TheAlgorithms
optimize split_matrix function
I optimized split_matrix function by removing duplicate code to the extract_submatrix function, and added tests to test_strassen_matrix_multiplication.py file
This increases readability and maintainability of the code, but certainly does not "optimize" the algorithm, for you have added the extra overhead of calling a function, which will actually slow down the algorithm.
This increases readability and maintainability of the code, but certainly does not "optimize" the algorithm, for you have added the extra overhead of calling a function, which will actually slow down the algorithm.
Actually, my code with extract_submatrix function is not less efficient than the original one, since there is exactly same number of conditional jumps. This can be checked using godbolt.org: comparison. If you follow the link, you can see disassembly of my revision of the code on the left and of the original code on the right. In original code, there are 8 FOR_ITER jumps and 8 JUMP_BACKWARD jumps. In new code, there are 2 FOR_ITER jumps and 2 JUMP_BACKWARD jumps in extract_submatrix function, which is called 4 times, so in sum we have same 8 FOR_ITER jumps and 8 JUMP_BACKWARD jumps. That's why I think code with extract_submatrix function is not slower and can be used for better readability.