Reduced Row-Echelon Form

Q: Is the reduced row-echelon form of a matrix really unique?

Yes. Existence follows from the reduction algorithm, and uniqueness is a theorem: the pivot columns and every entry of the reduced form are determined by the original matrix alone, independent of the sequence of row operations used. This is exactly what makes RREF a canonical form and what licenses phrases such as the rank of a matrix.

Q: What is the difference between row-echelon form and reduced row-echelon form?

Row-echelon form requires only the staircase pattern with zeros below each leading entry, and is the output of forward elimination. Reduced row-echelon form additionally requires every leading entry to be $1$ and to be the only non-zero entry in its column. Echelon form is not unique; reduced form is.

Q: Does row reduction change the solution set of the system?

No. Each elementary row operation is reversible and maps the solution set onto itself, so the reduced system has exactly the same solutions as the original. This invariance is the entire justification for the method — note that it applies to row operations only.

Q: How do I tell from the RREF whether a system has no solution, one solution, or infinitely many?

Reduce the augmented matrix. If the final column is a pivot column, there is no solution. Otherwise, if every one of the $n$ variable columns is a pivot column, the solution is unique; if any variable column is not a pivot column, the solution set is infinite with $n - r$ free parameters.

Q: Why do numerical libraries such as NumPy not provide an rref function?

Because the pivot pattern is discontinuous in the matrix entries. An arbitrarily small rounding error can promote a structurally zero entry into a pivot, producing a confidently wrong answer. Libraries therefore expose rank-revealing tools with explicit tolerances — SVD-based rank and pivoted QR — rather than an RREF that implies exactness it cannot deliver.

Q: Can the reduced row-echelon form have more non-zero rows than the original matrix?

No. The number of non-zero rows equals the rank, and row operations preserve rank, so it can only stay the same or appear smaller than the raw row count. Rows become zero when they were linearly dependent on other rows; no new non-zero rows can be created.

Q: Does the RREF depend on which entry I choose as a pivot?

The final result does not; the intermediate arithmetic does. Any valid pivot choice reaches the same reduced form, so pivot strategy can be selected purely for numerical stability or arithmetic convenience.

Overview

Row operations let you rewrite a matrix without changing the solution set of the system it represents. That freedom is useful but open-ended — there are infinitely many matrices row-equivalent to any given one. Reduced row-echelon form (RREF) closes the question by nominating a single distinguished member of that family, reached from any starting point and independent of the route taken.

The practical consequence is that RREF acts as a normal form. Two matrices are row-equivalent if and only if they share the same RREF, so a question about an entire equivalence class collapses into a comparison of two specific matrices. For an engineer this is the difference between an open-ended search and a deterministic computation.

Almost every structural quantity in the first half of linear algebra is a direct reading of the RREF: the number of non-zero rows is the rank, the pivot columns index a basis for the column space, the non-pivot columns index the free variables, and for an augmented matrix a pivot in the final column is the exact signature of an inconsistent system. Learning to read an RREF fluently is therefore a higher-leverage skill than learning to produce one by hand.

Definition

RREF

A matrix is in reduced row-echelon form when it satisfies all four conditions below. If a row has any non-zero entry, the leftmost such entry is a $1$ , called a leading one.

Every row that is entirely zero sits below every row that is not.
The leading one of each non-zero row lies strictly to the right of the leading one of the row above it.
Each leading one is the only non-zero entry in its column.
The leading entry of every non-zero row equals $1$ .

Row-echelon form (without the word reduced) requires only the first two conditions plus a leading one; it is the output of forward elimination alone and is not unique.

Zero Row of a Matrix

ZRM

A row of a matrix in which every entry is zero. In RREF all zero rows are collected at the bottom, so the count of non-zero rows is immediate.

Leading One

The leftmost non-zero entry of a non-zero row of a matrix in reduced row-echelon form. By definition this entry equals $1$ , and it is the only non-zero entry in its column.

Pivot Column

A column of a matrix in reduced row-echelon form that contains a leading one. Columns without a leading one are non-pivot (or free) columns. The pivot columns of an augmented matrix determine which variables are dependent.

Concepts

RREF as a canonical form

Row equivalence partitions the set of $m \times n$ matrices into disjoint classes. RREF selects one representative from each class, which makes it a canonical form in the same sense that lowest terms is a canonical form for a fraction. The two halves of this claim are proved separately: an algorithm establishes existence (every matrix can be driven to RREF), and a careful induction on pivot positions establishes uniqueness (the destination does not depend on the sequence of row operations chosen).

