How to Efficiently Find Row Indices Matching Values in a NumPy Array?

Find the Row Indexes of Various Values in a Numpy Array

1. Understanding the Problem

Given an array X and an array searched_values, the task is to find the indices of rows in X that match the corresponding rows in searched_values.

2. Approach Using NumPy Broadcasting

np.where((X==searched_values[:,None]).all(-1))[1]

3. Memory-Efficient Approach Using np.ravel_multi_index and np.in1d

dims = X.max(0)+1
out = np.where(np.in1d(np.ravel_multi_index(X.T,dims),\
                    np.ravel_multi_index(searched_values.T,dims)))[0]

4. Understanding np.ravel_multi_index

np.ravel_multi_index converts a 2D array of n-dimensional indices to linear index equivalents. For example, given X and dims, it would compute:

np.ravel_multi_index(X.T,dims)

Resulting in [30, 66, 61, 24, 41], where each number represents the linear index equivalent of the corresponding row in X.

5. Choosing Dimensions for Unique Linear Indices

When selecting dimensions for np.ravel_multi_index to generate unique linear indices, consider the following:

• Each column in the input array (X) represents one axis of the grid being mapped to.
• To ensure unique linear indices, the maximum stretch of each axis should be considered, which is the maximum value in each column plus 1.

For the given X:

dims = X.max(0)+1 # [10, 7]

This would create a grid with at least the specified dimensions, ensuring unique linear indices.

The above is the detailed content of How to Efficiently Find Row Indices Matching Values in a NumPy Array?. For more information, please follow other related articles on the PHP Chinese website!