使用Numpy和While循环生成所有可能的加法组合

Question

I am attempting to build an N-dimensional array from 2 base arrays that show all the possible combinations possible for the addition of all values such that:

Array A = [N0, N1, N2, ... , Nn]

Array B = [M0, M1, M2, ..., Mn]

Combinations = [[N0+M0, N0 + M1, ..., N0+Mn],

                [N1+M0, N1 + M1, ..., N1 + Mn],
                .
                .
                .,
                ]

I am aware that I can use the cartesian product to find all possible combinations (I do not care about repetition) however I do not need the product of all values, I need the addition. I cannot seem to get the while loop I've coded to be able to append the rows as it generates new data (code below).

import numpy as np
MassFlowOx = np.linspace(0.1,103,150)*10**-3
MassFlowFuel = np.linspace(0.1,75,150)*10**-3
size = len(MassFlowFuel)
size_final=size**2
MassFlowComb = np.zeros(size)
temp = np.zeros(size, dtype=float)
i=0
j=0
MassFlowCombArray = np.zeros(size)

# runs all possible combinations for the addition of values in arrays
while i<size-1:
    temp[i] = MassFlowFuel[i] + MassFlowOx[i]
    while j<size-1:
        MassFlowOx[i] = MassFlowOx[j]
        print MassFlowOx[i]        
        j += 1
    MassFlowComb[i] = temp[i]

    MassFlowCombArray[i] = np.append([MassFlowComb[0]], [MassFlowComb[i]], axis = 0)
    i+=1

print MassFlowComb[i]

Is there a particular fucntion I can use? I have not been succesfull in identifying a method that does the addition of all combinations of values in a while loop as it gives sequence error.

EDIT:

Both answers worked. The for loop makes it inN-dimensional and using np.add.outer makes an N dimensional array. For the purposes of this code the latter is more useful.

Thank you for all the responses!

Answer 1

What you need is already implemented in NumPy and is called np.add.outer . If you want to take the difference, simply replace add by subtract . If you want to take the product, simply use outer . For Nx1 and Mx1 arrays, the result will be of dimensions NxM .

Below is a simple example:

import numpy as np
a = np.array([1, 2, 3])
b = np.array([4, 5, 6])
np.add.outer(a, b)

---

array([[5, 6, 7], # [[1+4, 1+5, 1+6],
   [6, 7, 8],     #  [2+4, 2+5, 2+6],
   [7, 8, 9]])    #  [3+4, 3+5, 3+6]]

Answer 2

Why don't you just use double for loop?

import numpy as np
MassFlowOx = np.linspace(0.1,103,150)*10**-3
MassFlowFuel = np.linspace(0.1,75,150)*10**-3
MassFlowComb = []
for a in MassFlowOx:
    for b in MassFlowFuel:
        MassFlowComb.append(a+b)
MassFlowComb = np.array(MassFlowComb)

Answer 3

For this case is better to make a For-Loop than a While-Loop. Your code could be as simples as:

import numpy as np

MassFlowOx = np.linspace(0.1,103,150)*10**-3
MassFlowFuel = np.linspace(0.1,75,150)*10**-3

MassFlowComb = []

for  Fo in MassFlowOx:
  sub_list = []
  for  Ff in MassFlowFuel:
    sub_list.append(Fo+Ff)
  MassFlowComb.append(sub_list)

print(np.asarray(MassFlowComb))

Notice that I used a list to append the sums, but you can do it with numpy