Pivot positions carry the structure

Write $r$ for the number of non-zero rows and $D = \{d_1, d_2, \ldots, d_r\}$ for the indices of the pivot columns, with $d_1 < d_2 < \cdots < d_r$ . The set $D$ is the entire structural fingerprint of the matrix. Rank equals $r$ ; the free variables are indexed by the complement of $D$ ; the dimension of the null space is $n - r$ . Two matrices with the same pivot pattern share every structural property that row operations preserve, even when their numeric entries differ.

The augmented-column test for consistency

When RREF is applied to the augmented matrix $[\, A \mid \vec{b} \,]$ of a system with $n$ unknowns, column $n+1$ holds the constants. A leading one in that column corresponds to the equation $0 = 1$ , which no assignment of the variables can satisfy. This gives a decisive, mechanical consistency test: the system is consistent if and only if column $n+1$ is not a pivot column. No inspection of the individual equations is required.

Reading the solution set directly

For a consistent system, each pivot column corresponds to a dependent variable whose value is written explicitly in terms of the free variables by the row containing its leading one. Because the leading one is alone in its column, no back-substitution is needed — the row is already solved for that variable. This is precisely the labour that the extra reduction work of Gauss-Jordan buys over plain Gaussian elimination.

Why the reduced form matters more than the echelon form

Forward elimination alone reaches an echelon form in roughly $\tfrac{2}{3}n^3$ operations; the back-clearing phase that produces the reduced form costs about half as much again. In exchange, the answer requires no further arithmetic to interpret. In a numerical production setting the plain echelon form (as an LU factorisation) is usually preferred because the factors are reusable; in a symbolic, structural or teaching setting the reduced form is preferred because it is unique and directly readable.

Decision path: what an RREF tells you

Row-reduce the augmented matrixApply Gauss-Jordan elimination to

[\, A \mid \vec{b} \,]

until all four RREF conditions hold.

Is column $n+1$ a pivot column?If yes, the system is inconsistent and the solution set is empty. Stop here.

Count the pivot columns, $r$ This is the rank of

A

. The remaining

n - r

variables are free.

$r = n$ ?If yes there are no free variables and the solution is unique. If no, the solution set is infinite with

n - r

degrees of freedom.

Write the solution in vector formEach pivot row gives one dependent variable in terms of the free variables; collect the coefficients into vectors.

Equations

Shape of a matrix in reduced row-echelon form

EQ-RREF-01

[\begin{matrix} 1 & * & 0 & * & 0 & * \\ 0 & 0 & 1 & * & 0 & * \\ 0 & 0 & 0 & 0 & 1 & * \\ 0 & 0 & 0 & 0 & 0 & 0 \end{matrix}]

A representative RREF pattern with $r = 3$ non-zero rows and pivot columns $D = \{1, 3, 5\}$ . Entries marked $*$ are unconstrained; every pivot column is a column of the identity matrix.

Pivot column index set

EQ-RREF-02

D = {d_{1}, d_{2}, \dots, d_{r}}, 1 \leq d_{1} < d_{2} < \dots < d_{r} \leq n

The ordered indices of the pivot columns. The strictly increasing condition is the staircase requirement: each leading one lies to the right of the one above it.

Rank from reduced row-echelon form

EQ-RREF-03

Rank from reduced row-echelon form

The rank of a matrix equals the number of non-zero rows of its RREF, equivalently the number of pivot columns. Uniqueness of RREF is what makes this a well-defined quantity.

Degrees of freedom in the solution set

EQ-RREF-04

nullity (A) = n - r

Consistency test on the augmented matrix

EQ-RREF-05

S \neq \emptyset \Leftrightarrow (n + 1) \notin D

The system $A\vec{x} = \vec{b}$ has at least one solution exactly when the final column of the reduced augmented matrix contains no leading one.

Row equivalence via canonical form

EQ-RREF-06

A \sim B \Leftrightarrow RREF (A) = RREF (B)

Two matrices of the same size are row-equivalent precisely when their reduced row-echelon forms are identical entry-for-entry. This is the operational value of a canonical form.

Variable Definitions

Symbols used on this page
Symbol	Name	Meaning	Domain / type
$A$	Coefficient matrix	The $m \times n$ matrix being reduced	m x n matrix
$m$	Row count	Number of equations in the system	positive integer
$n$	Column count	Number of unknowns in the system	positive integer
$r$	Rank	Number of non-zero rows of the RREF, equal to the number of pivot columns	0 to min(m,n)
$D$	Pivot column set	Indices of the columns containing leading ones	subset of 1..n
$F$	Free column set	Complement of $D$ ; indices of the independent (free) variables	subset of 1..n
$[A ∣ \vec{b}]$	Augmented matrix	Coefficient matrix with the vector of constants appended as an extra column	m x (n+1) matrix
$S$	Solution set	The set of all vectors satisfying the system	subset of C^n

Worked Numerical Example

Problem statement

A three-equation, four-unknown system arises from a static force balance with a redundant member. Reduce the augmented matrix to RREF, decide consistency, and report the rank and the number of degrees of freedom.

Write the augmented matrix
The system is $x_1 + 2x_2 - x_3 + x_4 = 3$ , $2x_1 + 4x_2 - x_3 + 4x_4 = 9$ , $3x_1 + 6x_2 - 2x_3 + 5x_4 = 12$ . Stack the coefficients and constants:
$[\begin{matrix} 1 & 2 & - 1 & 1 & 3 \\ 2 & 4 & - 1 & 4 & 9 \\ 3 & 6 & - 2 & 5 & 12 \end{matrix}]$
Clear below the first pivot
Entry $(1,1)$ is already $1$ , so column 1 is a pivot column. Apply $R_2 \to R_2 - 2R_1$ and $R_3 \to R_3 - 3R_1$ to zero the rest of column 1.
$[\begin{matrix} 1 & 2 & - 1 & 1 & 3 \\ 0 & 0 & 1 & 2 & 3 \\ 0 & 0 & 1 & 2 & 3 \end{matrix}]$
Locate the next pivot
Column 2 below row 1 is entirely zero, so column 2 is not a pivot column — $x_2$ will be a free variable. Move right: entry $(2,3) = 1$ becomes the second leading one, so column 3 is a pivot column.
Clear the rest of the pivot column
Apply $R_3 \to R_3 - R_2$ to eliminate below, then $R_1 \to R_1 + R_2$ to eliminate above. Clearing above the pivot is the step that distinguishes reduced form from plain echelon form.
$[\begin{matrix} 1 & 2 & 0 & 3 & 6 \\ 0 & 0 & 1 & 2 & 3 \\ 0 & 0 & 0 & 0 & 0 \end{matrix}]$
Verify the four RREF conditions
The zero row is at the bottom; leading ones appear in columns 1 and 3 in strictly increasing order; each leading one equals $1$ and is alone in its column. The matrix is in reduced row-echelon form, and by uniqueness no other sequence of row operations could have produced a different result.
Read off the structure
Non-zero rows: $r = 2$ , so $\operatorname{rank}(A) = 2$ . Pivot columns $D = \{1, 3\}$ ; free columns $F = \{2, 4\}$ . Column 5 (the augmented column) is not a pivot column, so the system is consistent. Degrees of freedom: $n - r = 4 - 2 = 2$ .
Express the solution set
Row 1 reads $x_1 + 2x_2 + 3x_4 = 6$ and row 2 reads $x_3 + 2x_4 = 3$ . Each already isolates its dependent variable, so no back-substitution is needed:
$x_{1} = 6 - 2 x_{2} - 3 x_{4}, x_{3} = 3 - 2 x_{4}, x_{2}, x_{4} free$

Result

The system is consistent with rank $2$ and $2$ degrees of freedom, so the solution set is an infinite two-parameter family. In the physical reading, the structure is statically indeterminate to degree two: two independent internal force states satisfy equilibrium with zero external load.

Applications & Industry Use

Structural engineering

Degree of static indeterminacy

Assembling nodal equilibrium equations for a pin-jointed truss gives a system whose coefficient matrix encodes the geometry. The rank read from the RREF against the number of unknown member forces gives the degree of indeterminacy directly, and the free columns identify which members can carry self-stress with no external load.

Electrical engineering

Independent loop and node equations

Kirchhoff's laws applied naively to a network produce more equations than are independent. Reducing the incidence matrix to RREF reveals exactly how many of them are independent, and the pivot columns pick out a valid set of loop currents to solve for.

