Linear System Solver
Solve rectangular linear systems with exact fractions, reduced rows and a particular solution plus a basis when free variables remain.
This tool processes its input contents in your browser instead of sending them to a processing service. Page, asset and fixed usage-measurement requests are described on the Privacy page.
Solve rectangular linear systems with exact fractions, reduced rows and a particular solution plus a basis when free variables remain.
How to use it
- 1Choose equation and unknown counts, then fill the labelled coefficients and total for each equation.
- 2Check the equation preview and keep unknowns in the same column order.
- 3Solve and inspect the unique, inconsistent or infinite result.
- 4Download reduced rows, a particular solution and any basis vectors.
Example
- Input
- Rows: 1, 1, 3 and 2, -1, 0
- Output
- Unique solution: x1 = 1, x2 = 2. RREF rows: [1, 0, 1] and [0, 1, 2].
What happens to your data
The augmented matrix and row-reduction working are processed locally. Exported JSON contains the equations and solution description you chose to save; no account or matrix upload is used by the browser workspace.
Last updated November 2026
Solve a rectangular system of linear equations using exact rational row reduction. Choose the numbers of equations and unknowns, then fill the labelled coefficient boxes and the total after the equals sign. A paste view also accepts augmented-matrix rows, with the right-hand value last. The workspace previews the entered equations, then reports whether the system has one solution, no solution or infinitely many solutions. Reduced rows, ranks and calculation working remain available with the result.
How it works
Keep the unknowns in one consistent order in every row. For x1 + x2 = 3, enter 1, 1, 3. For 2x1 − x2 = 0, the next row is 2, -1, 0. A zero coefficient is required for an absent variable. The final column belongs to the right-hand side, so three entries per row describe two unknowns. Commas or whitespace separate entries. Blank interior rows, uneven column counts and unsupported numbers fail with an explanation.
The solver uses Gauss–Jordan row operations to build reduced row echelon form, also called RREF. It swaps a nonzero pivot into place, divides that row by the pivot and eliminates the pivot column in every other row. Each entry is an exact reduced fraction; there is no floating-point threshold that turns a small nonzero coefficient into zero. Row swaps and eliminations are recorded in the working export.
A row whose coefficients are all zero but whose final value is nonzero is a contradiction, so the result is inconsistent. Otherwise every non-pivot variable is free. When no variables are free, the particular solution is the unique solution. When free variables remain, the result includes one particular solution and one basis vector for each independent free parameter. All solutions have the form particular + t1 × basis[0] + t2 × basis[1] and so on. The parameter values may be chosen freely over the reals; exact rational arithmetic is used to construct the displayed vectors.
The matrix can contain up to eight equations and eight unknowns, including overdetermined and underdetermined systems. Exact entries use integers, finite decimals or integer fractions, with at most 160 characters per number. Numerators and denominators have a 5,000-bit intermediate budget, and calculations have a 30,000-step bound. The input field is at most 16 KiB; complete result and working exports are each at most 32 KiB. A breached limit gives an error rather than a rounded or incomplete answer.
This is an algebraic solver for the equations entered. It does not infer equations from prose, perform least-squares regression, reconcile contradictory measurements or determine whether an application model is appropriate. For an inconsistent measured system, choose a fitting method separately rather than treating this exact contradiction as a software failure.
Worked example: Solve two equations exactly
- Input
- Rows: 1, 1, 3 and 2, -1, 0
- Result
- Unique solution: x1 = 1, x2 = 2. RREF rows: [1, 0, 1] and [0, 1, 2].
Common use cases
- Check simultaneous equations without a floating-point pivot tolerance.
- Find the free-variable structure of an underdetermined system.
- Identify contradictory constraints in an overdetermined system.
- Export exact reduced rows and the full particular-plus-basis description.
Frequently asked questions
Does the linear system have to be square?+
No. Use up to eight equations and eight unknowns. Consistency and the number of pivots determine the result, rather than a requirement for matching equation and unknown counts.
What does an infinite solution result mean?+
At least one variable can vary freely. The particular solution is one valid choice. Adding any multiple of each displayed basis vector produces another solution of the same equations.
What is an augmented matrix?+
It is the coefficient table with the right-hand values added as its final column. The vertical bar in [A | b] is a label separating those roles; type only numbers and separators in the input.
Can it solve a system whose measurements disagree?+
It reports exact inconsistency if the equations contradict one another. It does not automatically replace your question with a least-squares approximation.
Standards and limitations
Related tools
Complex Number Calculator
Calculate with complex components, convert rectangular and polar forms, and review magnitude and principal angle in degrees or radians.
Vector Calculator
Calculate 2D or 3D vector operations with explicit projection direction, angle units and clear zero-vector errors.
Permutation & Combination Calculator
Count selections and arrangements exactly, choose whether repetition is allowed, and export large integer results without rounding.
Business Day Calculator
Count or offset working dates using chosen weekdays and your own excluded dates, then export the exact date schedule and assumptions.