Handling OverflowError: math range error in Python with NumPy

Question

我在Python

中有以下代码

z =  [[complex(x,y) for x in range(1000)]for y in range(1000)]
nexp = np.frompyfunc(cmath.exp,1,1)
nexp(z)

我有 OverflowError：数学范围错误。

有没有办法找到数组的哪些元素出现错误并处理？我的意思是，如果我只是简单地遍历数组，我就可以做到这一点

for x in range(1000):
    for y in range(1000):
        try:
            z[x,y] = cmath.exp(z[x,y])
        except:
            z[x,y] = 0

Answer 1

为什么不使用 numpy 类型和方法来代替 complex 和 cmath？

演示：

In [302]: x = np.arange(701, 720) + 1j*np.arange(701, 720)

In [303]: y = np.exp(x)
~\Anaconda3_5.0\envs\py36\Scripts\ipython3:1: RuntimeWarning: overflow encountered in exp

这产生：

In [304]: y
Out[304]:
array([ -2.51188899e+304 -1.13634915e+304j,  -1.08996190e+304 -7.41453166e+304j,   1.53588478e+305 -1.33828078e+305j,   5.31686829e+305 +1.5
4758929e+305j,   4.26896569e+305 +1.44345018e+306j,
        -2.67470300e+306 +3.09645066e+306j,  -1.10110029e+307 -1.57025773e+306j,  -1.25800559e+307 -2.74923036e+307j,   4.44084142e+307 -6.9
1528370e+307j,               inf +1.34684655e+304j,
                     inf             +infj,              -inf             +infj,              -inf             +infj,              -inf
        -infj,               inf             -infj,
                     inf             -infj,               inf             +infj,              -inf             +infj,              -inf
        +infj])

所有无限元素：

In [305]: y[np.isinf(y)]
Out[305]:
array([ inf +1.34684655e+304j,  inf             +infj, -inf             +infj, -inf             +infj, -inf             -infj,  inf
    -infj,  inf             -infj,  inf             +infj,
       -inf             +infj, -inf             +infj])

第一个无限元的索引：

In [306]: np.isinf(y).argmax()
Out[306]: 9

Handling OverflowError: math range error in Python with NumPy

Handling OverflowError: math range error in Python with NumPy

python

error-handling

numpy

cmath