计算每个网格网格点的值数组

Question

I have a function that depends on some parameters and which output is an array. For example:

def my_func(xs,param1,param2,param3):
    values = xs**param1 + xs*param2**2 + param3*xs
    return values

where xs is an array with values. Suposse I also have a list of values for each of the parameters:

xs = np.arange(0,10,1)
params1 = np.arange(5,10,1)
params2 = np.arange(1,30,1)
params3 = np.arange(1,20,1)

I would like to compute the output of my_func for each possible combination of params1, params2, and params3. The idea is to be able to compute chi squares and perform bayesian analysis with the output. I know it can be done with nested for loops, but I was wondering if that could be done with mehsgrid. I tried the following, but it breaks:

P1, P2, P3 = np.meshgrid(params1,params2,params3)
results = my_func(xs,P1,P2,P3)

      1 def my_func(xs,param1,param2,param3):
----> 2     values = xs**param1 + xs*param2**2 + param3*xs
      3 
      4 
      5 xs = np.arange(0,10,0.1)

      ValueError: operands could not be broadcast together with shapes (100,) (10,10,40) 

Any idea of how this can be done (if it can be done)?

Edit: The answer by @unutbu works, but no I have one extra question regarding the format of its output. I have changed the parameters ranges so it can be easily explained.

After passing xs as a parameter for np.meshgrid , the shape of results is

np.shape(results)
(5, 10, 29, 19)

Meaning that: axis0 is param1 , axis1 is xs , axis2 is param2 , and axis3 is param3 .

Why is xs put in axis=1 in the output? I would expect the order to follow what passed to np.mesgrid , ie, xs,param1,param2, param3 .

Edit2:

sorry, I have just discovered the "indexing" keyworkd for np.meshgrid. In case anyone needs the indexing as I was planning to use it, use np.meshgrig(arguments,indexing='ij') .

Answer 1

Pass xs also as an argument to np.meshgrid :

import numpy as np

def my_func(xs,param1,param2,param3):
    values = xs**param1 + xs*param2**2 + param3*xs
    return values

xs = np.arange(0,10,0.1)
params1 = np.arange(1,2,0.1)
params2 = np.arange(1,2,0.1)
params3 = np.arange(1,5,0.1)

X, P1, P2, P3 = np.meshgrid(xs, params1, params2, params3, sparse=True, indexing='ij')
my_func(X, P1, P2, P3)