How Gauss-Jordan Elimination Works
Gauss-Jordan elimination solves a system of linear equations by writing the coefficients and constants as an augmented matrix [A | b] and applying row operations until the left side becomes the identity matrix. For a 3×3 system a₁x + b₁y + c₁z = d₁, a₂x + b₂y + c₂z = d₂, a₃x + b₃y + c₃z = d₃, the calculator builds the augmented matrix, row-reduces it to reduced row echelon form (RREF), and reads the solution x, y, z directly from the last column.
Formula and method
Three elementary row operations are allowed, each of which preserves the solution set: (1) swap two rows, (2) multiply a row by a nonzero scalar, and (3) add a multiple of one row to another. Starting from column 1, the calculator selects the row with the largest available entry in that column as the pivot (partial pivoting, which reduces rounding error), scales that row so the pivot equals 1, then uses row operation (3) to zero out every other entry — above and below — in that column. Repeating this for each column leaves an identity matrix on the left and the solution on the right: [I | x, y, z]. This is what distinguishes Gauss-Jordan elimination from plain Gaussian elimination, which only zeroes out entries below the pivot (row echelon form) and then requires a separate back-substitution pass to recover each variable.
Common sources of error
- Zero pivot: if the entry you would pivot on is zero, you must swap with a row below it first — dividing by zero corrupts the whole row.
- Sign errors when eliminating: when subtracting a multiple of the pivot row from another row, double-check the sign of the multiplier — this is the most common hand-calculation mistake.
- Rounding mid-calculation: keep full-precision fractions or decimals through every row operation; rounding early compounds errors by the final step.
Checking your result
Once you have x, y, and z, substitute them back into all three original equations — each one must balance. If a row of the reduced matrix reads all zeros with a nonzero constant (like [0 0 0 | 5]), the system is inconsistent and has no solution. If a row reduces entirely to zeros (including the constant), the equations are dependent and the system has infinitely many solutions along a line or plane, rather than a single point.
Applications
Beyond solving linear systems, Gauss-Jordan elimination is the standard way to compute a matrix inverse (row-reduce [A | I] to get [I | A⁻¹]) and to determine a matrix's rank. It underlies circuit analysis (Kirchhoff's laws), balancing chemical equations, linear regression, and the simplex method in linear programming.