Systems and Matrices

Precise definition

A linear system $a_{i1}x_1+\cdots+a_{in}x_n=b_i$ is written $A\mathbf x=\mathbf b$, where rows encode equations and columns encode coefficients of variables. The augmented matrix $[A\mid\mathbf b]$ stores both sides for row reduction.

Notation and mathematical language

The vector equation is $x_1a_1+\cdots+x_na_n=\mathbf b$, so consistency is equivalent to $\mathbf b\in\operatorname{Col}A$. Rank criteria say a solution exists iff $\operatorname{rank}A=\operatorname{rank}[A\mid b]$, and it is unique iff that common rank equals the number of variables.

Conceptual picture

Each equation defines a hyperplane; solutions are intersections. Algebraically, one solution plus every null-space vector gives the full consistent solution set $x_p+\ker A$.

Conditions and key results

Variable order must remain fixed. Row operations change equations but preserve their common solutions. A square coefficient matrix is not necessarily invertible; rectangular systems can still have unique, none, or infinitely many solutions depending on rank.

A reliable strategy

1. Choose a variable order and translate coefficients—including zeros—into $A$ and constants into $b$.
2. Row-reduce the augmented matrix and identify contradictory rows, pivots, and free variables.
3. State no solution, one solution, or a parameterized affine family.
4. Substitute in the original equations and connect consistency to column-space membership.

Fully worked example

Interpretation and application

Matrix form scales to networks, balances, fitting, and discretized models. A consistent algebraic system can still be a poor real-world model; residuals and parameter uncertainty address different questions from exact solvability.

Common mistakes

Verification and reasonableness

• Translate the final vector back into every original equation.
• Compare coefficient and augmented ranks.
• For a consistent family, verify $A(x_p+z)=b$ for arbitrary $z\in\ker A$.

Practice

1. What does $b\in\operatorname{Col}A$ mean?
2. When is a consistent solution unique?
3. What does a contradictory row represent?

Further deduction

The homogeneous system $Ax=0$ is always consistent. If $A$ has $n$ columns, its solution dimension is $n-\operatorname{rank}A$. For a nonhomogeneous consistent system, the same number describes the directions of its affine solution set, although that set is not a vector subspace unless $b=0$.

For a square system, the condition number separates residual from forward error. A computed $\hat x$ may have tiny residual $b-A\hat x$ yet be far from the true $x$ when $A$ nearly collapses a direction. Exact classroom checks should be supplemented by conditioning awareness in numerical applications.

A parameterized augmented matrix must be split at values that make a prospective pivot zero. Dividing a row by $k-2$ silently assumes $k\ne2$; the case $k=2$ may produce a contradiction row, a free variable, or a different pivot. Recording the condition beside the operation preserves all systems in the family. The ranks of $A$ and $[A\mid b]$ then provide a compact final classification for each parameter case.

Related topics

• Gaussian Elimination
• Matrices
• Kernel and Range