1

I am using scipy's brute optimizer to compute the scores of a function given different input parameters. In order to analyze the results, I want to dump the resulting parameter vs score sets into a csv.

What I currently have is the return from the brute optimizer, which from the docs is of the form:

Returns:
x0 : ndarray
A 1-D array containing the coordinates of a point at which the objective function had its minimum value. (See Note 1 for which point is returned.)
fval : float
Function value at the point x0.
grid : tuple
Representation of the evaluation grid. It has the same length as x0. (Returned when full_output is True.)
Jout : ndarray
Function values at each point of the evaluation grid, i.e., Jout = func(*grid). (Returned when full_output is True.)

source

I want combine the "grid" value with the "Jout" parameter in the form of:

[Param1,Param2,Param3,Score1]
[Param1,Param2,Param3,Score2]
...

An example output from brute is:

     (array([ 0.,  0.,  0.]), -0.96868449202047968, array([[[[0, 0],
     [0, 0]],

    [[1, 1],
     [1, 1]]],


   [[[0, 0],
     [1, 1]],

    [[0, 0],
     [1, 1]]],


   [[[0, 1],
     [0, 1]],

    [[0, 1],
     [0, 1]]]]), array([[[-0.96868449, -0.96868449],
    [-0.96868449, -0.96868449]],

   [[-0.96868449, -0.96868449],
    [-0.96868449, -0.96868449]]]))

I cannot, for the life of me figure out how to combine the array in index 2 of above (the "grid") with the array in index 3 (the "Jout"). My np array foo is not up to snuff.

Any tips or pointers would be very appreciated and ease my frustrated brain of the last few hours.

ponderinghydrogen
  • 219
  • 5
  • 14

1 Answers1

1

So you want to stack the three (2,2,2) matrices in grid and the (2,2,2) matrix Jout? np.stack could be what you need, i.e.:

In [26]: a = np.array([[[[0, 0],
    ...:      [0, 0]],
    ...: 
    ...:     [[1, 1],
    ...:      [1, 1]]],
    ...: 
    ...: 
    ...:    [[[0, 0],
    ...:      [1, 1]],
    ...: 
    ...:     [[0, 0],
    ...:      [1, 1]]],
    ...: 
    ...: 
    ...:    [[[0, 1],
    ...:      [0, 1]],
    ...: 
    ...:     [[0, 1],
    ...:      [0, 1]]]])

In [27]: b = np.array([[[-0.96868449, -0.96868449],
    ...:     [-0.96868449, -0.96868449]],
    ...: 
    ...:    [[-0.96868449, -0.96868449],
    ...:     [-0.96868449, -0.96868449]]])

In [28]: np.stack([*a, b], -1)
Out[28]: 
array([[[[ 0.        ,  0.        ,  0.        , -0.96868449],
         [ 0.        ,  0.        ,  1.        , -0.96868449]],

        [[ 0.        ,  1.        ,  0.        , -0.96868449],
         [ 0.        ,  1.        ,  1.        , -0.96868449]]],


       [[[ 1.        ,  0.        ,  0.        , -0.96868449],
         [ 1.        ,  0.        ,  1.        , -0.96868449]],

        [[ 1.        ,  1.        ,  0.        , -0.96868449],
         [ 1.        ,  1.        ,  1.        , -0.96868449]]]])

In [29]:
tjollans
  • 717
  • 4
  • 18