This is a naive implementation of the Gauss/Jordan algorithm to invert matrices. It is reasonably fast and can invert a 2000×2000 matrix in 8.2 seconds. However, there are faster programs out there. This program was designed to have a very low memory footprint.
The determinant is calculated with no additional cost during the matrix inversion.
make
Run the program using
./matrix [order of matrix]
Then enter the matrix to be inverted. For example, you could do the following:
./matrix 3 < example.txt
The matrix in example.txt will be inverted and its determinant will be shown. We also supply a singular matrix in example2.txt. If the matrix is singular, the final Gauss/Jordan tableau is returned. From this final tableau, it follows trivially that the matrix does not have full rank and as such is not invertible.
Matrices are always entered "by row", to borrow a term from R. For example, the matrix in the file example.txt has the first row 4, 2, 8.
Due to the diceyness of the == operator as it relates to floating-point numbers, if the absolute value of a diagonal element is less than 10⁻¹⁰ and there is no suitable row below it to alleviate the problem using elementary row operations, the matrix is considered to be singular. You are invited to change this threshold using line 30 of matrix.c.