Using numpy to build an array of all combinations of two arrays

Question

I\'m trying to run over the parameters space of a 6 parameter function to study it\'s numerical behavior before trying to do anything complex with it so I\'m searching for a

Answer 1

The following numpy implementation should be approx. 2x the speed of the given answer:

def cartesian2(arrays):
    arrays = [np.asarray(a) for a in arrays]
    shape = (len(x) for x in arrays)

    ix = np.indices(shape, dtype=int)
    ix = ix.reshape(len(arrays), -1).T

    for n, arr in enumerate(arrays):
        ix[:, n] = arrays[n][ix[:, n]]

    return ix

Answer 2

It looks like you want a grid to evaluate your function, in which case you can use numpy.ogrid (open) or numpy.mgrid (fleshed out):

import numpy
my_grid = numpy.mgrid[[slice(0,1,0.1)]*6]

Answer 3

Here's yet another way, using pure NumPy, no recursion, no list comprehension, and no explicit for loops. It's about 20% slower than the original answer, and it's based on np.meshgrid.

def cartesian(*arrays):
    mesh = np.meshgrid(*arrays)  # standard numpy meshgrid
    dim = len(mesh)  # number of dimensions
    elements = mesh[0].size  # number of elements, any index will do
    flat = np.concatenate(mesh).ravel()  # flatten the whole meshgrid
    reshape = np.reshape(flat, (dim, elements)).T  # reshape and transpose
    return reshape

For example,

x = np.arange(3)
a = cartesian(x, x, x, x, x)
print(a)

gives

[[0 0 0 0 0]
 [0 0 0 0 1]
 [0 0 0 0 2]
 ..., 
 [2 2 2 2 0]
 [2 2 2 2 1]
 [2 2 2 2 2]]

Answer 4

Pandas merge offers a naive, fast solution to the problem:

# given the lists
x, y, z = [1, 2, 3], [4, 5], [6, 7]

# get dfs with same, constant index 
x = pd.DataFrame({'x': x}, index=np.repeat(0, len(x))
y = pd.DataFrame({'y': y}, index=np.repeat(0, len(y))
z = pd.DataFrame({'z': z}, index=np.repeat(0, len(z))

# get all permutations stored in a new df
df = pd.merge(x, pd.merge(y, z, left_index=True, righ_index=True),
              left_index=True, right_index=True)

Answer 5

In newer version of numpy (>1.8.x), numpy.meshgrid() provides a much faster implementation:

@pv's solution

In [113]:

%timeit cartesian(([1, 2, 3], [4, 5], [6, 7]))
10000 loops, best of 3: 135 µs per loop
In [114]:

cartesian(([1, 2, 3], [4, 5], [6, 7]))

Out[114]:
array([[1, 4, 6],
       [1, 4, 7],
       [1, 5, 6],
       [1, 5, 7],
       [2, 4, 6],
       [2, 4, 7],
       [2, 5, 6],
       [2, 5, 7],
       [3, 4, 6],
       [3, 4, 7],
       [3, 5, 6],
       [3, 5, 7]])

numpy.meshgrid() use to be 2D only, now it is capable of ND. In this case, 3D:

In [115]:

%timeit np.array(np.meshgrid([1, 2, 3], [4, 5], [6, 7])).T.reshape(-1,3)
10000 loops, best of 3: 74.1 µs per loop
In [116]:

np.array(np.meshgrid([1, 2, 3], [4, 5], [6, 7])).T.reshape(-1,3)

Out[116]:
array([[1, 4, 6],
       [1, 5, 6],
       [2, 4, 6],
       [2, 5, 6],
       [3, 4, 6],
       [3, 5, 6],
       [1, 4, 7],
       [1, 5, 7],
       [2, 4, 7],
       [2, 5, 7],
       [3, 4, 7],
       [3, 5, 7]])

Note that the order of the final resultant is slightly different.

Answer 6

Here's a pure-numpy implementation. It's about 5× faster than using itertools.

import numpy as np

def cartesian(arrays, out=None):
    """
    Generate a cartesian product of input arrays.

    Parameters
    ----------
    arrays : list of array-like
        1-D arrays to form the cartesian product of.
    out : ndarray
        Array to place the cartesian product in.

    Returns
    -------
    out : ndarray
        2-D array of shape (M, len(arrays)) containing cartesian products
        formed of input arrays.

    Examples
    --------
    >>> cartesian(([1, 2, 3], [4, 5], [6, 7]))
    array([[1, 4, 6],
           [1, 4, 7],
           [1, 5, 6],
           [1, 5, 7],
           [2, 4, 6],
           [2, 4, 7],
           [2, 5, 6],
           [2, 5, 7],
           [3, 4, 6],
           [3, 4, 7],
           [3, 5, 6],
           [3, 5, 7]])

    """

    arrays = [np.asarray(x) for x in arrays]
    dtype = arrays[0].dtype

    n = np.prod([x.size for x in arrays])
    if out is None:
        out = np.zeros([n, len(arrays)], dtype=dtype)

    m = n / arrays[0].size
    out[:,0] = np.repeat(arrays[0], m)
    if arrays[1:]:
        cartesian(arrays[1:], out=out[0:m, 1:])
        for j in xrange(1, arrays[0].size):
            out[j*m:(j+1)*m, 1:] = out[0:m, 1:]
    return out