Process & chemical engineering

Balancing reaction equations

A chemical reaction balance is a homogeneous linear system in the stoichiometric coefficients. The null space dimension — $n - r$ from the RREF — is the number of independent reactions, which distinguishes a uniquely balanced equation from a family of them.

Computing & data

Detecting redundant features

In a design matrix, exact linear dependence between predictors makes a regression unidentifiable. Row reduction of the matrix locates the non-pivot columns, which are exactly the redundant predictors that can be dropped without loss of information.

Control systems

Controllability and observability rank tests

The classical Kalman rank conditions ask whether a controllability or observability matrix has full rank. Row reduction answers that question and, when the rank is deficient, the pivot pattern identifies the reachable subspace.

Manufacturing & production planning

Blend and mix formulation

Blending problems with several constituents and specification targets reduce to a linear system. A pivot in the augmented column signals an infeasible specification before any optimisation is attempted; free columns signal formulation flexibility.

Design Considerations

Choose the reduced form only when you need to read it

Reduction to RREF costs roughly 50% more arithmetic than forward elimination alone. If the objective is to solve one system numerically, an LU factorisation with back-substitution is cheaper and reusable across right-hand sides. Choose RREF when the deliverable is the structure — rank, null space, independence — rather than a single numeric answer.

Reduce the augmented matrix, not the coefficient matrix alone

Consistency is a property of the pair $(A, \vec{b})$ , not of $A$ . Reducing $A$ by itself discards exactly the information needed to detect an inconsistent system. Carry the constants column through every operation.

Pivot selection changes the arithmetic, not the destination

Uniqueness guarantees the same RREF regardless of pivot order, so a pivoting strategy can be chosen freely on numerical or convenience grounds. In exact arithmetic, prefer a pivot of $1$ to avoid fractions; in floating point, prefer the largest available magnitude.

Track the operations if you need the transforming matrix

The reduction itself is a product of elementary matrices. Augmenting with an identity block and reducing the extended array records that product, which is how a matrix inverse and the extended echelon form are computed. Decide before you start whether you will need it.

Size the problem before committing to exact arithmetic

Symbolic reduction of a large integer matrix suffers coefficient explosion: intermediate entries can grow to enormous numerators and denominators even when the input and output are small. Above a few hundred rows, prefer a floating-point factorisation with a rank tolerance, or a fraction-free algorithm.

Define a rank tolerance in floating point

In exact arithmetic a pivot is either zero or not. In floating point, rounding means a structurally zero entry is merely small, so "is this a pivot?" becomes a threshold decision. Any production rank determination must state its tolerance, and a singular value decomposition is a more numerically defensible instrument than row reduction.

Standards & Codes

Notation, interchange and numerical standards that govern how this material is written down, stored and computed in production systems.

Applicable standards, conventions and reference implementations
Reference	Title	Relevance to this topic
`ISO 80000-2`	Quantities and units — Part 2: Mathematics	Governs the typographic conventions used throughout this page: upright letters for operators such as $\det$ , italic for scalar variables, bold or arrow notation for vectors, and bracket forms for matrices.
`IEEE 754-2019`	IEEE Standard for Floating-Point Arithmetic	Defines the binary32 and binary64 arithmetic in which row reduction is normally executed, and therefore the rounding behaviour that makes an exact-zero pivot test unsafe in production code.
`LAPACK / BLAS reference`	Linear Algebra PACKage reference implementation	The de facto interface standard for dense factorisation. LAPACK exposes LU (`xGETRF`) and rank-revealing QR (`xGEQP3`) rather than an RREF routine, which reflects industrial preference for factorisations over canonical forms.
`ISO/IEC 40314`	Mathematical Markup Language (MathML) Version 3.0	The markup standard used to encode every equation on this page, ensuring the notation is machine-readable, searchable and available to assistive technology.
`W3C WCAG 2.1 AA`	Web Content Accessibility Guidelines	Matrices are presented as semantic MathML with alternative text rather than as images, so their content is exposed to screen readers and remains legible at high zoom.

Material Selection

For a mathematical topic, "material" is the numeric representation: the scalar field, storage format and precision the computation is built from.

