Numpy how to vectorize python function use np.cumprod-CodePudding

I have a python function:

pval = np.array([1,1,1,1,1,0.999709, 0.99973,0.999743,0.999706, 0.999675, 0.99965,  0.999629])
age1=4
age2=8

def getnpxtocert(mt, age, age2):
    val = mt[age]
    for i in range(age   1,age2):
        val = val * mt[i]
    return val

getnpxtocertv(pval,age1,age2)

The output is:

0.9991822227268075

And then I tried to use cumprod to vectorize it:

def getnpxtocertv(mt, age, age2):
    return (mt[age]*np.cumprod(mt[age 1:age2])).sum()

getnpxtocert(pval,age1,age2)

But the output is:

2.998330301296807

What did I wrong?Any friend can help?

CodePudding user response：

You don't need cumprod and sum. Just use prod:

def getnpxtocert_v2(mt, age, age2):
    return np.prod(mt[age:age2])

Comparison:

In [23]: getnpxtocert(pval, age1, age2)
Out[23]: 0.9991822227268075

In [24]: getnpxtocert_v2(pval, age1, age2)
Out[24]: 0.9991822227268075

cumprod takes an array [x0, x1, x2, ...] and returns an array with the same length containing [x0, x0*x1, x0*x1*x2, ...]. prod returns the scalar that is the product of all the elements x0*x1*x2*....