Question

以下代码崩溃并出现以下错误： ValueError：操作数无法与形状一起广播（5,5）（5,23）我知道这两个矩阵可以相乘，所以发生了什么？

def recon_temp(recon_pc_matrix, ew, eof, n, p):
    new_temps = np.zeros((p, p))
    for i in range(0,p-1):
        a = np.array([ew[i]*(recon_pc_matrix[i,0:p])]).T
        b = np.array([eof[i, 0:n]])
        new_temps = np.add(new_temps, np.multiply(a, b))
    return new_temps

a = np.array([[-0.1712077   0.30916178  0.3639138   0.86173197  0.00957432]
 [ 0.09886268 -0.56593983  0.75287704 -0.09865605  0.30558374]
 [-0.23426853  0.07244316  0.4005179  -0.23221139 -0.85178256]
 [ 0.1941591  -0.70654452 -0.339463    0.4397842  -0.39300358]
 [-0.93184473 -0.28227279 -0.15840833 -0.01878062  0.16291583]])
b = np.array([[-0.89934942  1.51224286  1.17335825 -0.08218979  0.98625991  1.13124679
   0.12082299 -1.45882345 -1.6897246   0.70884387 -1.21370037 -0.03640097
   0.25423117 -0.34026433 -0.94905651 -0.48517389  0.05775344  0.71212526
   1.11442191  0.31825772 -1.41012823 -0.59093546  1.06618286]
 [ 0.85264281  1.63168765  0.80196132  0.04534459  0.48561984 -0.11086797
   1.11591161 -0.24153581  0.65829793 -0.07172239  1.12920778 -0.52778775
  -2.2780791   0.90289084 -0.20018346 -0.20287233 -0.31416035 -0.83016084
  -0.8077091  -0.29606383 -1.19852534 -0.0768147  -0.46708139]
 [ 0.67870887 -0.01875505  0.26984334 -0.6223331   0.32559298 -0.94280004
  -0.81414199 -0.19893921  0.84574502  1.09421435 -0.15831082  1.81599507
   0.00648578 -1.21146774 -0.3179769   1.04116227 -0.11486882 -0.31177898
  -0.19821291  0.10022766 -1.09882425 -0.1149134  -0.05465211]
 [-0.03707971  0.53168148  0.41874141  0.03673735 -1.33893753  0.12464263
  -0.1343872   0.46679816 -0.75343034 -0.23104592 -0.22133639  0.28104257
  -0.37346672 -0.65672427  0.54146026 -0.13421286 -0.18973127 -1.0554372
  -0.44792916  0.85063804 -0.02023537  1.22571478  1.11649726]
 [ 0.31681713 -0.12923957 -0.04577132  0.2485488  -0.03239199 -0.07354265
   0.35762585 -0.97073981 -0.59382981  0.00892815  0.74089591 -0.04775075
   0.30722372 -0.51731235  0.43637896  0.21612401  0.83204179 -0.29326734
  -0.59640106  0.22724823  0.21728589 -0.64666786  0.03779609]])
c = np.array([ 0.92504802  0.7660462   0.540457    0.40724992  0.20059767])

temp = recon_temp(a, c, b, 23, 5)

在乘以a和b

时崩溃

Answer 1

np.multiply进行元素乘法运算。形状为(5, 5)和(5, 23)的两个数组不能按元素相乘，因此会出错。如果要执行矩阵乘法，请使用np.dot：

>>> a = np.random.rand(5, 5)
>>> b = np.random.rand(5, 23)
>>> np.multiply(a, b) <-- This gives a broadcasting error
>>> np.dot(a, b) <-- This gives the matrix multiplicaitn of a and b

为什么numpy乘以win（5,5）和（5,23）数组

1 个答案: