Solvability: Existence & Uniqueness

The Two Questions

Given the system

A\mathbf{x} = \mathbf{b}

with

A

of size

m \times n

, two logically independent questions arise.

Existence: is there at least one vector

\mathbf{x}

satisfying

A\mathbf{x} = \mathbf{b}

? This asks whether

\mathbf{b}

lies in the column space of

A

.

Uniqueness: if a solution exists, is it the only one? This asks whether the null space of

A

is trivial.

The two questions are independent — existence can hold without uniqueness, and non-existence makes uniqueness moot. Both are answered by the rank of

A

.

Solvability Notation

Notation

Solvability Notation

Two words that classify a system before any solving happens, a rank comparison that answers the existence question without producing a single solution, and a count of solutions that is never anything but zero, one, or infinity. All catalogued among the linear algebra symbols.

A\mathbf{x} = \mathbf{b}

, the augmented matrix

[A \mid \mathbf{b}]

and its bar come from Gaussian elimination notation;

\operatorname{rank}

from rank notation; free variables and pivot positions from echelon form notation;

\operatorname{Col}

and

\operatorname{Nul}

from fundamental-space notation.

consistent / inconsistent

the system has at least one solution — or none at all

A two-word classification applied to the system, not to any solution: consistent means at least one exists, inconsistent means none does. The words settle existence only and say nothing about how many, which is why The Two Questions above keeps existence and uniqueness apart.

CasesThe marks that detect inconsistency are visual: a row reading

[\,0 \; \cdots \; 0 \mid d\,]

with

d \neq 0

asserts

0 = d

, so a single such row in the augmented matrix decides the whole system.

Do not confuse"Inconsistent" as a comment on the equations being contradictory in some loose sense. It is a precise verdict about the solution set being empty — and a system can look perfectly reasonable and still be inconsistent, which is what the rank test below is for.

\operatorname{rank}(A) = \operatorname{rank}([A \mid \mathbf{b}])

the ranks agree, so the system is consistent

Existence written as a comparison rather than a search: appending

\mathbf{b}

adds no new independent direction exactly when

\mathbf{b}

already lies in the column space. The test answers whether a solution exists while producing none — the point of The Existence Condition below.

CasesAdding the second rank gives the whole classification, the Rouché–Capelli statement below: ranks unequal means no solution, ranks equal to

n

means exactly one, ranks equal but less than

n

means infinitely many — with

n - \operatorname{rank}(A)

counting the free parameters.

Do not confuseReading

\operatorname{rank}([A \mid \mathbf{b}])

as a rank of two things. The bar is punctuation, not an operation — the argument is one matrix that happens to have been assembled from a coefficient block and a column.

0

,

1

, or

\infty

a linear system has no solutions, exactly one, or infinitely many

The count of solutions is not an arbitrary number: The Three Cases Combined below shows only these three occur. A system with two distinct solutions has infinitely many, because the whole line through them solves it as well — linearity forbids any finite count above one.

CasesThe infinite case is written as a parametrised family rather than the bare

\infty

— a particular solution plus the null space, which is why "infinitely many" still admits a precise description and a dimension.

Do not confuseExpecting "more equations than unknowns" to force zero and "fewer" to force infinitely many. Neither holds: an overdetermined system can be consistent and an underdetermined one inconsistent — the shape suggests the outcome, the ranks decide it.

All linear algebra symbols →↑ Linear Systems

The Existence Condition

The system

A\mathbf{x} = \mathbf{b}

has at least one solution if and only if

\text{rank}(A) = \text{rank}([A \mid \mathbf{b}])

When this condition holds,

\mathbf{b}

is already a linear combination of the columns of

A

, so appending

\mathbf{b}

as an extra column does not introduce a new independent direction — the rank stays the same.

When the condition fails —

\text{rank}([A \mid \mathbf{b}]) > \text{rank}(A)

— the vector

\mathbf{b}

is not in the column space. In the echelon form of the augmented matrix, this appears as a row

[0 \; 0 \; \cdots \; 0 \mid d]

with

d \neq 0

: the equation

0 = d

has no solution, and the system is inconsistent.

Since appending one column can increase the rank by at most

1

, the only possibilities are

\text{rank}([A \mid \mathbf{b}]) = \text{rank}(A)

(consistent) or

\text{rank}([A \mid \mathbf{b}]) = \text{rank}(A) + 1

(inconsistent).

The Uniqueness Condition

When a solution exists, it is unique if and only if

\text{rank}(A) = n

where

n

is the number of unknowns. Full column rank means every column of

A

contains a pivot, leaving no free variables. The null space is

\{\mathbf{0}\}

, so the particular solution

\mathbf{x}_p

stands alone with nothing to add.

When

\text{rank}(A) < n

, there are

n - \text{rank}(A)

free variables. Each free variable parametrizes a direction along which the solution can move without violating the equations. The solution set is infinite — a translated copy of the null space, which has dimension

n - \text{rank}(A)

.

The Three Cases Combined

Putting existence and uniqueness together, every linear system falls into exactly one of three cases.

\text{rank}(A) < \text{rank}([A \mid \mathbf{b}])

: no solution. The system is inconsistent.

\text{rank}(A) = \text{rank}([A \mid \mathbf{b}]) = n

: exactly one solution. The system is consistent and fully determined.

\text{rank}(A) = \text{rank}([A \mid \mathbf{b}]) < n

: infinitely many solutions. The system is consistent but underdetermined, with

n - \text{rank}(A)

free parameters.

There is no case with a finite number of solutions greater than one. If two distinct solutions exist, their difference lies in the null space, which is a subspace — and a nontrivial subspace contains infinitely many vectors, generating infinitely many solutions.

The three endings, side by side

Existence fails only in the middle case, where a contradiction row appears. Uniqueness fails only in the last, where a free column leaves an unknown unpinned. The two questions are independent of each other, and between them they exhaust what can happen. Build a system for each ending on the linear system solutions visualizer.

The rank conditions in the next section are these three pictures restated as arithmetic on two numbers.

The Rouché–Capelli Theorem

The existence condition

\text{rank}(A) = \text{rank}([A \mid \mathbf{b}])

is known as the Rouché–Capelli theorem (or the Kronecker–Capelli theorem in some traditions):

Solvability Rank Criterion

A\mathbf{x} = \mathbf{b} \text{ consistent} \iff \text{rank}(A) = \text{rank}([A \mid \mathbf{b}])

Solvability rank criterion →

When consistent, the solution set has dimension

n - \text{rank}(A)

.

The theorem unifies the existence and dimension questions into a single rank comparison. It applies to every linear system regardless of the shape of

A

— square, tall, or wide. The proof follows directly from the column-space interpretation:

\mathbf{b} \in \text{Col}(A)

if and only if adding

\mathbf{b}

as a column does not increase the rank.

Square Systems

When

A

is

n \times n

, the determinant provides the sharpest diagnostic.

If

\det(A) \neq 0

: the matrix is invertible, the rank is

n

, and the system

A\mathbf{x} = \mathbf{b}

has exactly one solution for every

\mathbf{b}

. The solution is

\mathbf{x} = A^{-1}\mathbf{b}

, or equivalently, the solution obtained by Gaussian elimination. Existence and uniqueness both hold universally — the right-hand side does not matter.

If

\det(A) = 0

: the matrix is singular, the rank is less than

n

, and the outcome depends on

\mathbf{b}

. For some

\mathbf{b}

(those in the column space), infinitely many solutions exist. For other

\mathbf{b}

(those outside the column space), no solution exists. The determinant test determines whether the coefficient matrix is adequate; the rank comparison with the augmented matrix determines whether the specific

\mathbf{b}

is reachable.

Overdetermined Systems

When

m > n

— more equations than unknowns — the system is overdetermined. The coefficient matrix is tall, and generically, no solution exists.

The column space of

A

is at most

n

-dimensional inside

\mathbb{R}^m

. When

m > n

, this column space is a proper subspace — most vectors in

\mathbb{R}^m

lie outside it. A randomly chosen

\mathbf{b}

will almost certainly not be in the column space, making the system inconsistent.

A solution exists only when

\mathbf{b}

happens to lie in the column space — when the extra equations are consistent with the first

n

. When a solution does exist and

\text{rank}(A) = n

, it is unique.

When no exact solution exists, the least-squares approach finds the

\hat{\mathbf{x}}

that minimizes

\|A\mathbf{x} - \mathbf{b}\|^2

— the closest approximation. The least-squares solution satisfies the normal equations

A^T A \hat{\mathbf{x}} = A^T \mathbf{b}

and is the projection of

\mathbf{b}

onto the column space.

Underdetermined Systems

When

m < n

— fewer equations than unknowns — the system is underdetermined. If a solution exists, it is never unique: the rank cannot exceed

m

, which is less than

n

, so at least

n - m

free variables remain.

The solution set, when nonempty, is an affine subspace of dimension at least

n - m

. Multiple vectors

\mathbf{x}

satisfy the equations, and additional criteria beyond the linear system itself are needed to select a preferred one.

Common selection criteria include minimum norm (the

\mathbf{x}

closest to

\mathbf{0}

, given by the pseudoinverse

\mathbf{x} = A^T(AA^T)^{-1}\mathbf{b}

), sparsity (the

\mathbf{x}

with the fewest nonzero entries, central to compressed sensing), and physical constraints (bounds or nonnegativity in engineering applications).

Underdetermined systems are not defective — they arise naturally whenever a problem has more degrees of freedom than constraints. The linear system identifies the feasible set; the selection criterion picks a point within it.

Download

PNGRaster · 2×PDFPrint quality

Linear systems · by shape

What the shape of A predicts

Before any reduction, the shape of $A$ narrows what can happen. It never settles the question — $\mathbf{b}$ still decides existence — but it says which outcomes are available and which are ruled out.

4shapes

Type	Condition	existence	uniqueness	practical approach
Square
Nonsingular§ 6	$m = n$ , $\det A \neq 0$	always, for every $\mathbf{b}$	always unique	factorize once, solve per $\mathbf{b}$
Singular§ 6	$m = n$ , $\det A = 0$	iff $\mathbf{b} \in \operatorname{Col}(A)$	never, when it exists	reduce $[A \mid \mathbf{b}]$ , compare ranks
More equations than unknowns
Overdetermined§ 7	$m > n$ — tall	unusual — most $\mathbf{b}$ miss $\operatorname{Col}(A)$	unique when it exists and $\operatorname{rank}(A) = n$	least squares when inconsistent
More unknowns than equations
Underdetermined§ 8	$m < n$ — wide	iff $\operatorname{rank}(A) = \operatorname{rank}([A \mid \mathbf{b}])$	never — at least $n - m$ free parameters	impose a criterion: minimum norm, sparsity

Shape constrains, rank decides. A wide matrix cannot have a unique solution whatever

\mathbf{b}

is; a tall one usually has none. What no shape determines is existence, which depends on whether

\mathbf{b}

lies in the column space — and that is a fact about

\mathbf{b}

, not about the dimensions.

What the shape of A predicts·/linear-algebra/linear-systems/solvabilityLearn Math Class

Read down the uniqueness column and the shape does more work than down the existence column. A wide matrix can never have a unique solution — rank is capped by

m

, which is below

n

, so free variables are guaranteed before

\mathbf{b}

is even mentioned. Existence is different: every row except the first says "depends on

\mathbf{b}

", because whether

\mathbf{b}

lies in the column space is not a fact about the dimensions of

A

at all.

Geometric Interpretation

Each equation

a_{i1}x_1 + a_{i2}x_2 + \cdots + a_{in}x_n = b_i

defines a hyperplane in

\mathbb{R}^n

— a flat set of dimension

n - 1

. The solution set of the full system is the intersection of all

m

hyperplanes.

When the system is inconsistent, the hyperplanes have no common point. In

\mathbb{R}^2

this means parallel lines. In

\mathbb{R}^3

it can mean parallel planes, or planes arranged in a triangular prism where each pair intersects but no point lies on all three.

When the system has a unique solution, the hyperplanes meet at a single point. This requires at least

n

independent equations — enough to cut the solution space down from

n

dimensions to

0

.

When the system has infinitely many solutions, the hyperplanes meet along a flat of dimension

n - r

, where

r = \text{rank}(A)

. Each independent equation reduces the dimension of the intersection by one, and the rank counts how many equations cut independently. The remaining

n - r

dimensions are the free directions in the solution set.

Structure of the Solution Set

When

A\mathbf{x} = \mathbf{b}

is consistent, the complete solution set has the form

Solution Structure Decomposition

\mathbf{x} = \mathbf{x}_p + \mathbf{x}_h, \quad \mathbf{x}_h \in \text{Null}(A)

Solution structure decomposition →

where

\mathbf{x}_p

is any one particular solution and

\mathbf{x}_h

ranges over the null space. Equivalently, the solution set is

\{\mathbf{x}_p + \mathbf{x}_h : \mathbf{x}_h \in \text{Null}(A)\}

— a coset of the null space, the null space translated by

\mathbf{x}_p

. In geometric terms, it is an affine subspace: a subspace shifted away from the origin.

The dimension of the solution set equals the dimension of the null space:

n - \text{rank}(A)

. When this is

0

, the solution set is a single point

\{\mathbf{x}_p\}

. When it is

1

, the solution set is a line. When it is

2

, a plane. The shape is always a flat, and the null space determines its orientation.

The particular solution

\mathbf{x}_p

captures the effect of the right-hand side

\mathbf{b}

. The homogeneous component

\mathbf{x}_h

captures the inherent freedom in the system — the directions along which the solution can shift without violating any equation. This decomposition into "forced" and "free" parts is one of the most fundamental structural results in linear algebra.

Component	Symbol	Where it comes from	What it captures
Particular solution	x_p	any one vector satisfying A x_p = b	the "forced" part; the response to the right-hand side b
Homogeneous part	x_h	any vector in Null(A), i.e. A x_h = 0	the "free" directions along which the solution can shift
Complete solution	x = x_p + x_h	sum over all x_h ∈ Null(A)	affine flat of dimension n − rank(A) through x_p

Summary: Solvability Questions Answered

The two foundational questions about A·x = b — existence and uniqueness — together with the structure of the solution set and the special behavior of square, overdetermined, and underdetermined shapes, can be collected as a six-row Q&A reference. The table below pairs each question a reader might bring to a linear system with the answer in terms of rank, determinant, or solution-set structure.

The first group below is the theory and the second is what to do about it. Existence and uniqueness are each a single rank comparison, and they are independent — a system can have solutions without having one solution, or have at most one without having any. The second group takes the cases where the answer is unwelcome and says what replaces the original question.

Download

PNGRaster · 2×PDFPrint quality

Linear systems · solvability

Every solvability question, answered by rank

Two comparisons settle existence and uniqueness completely. Everything below them is what to do once the answer is known — and the shape of the system determines which of those situations you are in.

6questions

The two questions3

1

Does a solution exist?§ 2

equivalently

\mathbf{b} \in \operatorname{Col}(A)

\operatorname{rank}(A) = \operatorname{rank}([A \mid \mathbf{b}])

Rouché–Capelli. If the augmented column raises the rank, it contributes a direction the columns of

A

cannot reach, and the reduction reports this as a row reading

0 = d

. Existence is a question about

\mathbf{b}

, not about

A

.

2

Is the solution unique?§ 3

equivalently

\operatorname{Null}(A) = \{\mathbf{0}\}

\operatorname{rank}(A) = n

A pivot in every column means no free variables. Note this says nothing about whether a solution exists — uniqueness is a question about

A

alone, which is why the two conditions are independent and both must be checked.

3

What does the solution set look like?§ 10

when a solution exists

\mathbf{x}_p + \operatorname{Null}(A)

One particular solution plus the entire null space. The set is a point when the nullity is zero, a line when it is one, a plane when it is two — never anything else, because it is always a translated subspace.

What the shape of the system implies3

4

Square and nonsingular§ 6

m = n

and

\det A \neq 0

\mathbf{x} = A^{-1}\mathbf{b}

Both conditions hold at once and for every

\mathbf{b}

— the only case where the answer is unconditional. In practice solve by factorization rather than by forming the inverse; the formula is a statement about existence, not a method.

