发现这个神奇的用法,以后写博客就可以很好的演示矩阵乘法了
原文知乎
这里再分享一个可以把latex转成图片的在线网站quicklatex
markdown 显示矩阵
from IPython.display import display,Latex,Math
%matplotlib inline
import numpy as np
from IPython.core.interactiveshell import InteractiveShell
sh = InteractiveShell.instance()
def number_to_str(n,cut=):
ns=str(n)
format_='{0:.'+str(cut)+'f}'
if 'e' in ns or ('.' in ns and len(ns)>cut+):
return format_.format(n)
else:
return str(n)
def matrix_to_latex(mat,style='bmatrix'):
if type(mat)==np.matrixlib.defmatrix.matrix:
mat=mat.A
head=r'\begin{'+style+'}'
tail=r'\end{'+style+'}'
if len(mat.shape)==:
body=r'\\'.join([str(el) for el in mat])
return head+body+tail
elif len(mat.shape)==:
lines=[]
for row in mat:
lines.append('&'.join([number_to_str(el) for el in row])+r'\\')
s=head+' '.join(lines)+tail
return s
return None
sh.display_formatter.formatters['text/latex'].type_printers[np.ndarray]=matrix_to_latex
def show_decomposition(*args):
latex=''
for arg in args:
if type(arg)==str:
latex+=arg
else:
latex+=matrix_to_latex(arg)
latex='$'+latex+'$'
display(Math(latex))
效果如下
A = arange().reshape(, )
omega = random.randn(, )
show_decomposition(A,"*",omega,"=", np.dot(A,omega))
⎡⎣⎢⎢⎢⎢⎢⎢0510152016111621271217223813182349141924⎤⎦⎥⎥⎥⎥⎥⎥∗⎡⎣⎢⎢⎢⎢⎢⎢1.373280.27555−1.66708−0.655180.329181.41954−0.09314−1.816800.771331.37744⎤⎦⎥⎥⎥⎥⎥⎥=⎡⎣⎢⎢⎢⎢⎢⎢−3.70743−5.42866−7.14990−8.87113−10.592374.0970112.3888620.6807228.9725837.26443⎤⎦⎥⎥⎥⎥⎥⎥ [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 ] ∗ [ 1.37328 1.41954 0.27555 − 0.09314 − 1.66708 − 1.81680 − 0.65518 0.77133 0.32918 1.37744 ] = [ − 3.70743 4.09701 − 5.42866 12.38886 − 7.14990 20.68072 − 8.87113 28.97258 − 10.59237 37.26443 ]
基于以上代码,我修改了部分,让其能够显示一层嵌套的矩阵,或者显示矩阵代表的符号,这样方便写博客演示
def matrix_object_to_latex(col, *args):
latex='$ \\begin{matrix} '
i =
for arg in args:
i = i +
latex += " & "
if type(arg)==str:
latex+=arg
else:
latex+=matrix_to_latex(arg)
if i % col == :
latex += " \\\\ "
latex += "\end{matrix} $"
print(latex)
display(Math(latex))
return latex
测试代码
d =
matrix_a = np.arange(d*d).reshape(d,d)
matrix_b = np.arange(d*d).reshape(d,d)
tex = matrix_object_to_latex(, "A","*","B","=","C",matrix_a,"*",
matrix_b,"=",np.dot(matrix_a,matrix_b))
效果如下
A⎡⎣⎢⎢⎢0481215913261014371115⎤⎦⎥⎥⎥∗∗B⎡⎣⎢⎢⎢0481215913261014371115⎤⎦⎥⎥⎥==C⎡⎣⎢⎢⎢56152248344621742863986819632445274218362506⎤⎦⎥⎥⎥ A ∗ B = C [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 ] ∗ [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 ] = [ 56 62 68 74 152 174 196 218 248 286 324 362 344 398 452 506 ]