Representation and precision selection
Representation	Select when	Trade-off
Exact rational (fraction) arithmetic	Structural questions where the answer is a rank, a pivot pattern or a symbolic null space basis, and the matrix is small to moderate.	Guarantees a correct pivot pattern with no tolerance decisions, but intermediate coefficients can grow explosively and memory use is unbounded.
Arbitrary-precision integers with fraction-free elimination	Integer input matrices where exact results are required but rational blow-up must be contained.	Controls coefficient growth via the Bareiss algorithm at the cost of a more complex implementation and results in echelon rather than fully reduced form.
IEEE 754 binary64 (double precision)	Numerical solution of engineering systems where the input data already carries measurement uncertainty.	Fast and memory-efficient, but rank becomes a tolerance-dependent judgement and near-dependent columns may be misclassified.
IEEE 754 binary32 (single precision)	Large-scale or hardware-accelerated computation where throughput dominates and the conditioning is known to be benign.	Halves memory traffic but leaves roughly seven significant decimal digits, which is rarely enough for a defensible rank decision.
Finite field arithmetic (modulo a prime)	Coding theory, cryptography and rank certification by modular reduction.	Exact and fast with no growth in operand size, but the pivot pattern over a finite field can differ from the pattern over the rationals for unlucky primes.
Sparse row storage (CSR) with symbolic ordering	Very large systems from finite element or network models where most entries are zero.	Row reduction causes fill-in that can destroy sparsity, so RREF is usually the wrong tool; a sparse LU or QR with a fill-reducing permutation is preferred.

Manufacturing Notes

Implementation notes — how the result is actually produced by hand, by algorithm and by library, including cost and numerical behaviour.

Operation count and cost

Forward elimination on an $n \times n$ system costs about $\tfrac{2}{3}n^3$ floating-point operations. Completing the reduction by clearing above every pivot adds roughly a further $\tfrac{1}{3}n^3$ , so full Gauss-Jordan reduction is around $n^3$ operations — about 50% more work than elimination with back-substitution for the same answer.

Producing the form by hand

Work strictly left to right, one pivot column at a time. Fix the pivot row, scale it so the leading entry is $1$ , then clear the entire column both above and below before moving on. Locking the pivot row before clearing avoids the most common hand-computation error, which is re-using a row that has already been modified in the same step.

Verifying a completed reduction

Two independent checks are worth the effort. First, confirm all four defining conditions explicitly rather than by eye. Second, substitute a specific solution vector back into the original equations — not the reduced ones — since an arithmetic slip during reduction produces a matrix that is internally consistent but no longer row-equivalent to the input.

Library behaviour differs by design

Symbolic systems such as SymPy and Maxima expose an explicit rref operation. NumPy deliberately does not, offering numpy.linalg.matrix_rank (via SVD) and scipy.linalg.lu instead, on the grounds that a floating-point RREF invites unjustified confidence in the pivot pattern. Treat any floating-point RREF result as an estimate accompanied by a tolerance.

Recording the transformation

Augmenting $A$ with an identity block and reducing $[\, A \mid I_m \,]$ yields both the reduced form and the non-singular matrix $J$ with $JA = \operatorname{RREF}(A)$ . This extended reduction is the mechanism behind matrix inversion and behind the extended echelon form used to extract all four fundamental subspaces in a single pass.

Failure Modes & Common Mistakes

Failure modes, root causes and prevention
Failure mode / mistake	Impact	Root cause	Prevention & detection
Stopping at echelon form	high	Entries above a leading one are left non-zero because forward elimination alone was performed.	Check condition three explicitly: each leading one must be the only non-zero entry in its column, above as well as below.
Reducing the coefficient matrix without the constants	high	The augmented column is dropped, discarding the information that detects an inconsistent system.	Carry the constants column through every operation and test whether column $n+1$ is a pivot column before interpreting anything else.
Treating a zero row as an inconsistency	medium	Confusing a row of all zeros (a redundant equation, harmless) with a row of the form $[\,0\ \cdots\ 0 \mid 1\,]$ (the contradiction $0 = 1$ ).	Inspect the augmented entry of every zero row of the coefficient part; only a non-zero constant there signals inconsistency.
Assuming the pivots lie on the diagonal	medium	Reasoning from square, full-rank examples where pivot column $k$ happens to sit in row $k$ .	Record the pivot index set $D$ explicitly during reduction rather than inferring positions from row numbers.
Miscounting free variables	medium	Counting zero rows instead of non-pivot columns, which coincide only when $m = n$ .	Compute degrees of freedom as $n - r$ where $n$ is the number of unknowns, never as the number of zero rows.
Exact-zero pivot test in floating point	high	Testing `if entry == 0` on rounded data, so a structurally zero entry of magnitude $10^{-17}$ is chosen as a pivot.	Apply a scaled tolerance relative to the matrix norm, or determine rank from a singular value decomposition instead.
Applying column operations	high	Swapping or combining columns by analogy with row operations.	Column operations change the variables, not the equations, and do not preserve the solution set. Restrict every step to the three elementary row operations.
Coefficient explosion in symbolic reduction	low	Naive rational arithmetic on a large integer matrix, where intermediate numerators grow exponentially.	Use a fraction-free (Bareiss) elimination, or reduce modulo several primes and reconstruct.