5

Overdetermined, inconsistent§ 7

m > n

, no exact solution

A^{\mathsf{T}}A\hat{\mathbf{x}} = A^{\mathsf{T}}\mathbf{b}

More equations than unknowns usually means no solution at all, so the question changes: least squares minimises

\|A\mathbf{x} - \mathbf{b}\|^2

instead of solving. The answer is the projection of

\mathbf{b}

onto the column space — the closest reachable point.

6

Underdetermined§ 8

m < n

, consistent

infinitely many; choose one

Fewer equations than unknowns leaves free variables, so a solution is never unique and the algebra cannot pick between them. The criterion comes from outside: minimum norm via the pseudoinverse, sparsity, or a physical constraint the model imposes.

The whole of solvability is two numbers compared three ways:

\operatorname{rank}(A)

against

\operatorname{rank}([A \mid \mathbf{b}])

decides existence, and

\operatorname{rank}(A)

against

n

decides uniqueness. Both come from one reduction, which is why the two questions are never asked separately in practice.

Every solvability question, answered by rank·/linear-algebra/linear-systems/solvabilityLearn Math Class

Both comparisons come out of the same reduction, which is why they are never computed separately. Reduce

[A \mid \mathbf{b}]

once: a contradiction row settles existence, and the count of pivot columns settles uniqueness. Nothing else is needed, and no determinant is involved — which matters because the determinant test only applies to square systems, while these two apply to every system.

The last two rows are worth reading as a change of question rather than a failure. An overdetermined system usually has no solution, so least squares asks for the closest approximation instead and answers with a projection. An underdetermined system has too many, so a selection criterion is imposed from outside the algebra — minimum norm, sparsity, or whatever the model requires. In both cases the linear algebra is finished; what remains is deciding what you wanted.

Solvability FAQ

Does having more equations than unknowns mean there is no solution?

+

No, and the converse fails too. An overdetermined system can be perfectly consistent, and an underdetermined one can be inconsistent. The shape of a system suggests an outcome without deciding it; comparing the rank of the coefficient matrix with the rank of the augmented matrix is what actually settles the question.Read more →

Is $\operatorname{rank}([A \mid \mathbf{b}])$ the rank of two things?

+

No, it is the rank of one matrix that happens to have been assembled from a coefficient block and an extra column. The bar is punctuation rather than an operation, so nothing is being combined or compared inside the brackets. The comparison happens outside, between this number and

\operatorname{rank}(A)

.Read more →

What does "inconsistent" mean exactly?

+

That the solution set is empty, which is a precise verdict rather than a loose remark about the equations disagreeing. A system can look entirely reasonable and still be inconsistent. The visual tell during elimination is a row reading all zeros to the left of the bar with a nonzero entry to its right, asserting

0 = d

.Read more →

Solvability of Linear Systems

When Solutions Exist and When They Are Unique

The Two Questions

Solvability Notation

Solvability Notation

The Existence Condition

The Uniqueness Condition

The Three Cases Combined

The Rouché–Capelli Theorem

Square Systems

Overdetermined Systems

Underdetermined Systems

What the shape of A predicts

Geometric Interpretation

Structure of the Solution Set

Summary: Solvability Questions Answered

Every solvability question, answered by rank

Solvability FAQ

Does having more equations than unknowns mean there is no solution?

Is $\operatorname{rank}([A \mid \mathbf{b}])$ the rank of two things?

What does "inconsistent" mean exactly?

Solvability of Linear Systems

When Solutions Exist and When They Are Unique

The Two Questions

Solvability Notation

Solvability Notation

The Existence Condition

The Uniqueness Condition

The Three Cases Combined

The Rouché–Capelli Theorem

Square Systems

Overdetermined Systems

Underdetermined Systems

What the shape of A predicts

Geometric Interpretation

Structure of the Solution Set

Summary: Solvability Questions Answered

Every solvability question, answered by rank

Solvability FAQ

Does having more equations than unknowns mean there is no solution?

Is rank⁡([A∣b])\operatorname{rank}([A \mid \mathbf{b}])rank([A∣b]) the rank of two things?

What does "inconsistent" mean exactly?

Is $\operatorname{rank}([A \mid \mathbf{b}])$ the rank of two things?