How this calculation works
Gauss–Jordan elimination swaps rows, scales each pivot to one and clears the other entries in its column. The remaining pivots identify the rank. Nullity counts the free variables in the homogeneous system A x = 0.
rank = pivot count; nullity = number of columns − rankA worked example
For [1 2 3; 2 4 6], the RREF is [1 2 3; 0 0 0]. Rank is 1, nullity is 2 and the only pivot is column 1.
Good to know
For small matrices of exact rational numbers, up to 6 rows and 6 columns. Entries are limited to ±1,000,000; integer fraction denominators to 1,000,000. Decimal input allows up to 6 places. No expressions, scientific notation, complex numbers or uncertain measurements. Large intermediate fractions are bounded to keep the tool responsive.
Reference: University of Nebraska–Lincoln: row echelon forms (opens in a new tab) ↗
Row reduction preserves the represented row relationships
Reduced row echelon form uses three elementary operations: exchange two rows, multiply a row by a nonzero value, and add a multiple of one row to another. The pivot in each leading column is normalized to one, with its other column entries cleared.
The tool keeps supported rational inputs exact rather than guessing a pivot from a tiny floating-point residual. Rank counts retained pivot positions. Nullity describes the number of free variables in the homogeneous system with all entered columns treated as coefficients.
An augmented constant column needs separate interpretation
A rectangular matrix is not automatically a system with a final constant column. If you intend to solve equations, use the linear system solver, which explicitly distinguishes coefficient columns from the constants. Otherwise the last column participates in the rank/nullity calculation just like every other matrix column.
A zero row can indicate a redundant row relationship. It does not by itself prove an inconsistent equation. In an augmented system, the contradictory pattern is zero coefficients paired with a nonzero constant. Keep that interpretation separate from reducing a general matrix.
A few questions, answered
Must the input matrix be square?
No. Rectangular matrices are supported within the stated row and column limits.
Are fractions converted to approximate decimals first?
No. Supported rational entries are retained exactly during elimination.
Does every zero row mean no solution?
No. Inconsistency depends on the coefficient/constant relationship of an augmented system.