gpt4 book ai didi

python - matshow() 的替代方案

转载 作者:太空宇宙 更新时间:2023-11-03 14:58:56 24 4
gpt4 key购买 nike

我有一个 101x101 矩阵,我想以图形方式显示它。到目前为止,我使用了 matplotlib.pyplot 中的函数 matshow,如下所示:

import numpy as np
import random
import matplotlib.pyplot as plt

A = np.zeros((101,101))
# do stuff to randomly change some values of A
plt.ion()
plt.matshow(A,cmap='PuBuGn')
plt.colorbar()
plt.show()

输出如下所示: enter image description here

您应该将其视为物种之间的相互作用矩阵,正如您所见,只有三个物种存在强烈相互作用。这就是为什么考虑像我在 paper 中找到的下图这样的可视化看起来是合适的。 ,但我不知道这是否以及如何在 Python 中实现:

enter image description here

最佳答案

这应该可以解决问题:

import numpy as np
import matplotlib.pyplot as plt

def old_graph(A):
plt.matshow(A,cmap='PuBuGn')
plt.colorbar()
plt.title(r"abs$\left(\left[\mathbf{A}_{ij} \right ] \right )$ ; SIS=%d"%(sis,), va='bottom')
plt.show()

def new_graph(A, sis_list=np.zeros(0,int), symmetric=True, fig=None, pos=111):
#create and edit figure:
if fig is None:
fig = plt.figure()
ax = fig.add_subplot(pos, projection='polar')
ax.set_rgrids([1],[' '])
ax.set_rmax(1)
ax.set_thetagrids([])
ax.set_title(r"abs$\left(\left[\mathbf{A}_{ij} \right ] \right )$ ; SIS=%d"%(sis,), va='bottom')
colormap = plt.get_cmap('PuBuGn')

# make each species an angle value:
n_species = A.shape[0]
angles = np.linspace(0, 2*np.pi, n_species+1)
# the radius will always be r_max, and each line
# will always unite two points:
r = np.ones((2,))
# prepare list of lines to sort:
unordered_pairs_and_values = []
for index_line in xrange(n_species):
for index_column in xrange(index_line):
if symmetric:
value = A[index_line,index_column]
else: # not symmetric
value= abs(A[index_line,index_column]- A[index_column,index_line])
unordered_pairs_and_values.append([[angles[index_line],angles[index_column]], value])
# sort the lines (otherwise white lines would cover the 'important' ones):
ordered_pairs_and_values = sorted(unordered_pairs_and_values, key=lambda pair: pair[1])
# get the maximum value for scaling:
I_max = ordered_pairs_and_values[-1][1]
# plot every line in order:
for pair in ordered_pairs_and_values:
ax.plot(pair[0], r, color=colormap(pair[1]/I_max), linewidth=2, alpha=0.8)
# don't know how to add the colorbar:
#fig.colorbar(orientation='horizontal')
# mark the angles:
ax.plot(angles, np.ones(angles.shape), 'ko')
# mark the important angles (comment if you don't know which ones are these):
ax.plot(angles[sis_list], np.ones(sis_list.shape), 'ro')
fig.show()

if __name__ == '__main__':
n_species = 51
sis = 3 # strongly interacting species
sis_factor = 4.
A = np.zeros((n_species,n_species))
# do stuff to randomly change some values of A:
for index_line in xrange(n_species):
for index_column in xrange(index_line+1):
A[index_line,index_column] = np.random.random()
A[index_column,index_line] = A[index_line,index_column]

sis_list = np.random.randint(0,n_species,sis)
for species in sis_list:
A[species,:] *= sis_factor
A[:,species] *= sis_factor
for species2 in sis_list: # correct crossings
A[species,species2] /= sis_factor
# stuff to randomly change some values of A done
old_graph(A=A)
new_graph(A=A, sis_list=sis_list, symmetric=True)

旧图: old_graph

新图表: enter image description here我还是不知道:

  • 如何输入颜色栏,因为它需要使用颜色图绘制的绘图。
  • 如何在不重新缩放图形的情况下删除 r-ticks(您可以尝试取消注释 #ax.set_rticks([]))

并且情节在相互作用较少的物种中看起来更好: enter image description here

关于python - matshow() 的替代方案,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/40103855/

24 4 0
Copyright 2021 - 2024 cfsdn All Rights Reserved 蜀ICP备2022000587号
广告合作:1813099741@qq.com 6ren.com