TheAlgorithms/Python · error · ValueError

The input array is not a square matrix

Error message

The input array is not a square matrix

What it means

maxpooling() in computer_vision/pooling_functions.py raises this ValueError when the input array's row count differs from its column count. The pooling loop assumes a square matrix: it computes one edge length for the output and iterates rows/columns symmetrically, so rectangular input would corrupt indices rather than pool correctly.

Source

Thrown at computer_vision/pooling_functions.py:27

    """
    This function is used to perform maxpooling on the input array of 2D matrix(image)
    Args:
        arr: numpy array
        size: size of pooling matrix
        stride: the number of pixels shifts over the input matrix
    Returns:
        numpy array of maxpooled matrix
    Sample Input Output:
    >>> maxpooling([[1,2,3,4],[5,6,7,8],[9,10,11,12],[13,14,15,16]], 2, 2)
    array([[ 6.,  8.],
           [14., 16.]])
    >>> maxpooling([[147, 180, 122],[241, 76, 32],[126, 13, 157]], 2, 1)
    array([[241., 180.],
           [241., 157.]])
    """
    arr = np.array(arr)
    if arr.shape[0] != arr.shape[1]:
        raise ValueError("The input array is not a square matrix")
    i = 0
    j = 0
    mat_i = 0
    mat_j = 0

    # compute the shape of the output matrix
    maxpool_shape = (arr.shape[0] - size) // stride + 1
    # initialize the output matrix with zeros of shape maxpool_shape
    updated_arr = np.zeros((maxpool_shape, maxpool_shape))

    while i < arr.shape[0]:
        if i + size > arr.shape[0]:
            # if the end of the matrix is reached, break
            break
        while j < arr.shape[1]:
            # if the end of the matrix is reached, break
            if j + size > arr.shape[1]:
                break

View on GitHub (pinned to f5988cc097)

Solutions

1. Crop or pad the input to square before calling: arr[:n, :n] with n = min(arr.shape)
2. Or use a library pooling op that supports rectangles: torch.nn.functional.max_pool2d, cv2, or numpy strides
3. If you own the code, generalize the output shape to ((h-size)//stride+1, (w-size)//stride+1)

Example fix

# before
maxpooling([[1,2,3],[4,5,6],[7,8,9],[10,11,12]], 2, 2)
# ValueError: The input array is not a square matrix

# after
import numpy as np
arr = np.array([[1,2,3],[4,5,6],[7,8,9],[10,11,12]])
n = min(arr.shape)
maxpooling(arr[:n, :n], 2, 2)

import numpy as np
arr = np.asarray(arr)
if arr.ndim != 2 or arr.shape[0] != arr.shape[1]:
    n = min(arr.shape)
    arr = arr[:n, :n]  # or pad
maxpooling(arr.tolist(), size, stride)

def is_square(arr) -> bool:
    a = np.asarray(arr)
    return a.ndim == 2 and a.shape[0] == a.shape[1]

try:
    maxpooling(arr, size, stride)
except ValueError as e:
    if 'square matrix' in str(e):
        n = min(np.asarray(arr).shape)
        return maxpooling(np.asarray(arr)[:n, :n].tolist(), size, stride)
    raise

When it happens

Trigger: maxpooling([[1,2,3],[4,5,6]], 2, 2) — a 2x3 list; passing a grayscale-cropped or resized image array whose width != height; passing a numpy array of shape (h, w) with h != w.

Common situations: Feeding non-square image crops/patches into a hand-rolled pooling step; assuming the function pads or crops rectangular input like torch.nn.MaxPool2d does; converting RGBA or channel-last arrays without squeezing to a square 2-D grid.

Related errors

• invalid k value
• Vector length must match the number of nodes in the graph.
• Coefficient matrix dimensions must be nxn but received {rows
• Constant matrix must be nx1 but received {rows2}x{cols2}
• Coefficient and constant matrices dimensions must be nxn and

AI-assisted analysis of TheAlgorithms/Python@f5988cc097 (2026-08-14). Data as JSON: /api/errors/a0fc44602874d94c. Report an issue: GitHub.