FAQs

Is the reduced row-echelon form of a matrix really unique?

Yes. Existence follows from the reduction algorithm, and uniqueness is a theorem: the pivot columns and every entry of the reduced form are determined by the original matrix alone, independent of the sequence of row operations used. This is exactly what makes RREF a canonical form and what licenses phrases such as the rank of a matrix.

What is the difference between row-echelon form and reduced row-echelon form?

Row-echelon form requires only the staircase pattern with zeros below each leading entry, and is the output of forward elimination. Reduced row-echelon form additionally requires every leading entry to be $1$ and to be the only non-zero entry in its column. Echelon form is not unique; reduced form is.

Does row reduction change the solution set of the system?

No. Each elementary row operation is reversible and maps the solution set onto itself, so the reduced system has exactly the same solutions as the original. This invariance is the entire justification for the method — note that it applies to row operations only.

How do I tell from the RREF whether a system has no solution, one solution, or infinitely many?

Reduce the augmented matrix. If the final column is a pivot column, there is no solution. Otherwise, if every one of the $n$ variable columns is a pivot column, the solution is unique; if any variable column is not a pivot column, the solution set is infinite with $n - r$ free parameters.

Why do numerical libraries such as NumPy not provide an rref function?

Because the pivot pattern is discontinuous in the matrix entries. An arbitrarily small rounding error can promote a structurally zero entry into a pivot, producing a confidently wrong answer. Libraries therefore expose rank-revealing tools with explicit tolerances — SVD-based rank and pivoted QR — rather than an RREF that implies exactness it cannot deliver.

Can the reduced row-echelon form have more non-zero rows than the original matrix?

No. The number of non-zero rows equals the rank, and row operations preserve rank, so it can only stay the same or appear smaller than the raw row count. Rows become zero when they were linearly dependent on other rows; no new non-zero rows can be created.

Does the RREF depend on which entry I choose as a pivot?

The final result does not; the intermediate arithmetic does. Any valid pivot choice reaches the same reduced form, so pivot strategy can be selected purely for numerical stability or arithmetic convenience.

References

Beezer, R. A. A First Course in Linear Algebra, Version 0.70. University of Puget Sound, 2006. Section RREF. Licensed under the GNU Free Documentation License v1.2.
ISO 80000-2:2019, Quantities and units — Part 2: Mathematics. International Organization for Standardization.
IEEE 754-2019, IEEE Standard for Floating-Point Arithmetic. Institute of Electrical and Electronics Engineers.
Anderson, E. et al. LAPACK Users' Guide, 3rd edition. Society for Industrial and Applied Mathematics, 1999.
W3C. Mathematical Markup Language (MathML) Version 3.0, 2nd edition (ISO/IEC 40314).

AI Suggested Questions

Show me a 4x6 matrix whose RREF has pivot columns in positions 1, 2 and 5, and explain what that pattern implies about the solution set.
Why does uniqueness of the reduced row-echelon form fail if I allow column operations as well as row operations?
Compare the operation count of Gauss-Jordan reduction against LU factorisation with back-substitution for a 500x500 system.
Given a floating-point matrix, how should I choose a numerical tolerance for deciding whether a pivot candidate is really non-zero?
Walk me through reducing an augmented matrix that turns out to be inconsistent, and show exactly where the contradiction appears.
How does the extended echelon form of a matrix relate to its reduced row-echelon form?

Related Calculators

RREF Calculator

Reduce any matrix to reduced row-echelon form with a full audit trail of the row operations applied.

Linear System Solver

Solve $A\vec{x} = \vec{b}$ and report consistency, rank and the vector form of the solution set.

Matrix Rank & Nullity Calculator

Compute rank, nullity and the pivot column set with a selectable numerical tolerance.

Overview

Definition