import matplotlib
import matplotlib.pyplot as plt
import numpy as np
[文档]def plot_2d_heatmap(array: np.ndarray, title: str, xlabel: str, ylabel: str, int_x_ticks=True, int_y_ticks=True,
plot_colorbar=True, colorbar_y_label='magnitude', x_max=None, figsize=(12, 8), dpi=200):
'''
:param array: shape=[T, N]的任意数组
:param title: 热力图的标题
:param xlabel: 热力图的x轴的label
:param ylabel: 热力图的y轴的label
:param int_x_ticks: x轴上是否只显示整数刻度
:param int_y_ticks: y轴上是否只显示整数刻度
:param plot_colorbar: 是否画出显示颜色和数值对应关系的colorbar
:param colorbar_y_label: colorbar的y轴label
:param x_max: 横轴的最大刻度。若设置为 ``None``，则认为横轴的最大刻度是 ``array.shape[1]``
:param dpi: 绘图的dpi
:return: 绘制好的figure
绘制一张二维的热力图。可以用来绘制一张表示多个神经元在不同时刻的电压的热力图，示例代码：
.. code-block:: python
import torch
from spikingjelly.activation_based import neuron
from spikingjelly import visualizing
from matplotlib import pyplot as plt
import numpy as np
lif = neuron.LIFNode(tau=100.)
x = torch.rand(size=[32]) * 4
T = 50
s_list = []
v_list = []
for t in range(T):
s_list.append(lif(x).unsqueeze(0))
v_list.append(lif.v.unsqueeze(0))
s_list = torch.cat(s_list)
v_list = torch.cat(v_list)
visualizing.plot_2d_heatmap(array=np.asarray(v_list), title='Membrane Potentials', xlabel='Simulating Step',
ylabel='Neuron Index', int_x_ticks=True, x_max=T, dpi=200)
plt.show()
.. image:: ./_static/API/visualizing/plot_2d_heatmap.*
:width: 100%
'''
if array.ndim != 2:
raise ValueError(f"Expected 2D array, got {array.ndim}D array instead")
fig, heatmap = plt.subplots(figsize=figsize, dpi=dpi)
if x_max is not None:
im = heatmap.imshow(array.T, aspect='auto', extent=[-0.5, x_max, array.shape[1] - 0.5, -0.5])
else:
im = heatmap.imshow(array.T, aspect='auto')
heatmap.set_title(title)
heatmap.set_xlabel(xlabel)
heatmap.set_ylabel(ylabel)
heatmap.xaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_x_ticks))
heatmap.yaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_y_ticks))
heatmap.xaxis.set_minor_locator(matplotlib.ticker.NullLocator())
heatmap.yaxis.set_minor_locator(matplotlib.ticker.NullLocator())
if plot_colorbar:
cbar = heatmap.figure.colorbar(im)
cbar.ax.set_ylabel(colorbar_y_label, rotation=90, va='top')
cbar.ax.yaxis.set_minor_locator(matplotlib.ticker.NullLocator())
return fig
[文档]def plot_2d_bar_in_3d(array: np.ndarray, title: str, xlabel: str, ylabel: str, zlabel: str, int_x_ticks=True, int_y_ticks=True, int_z_ticks=False, dpi=200):
'''
:param array: shape=[T, N]的任意数组
:param title: 图的标题
:param xlabel: x轴的label
:param ylabel: y轴的label
:param zlabel: z轴的label
:param int_x_ticks: x轴上是否只显示整数刻度
:param int_y_ticks: y轴上是否只显示整数刻度
:param int_z_ticks: z轴上是否只显示整数刻度
:param dpi: 绘图的dpi
:return: 绘制好的figure
将shape=[T, N]的任意数组，绘制为三维的柱状图。可以用来绘制多个神经元的脉冲发放频率，随着时间的变化情况，示例代码：
.. code-block:: python
import torch
from spikingjelly import visualizing
from matplotlib import pyplot as plt
Epochs = 5
N = 10
firing_rate = torch.zeros(Epochs, N)
init_firing_rate = torch.rand(size=[N])
for i in range(Epochs):
firing_rate[i] = torch.softmax(init_firing_rate * (i + 1) ** 2, dim=0)
visualizing.plot_2d_bar_in_3d(firing_rate.numpy(), title='spiking rates of output layer', xlabel='neuron index',
ylabel='training epoch', zlabel='spiking rate', int_x_ticks=True, int_y_ticks=True,
int_z_ticks=False, dpi=200)
plt.show()
.. image:: ./_static/API/visualizing/plot_2d_bar_in_3d.png
也可以用来绘制一张表示多个神经元在不同时刻的电压的热力图，示例代码：
.. code-block:: python
import torch
from spikingjelly import visualizing
from matplotlib import pyplot as plt
from spikingjelly.activation_based import neuron
neuron_num = 4
T = 50
lif_node = neuron.LIFNode(tau=100.)
