使用非零值的索引错误地覆盖了数组

Question

我有一个遵循泊松分布的数字数组，我想覆盖每个非零值本身乘以某个随机数

from scipy import stats
import numpy as np

jumps = stats.poisson(mu=0.05).rvs(200) # generate 200 random poisson numbers
idx = np.where(jumps!=0)[0] # get the index where jumps is non zero
for i in idx:
    # want to overwrite the i'th index in jumps with itself
    # multiplied by a random float
    jumps[i] = jumps[i]*np.random.normal()

print(jumps) # we get only 0,-1,1
>>> [0 0 -1 ....
     0 0 0]

即使我包含一个临时变量，我仍然只能得到0，-1,1。

for i in idx:
    # want to overwrite the i'th index in jumps with itself
    # multiplied by a random float
    tmp = jumps[i]*np.random.normal()
    jumps[i] = tmp

print(jumps) # we still get only 0,-1,1
>>> [0 0 -1 ....
     0 0 0]

如果在每次迭代中打印出tmp，您会发现它工作正常。但是由于某些原因，jumps[i]没有被正确覆盖。

这是怎么回事？