, y))
+    aux = softmax(array, minicolumns=2)
+    distance[index] = np.linalg.norm(aux[0] - aux[1])
 
-P = calculate_coactivations(patterns)
-p = calculate_probability(patterns)
-
-w = get_w(P, p)
-beta = get_beta(p)
-
-dt = 0.001
-T_simulation = 10.0
-T_training = 1.0
-simulation_time = np.arange(0, T_simulation + dt, dt)
-training_time = np.arange(0, T_simulation + dt, dt)
-
-prng = np.random.RandomState(seed=0)
-
-tolerance = 1e-5
-
-g_a_set = np.arange(0, 110, 10)
-g_beta_set = np.arange(0, 22, 2)
-g_w_set = np.arange(0, 12, 2)
-
-nn = BCPNN(hypercolumns, minicolumns, beta, w, g_a=1, g_beta=1.0, g_w=1.0, g_I=10.0, prng=prng)
-
-
-if False:
-    for pattern in reversed(patterns):
-        I = pattern
-        nn.randomize_pattern()
-        print('---------')
-        print('I')
-        print(I)
-        # Run the network for one minut clamped to the patternr
-        nn.k = 1.0
-        nn.run_network_simulation(training_time, I=I)
-        # Run the network free
-        print('---------')
-        print('This should look like I')
-        print(nn.o)
-        nn.k = 0.0
-        nn.run_network_simulation(training_time)
-        print('---------')
-        print('---------')
-        print(nn.o)
-
-
-if True:
-    nn.randomize_pattern()
-    nn.k = 1.0
-    history = nn.run_network_simulation(training_time, save=True, I=patterns[0])
-    nn.run_network_simulation(training_time, save=True, I=None)
-    history = nn.run_network_simulation(training_time, save=True, I=patterns[1])
-    history = nn.run_network_simulation(training_time, save=True, I=patterns[2])
-
-    o = history['o']
-    s = history['s']
-    z_pre = history['z_pre']
-    p_pre = history['p_pre']
-    p_post = history['p_post']
-    p_co = history['p_co']
-    beta = history['beta']
-    w = history['w']
-    adaptation = history['a']
-
-    fig = plt.figure(figsize=(16, 12))
-
-    ax1 = fig.add_subplot(221)
-    im1 = ax1.imshow(o, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax1)
-    cax1 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im1, cax=cax1, ax=ax1)
-
-    ax2 = fig.add_subplot(222)
-    im2 = ax2.imshow(z_pre, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax2)
-    cax2 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im2, cax=cax2, ax=ax2)
-
-    ax3 = fig.add_subplot(223)
-    im3 = ax3.imshow(adaptation, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax3)
-    cax3 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im3, cax=cax3, ax=ax3)
-
-    ax4 = fig.add_subplot(224)
-    im4 = ax4.imshow(p_pre, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax4)
-    cax4 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im4, cax=cax4, ax=ax4)
-
-    plt.show()
-
-if False:
-    nn.randomize_pattern()
-    nn.k = 1.0
-
-    history = nn.run_network_simulation(training_time, save=True, I=patterns[0])
-    o = history['o']
-    s = history['s']
-    z_pre = history['z_pre']
-    p_pre = history['p_pre']
-    p_post = history['p_post']
-    p_co = history['p_co']
-    beta = history['beta']
-    w = history['w']
-    adaptation = history['a']
-
-    history = nn.run_network_simulation(training_time, save=True, I=patterns[1])
-    o = np.concatenate((o, history['o']))
-    s = np.concatenate((s, history['s']))
-    z_pre = np.concatenate((z_pre, history['z_pre']))
-    p_pre = np.concatenate((p_pre, history['p_pre']))
-    p_post = history['p_post']
-    p_co = history['p_co']
-    beta = history['beta']
-    w = history['w']
-    adaptation = np.concatenate((adaptation, history['a']))
-
-    history = nn.run_network_simulation(training_time, save=True, I=patterns[2])
-    o = np.concatenate((o, history['o']))
-    s = np.concatenate((s, history['s']))
-    z_pre = np.concatenate((z_pre, history['z_pre']))
-    p_pre = np.concatenate((p_pre, history['p_pre']))
-    p_post = history['p_post']
-    p_co = history['p_co']
-    beta = history['beta']
-    w = history['w']
-    adaptation = np.concatenate((adaptation, history['a']))
-
-    fig = plt.figure(figsize=(16, 12))
-
-    ax1 = fig.add_subplot(221)
-    im1 = ax1.imshow(o, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax1)
-    cax1 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im1, cax=cax1, ax=ax1)
-
-    ax2 = fig.add_subplot(222)
-    im2 = ax2.imshow(z_pre, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax2)
-    cax2 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im2, cax=cax2, ax=ax2)
-
-    ax3 = fig.add_subplot(223)
-    im3 = ax3.imshow(adaptation, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax3)
-    cax3 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im3, cax=cax3, ax=ax3)
-
-    ax4 = fig.add_subplot(224)
-    im4 = ax4.imshow(p_pre, aspect='auto', interpolation='None', cmap='viridis', vmax=1, vmin=0)
-    divider = make_axes_locatable(ax4)
-    cax4 = divider.append_axes('right', size='5%', pad=0.05)
-    fig.colorbar(im4, cax=cax4, ax=ax4)
-
-    plt.show()
+plt.plot(points, distance)

test_convergence.py

@@ -33,7 +33,7 @@ def test_basic_convergence(self):
         w = get_w(P, p)
         beta = get_beta(p)
 
-        dt = 0.01
+        dt = 0.001
         T = 1
         time = np.arange(0, T + dt, dt)