w = torch.rand([neuron_num]) * 10
v_list = []
for t in range(T):
lif_node(w * torch.rand(size=[neuron_num]))
v_list.append(lif_node.v.unsqueeze(0))
v_list = torch.cat(v_list)
visualizing.plot_2d_bar_in_3d(v_list, title='voltage of neurons', xlabel='neuron index',
ylabel='simulating step', zlabel='voltage', int_x_ticks=True, int_y_ticks=True,
int_z_ticks=False, dpi=200)
plt.show()
.. image:: ./_static/API/visualizing/plot_2d_bar_in_3d_1.png
'''
if array.ndim != 2:
raise ValueError(f"Expected 2D array, got {array.ndim}D array instead")
fig = plt.figure(dpi=dpi)
ax = fig.add_subplot(111, projection='3d')
ax.set_title(title)
colormap = plt.get_cmap('tab10') # cmap的种类参见https://matplotlib.org/gallery/color/colormap_reference.html
array_T = array.T
xs = np.arange(array_T.shape[1])
for i in range(array_T.shape[0]):
ax.bar(xs, array_T[i], i, zdir='x', color=colormap(i % 10), alpha=0.8)
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
ax.set_zlabel(zlabel)
ax.xaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_x_ticks))
ax.yaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_y_ticks))
ax.zaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_z_ticks))
return fig
[文档]def plot_1d_spikes(spikes: np.asarray, title: str, xlabel: str, ylabel: str, int_x_ticks=True, int_y_ticks=True,
plot_firing_rate=True, firing_rate_map_title='firing rate', figsize=(12, 8), dpi=200):
'''
:param spikes: shape=[T, N]的np数组，其中的元素只为0或1，表示N个时长为T的脉冲数据
:param title: 热力图的标题
:param xlabel: 热力图的x轴的label
:param ylabel: 热力图的y轴的label
:param int_x_ticks: x轴上是否只显示整数刻度
:param int_y_ticks: y轴上是否只显示整数刻度
:param plot_firing_rate: 是否画出各个脉冲发放频率
:param firing_rate_map_title: 脉冲频率发放图的标题
:param dpi: 绘图的dpi
:return: 绘制好的figure
画出N个时长为T的脉冲数据。可以用来画N个神经元在T个时刻的脉冲发放情况，示例代码：
.. code-block:: python
import torch
from spikingjelly.activation_based import neuron
from spikingjelly import visualizing
from matplotlib import pyplot as plt
import numpy as np
lif = neuron.LIFNode(tau=100.)
x = torch.rand(size=[32]) * 4
T = 50
s_list = []
v_list = []
for t in range(T):
s_list.append(lif(x).unsqueeze(0))
v_list.append(lif.v.unsqueeze(0))
s_list = torch.cat(s_list)
v_list = torch.cat(v_list)
visualizing.plot_1d_spikes(spikes=np.asarray(s_list), title='Membrane Potentials', xlabel='Simulating Step',
ylabel='Neuron Index', dpi=200)
plt.show()
.. image:: ./_static/API/visualizing/plot_1d_spikes.*
:width: 100%
'''
if spikes.ndim != 2:
raise ValueError(f"Expected 2D array, got {spikes.ndim}D array instead")
spikes_T = spikes.T
if plot_firing_rate:
fig = plt.figure(tight_layout=True, figsize=figsize, dpi=dpi)
gs = matplotlib.gridspec.GridSpec(1, 5)
spikes_map = fig.add_subplot(gs[0, 0:4])
firing_rate_map = fig.add_subplot(gs[0, 4])
else:
fig, spikes_map = plt.subplots()
spikes_map.set_title(title)
spikes_map.set_xlabel(xlabel)
spikes_map.set_ylabel(ylabel)
spikes_map.xaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_x_ticks))
spikes_map.yaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=int_y_ticks))
spikes_map.xaxis.set_minor_locator(matplotlib.ticker.NullLocator())
spikes_map.yaxis.set_minor_locator(matplotlib.ticker.NullLocator())
spikes_map.set_xlim(-0.5, spikes_T.shape[1] - 0.5)
spikes_map.set_ylim(-0.5, spikes_T.shape[0] - 0.5)
spikes_map.invert_yaxis()
N = spikes_T.shape[0]
T = spikes_T.shape[1]
t = np.arange(0, T)
t_spike = spikes_T * t
mask = (spikes_T == 1) # eventplot中的数值是时间发生的时刻，因此需要用mask筛选出
colormap = plt.get_cmap('tab10') # cmap的种类参见https://matplotlib.org/gallery/color/colormap_reference.html
for i in range(N):
spikes_map.eventplot(t_spike[i][mask[i]], lineoffsets=i, colors=colormap(i % 10))
if plot_firing_rate:
firing_rate = np.mean(spikes_T, axis=1, keepdims=True)
