Determinant is computed via Gaussian elimination: the product of the diagonal pivots (accounting for row swaps).
Rank is the number of non-zero pivots after reduction to row-echelon form.
Inverse (when it exists) is obtained by augmenting the matrix with the identity and reducing the left side to I; the right side becomes A⁻¹.
Partial pivoting is used for numerical stability. All arithmetic is floating-point; results are rounded for display.