max_rate = firing_rate.max()
min_rate = firing_rate.min()
firing_rate_map.yaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=True))
firing_rate_map.yaxis.set_minor_locator(matplotlib.ticker.NullLocator())
firing_rate_map.imshow(firing_rate, cmap='magma', aspect='auto')
for i in range(firing_rate.shape[0]):
firing_rate_map.text(0, i, f'{firing_rate[i][0]:.2f}', ha='center', va='center', color='w' if firing_rate[i][0] < 0.7 * max_rate or min_rate == max_rate else 'black')
firing_rate_map.get_xaxis().set_visible(False)
firing_rate_map.set_title(firing_rate_map_title)
return fig
[文档]def plot_2d_feature_map(x3d: np.asarray, nrows, ncols, space, title: str, figsize=(12, 8), dpi=200):
'''
:param x3d: shape=[C, W, H]，C个尺寸为W * H的矩阵。这样的矩阵一般来源于卷积层后的脉冲神经元的输出
:param nrows: 画成多少行
:param ncols: 画成多少列
:param space: 矩阵之间的间隙
:param title: 图的标题
:param figsize: 图片大小
:param dpi: 绘图的dpi
:return: 一个figure，将C个矩阵全部画出，然后排列成nrows行ncols列
将C个尺寸为W * H的矩阵，全部画出，然后排列成nrows行ncols列。这样的矩阵一般来源于卷积层后的脉冲神经元的输出，通过这个函数\\
可以对输出进行可视化。示例代码：
.. code-block:: python
from spikingjelly import visualizing
import numpy as np
from matplotlib import pyplot as plt
C = 48
W = 8
H = 8
spikes = (np.random.rand(C, W, H) > 0.8).astype(float)
visualizing.plot_2d_feature_map(spikes=spikes, nrows=6, ncols=8, space=2, title='Spiking Feature Maps', dpi=200)
plt.show()
.. image:: ./_static/API/visualizing/plot_2d_feature_map.*
:width: 100%
'''
if x3d.ndim != 3:
raise ValueError(f"Expected 3D array, got {x3d.ndim}D array instead")
C = x3d.shape[0]
assert nrows * ncols == C, 'nrows * ncols != C'
h = x3d.shape[1]
w = x3d.shape[2]
y = np.ones(shape=[(h + space) * nrows, (w + space) * ncols]) * x3d.max().item()
index = 0
for i in range(space // 2, y.shape[0], h + space):
for j in range(space // 2, y.shape[1], w + space):
y[i:i + h, j:j + w] = x3d[index]
index += 1
fig, maps = plt.subplots(figsize=figsize, dpi=dpi)
maps.set_title(title)
maps.imshow(y, cmap='gray')
maps.get_xaxis().set_visible(False)
maps.get_yaxis().set_visible(False)
return fig, maps
[文档]def plot_one_neuron_v_s(v: np.ndarray, s: np.ndarray, v_threshold=1.0, v_reset=0.0,
title='$V[t]$ and $S[t]$ of the neuron', figsize=(12, 8), dpi=200):
'''
:param v: shape=[T], 存放神经元不同时刻的电压
:param s: shape=[T], 存放神经元不同时刻释放的脉冲
:param v_threshold: 神经元的阈值电压
:param v_reset: 神经元的重置电压。也可以为 ``None``
:param title: 图的标题
:param dpi: 绘图的dpi
:return: 一个figure
绘制单个神经元的电压、脉冲随着时间的变化情况。示例代码：
.. code-block:: python
import torch
from spikingjelly.activation_based import neuron
from spikingjelly import visualizing
from matplotlib import pyplot as plt
lif = neuron.LIFNode(tau=100.)
x = torch.Tensor([2.0])
T = 150
s_list = []
v_list = []
for t in range(T):
s_list.append(lif(x))
v_list.append(lif.v)
visualizing.plot_one_neuron_v_s(v_list, s_list, v_threshold=lif.v_threshold, v_reset=lif.v_reset,
dpi=200)
plt.show()
.. image:: ./_static/API/visualizing/plot_one_neuron_v_s.*
:width: 100%
'''
fig = plt.figure(figsize=figsize, dpi=dpi)
ax0 = plt.subplot2grid((3, 1), (0, 0), rowspan=2)
ax0.set_title(title)
T = s.shape[0]
t = np.arange(0, T)
ax0.plot(t, v)
ax0.set_xlim(-0.5, T - 0.5)
ax0.set_ylabel('voltage')
ax0.axhline(v_threshold, label='$V_{threshold}$', linestyle='-.', c='r')
if v_reset is not None:
ax0.axhline(v_reset, label='$V_{reset}$', linestyle='-.', c='g')
ax0.legend(frameon=True)
t_spike = s * t
mask = (s == 1) # eventplot中的数值是时间发生的时刻，因此需要用mask筛选出
ax1 = plt.subplot2grid((3, 1), (2, 0))
ax1.eventplot(t_spike[mask], lineoffsets=0, colors='r')
ax1.set_xlim(-0.5, T - 0.5)
ax1.set_xlabel('simulating step')
ax1.set_ylabel('spike')
ax1.set_yticks([])
ax1.xaxis.set_major_locator(matplotlib.ticker.MaxNLocator(integer=True))
return fig, ax0, ax1