Reading Equations from Augmented Matrices

From Rows to Equations

An augmented matrix is a compact way to record a system of linear equations. Reading one correctly means keeping track of two things at once: which variables the coefficient columns belong to, and which column records the right-hand sides. A zero entry is still part of that bookkeeping. It means the corresponding variable has coefficient zero in that equation; it does not remove the variable or change the positions of the later columns.

Suppose the variables have been chosen and ordered as \(x_1,\dots,x_n\). A system with \(m\) equations can be written row by row as \(a_{i1}x_1+\cdots+a_{in}x_n=b_i\), for \(i=1,\dots,m\). The coefficient \(a_{ij}\) is the number multiplying variable \(x_j\) in equation \(i\), and \(b_i\) is the right-hand-side constant in that equation. The order of variables must be known before the entries can be read as coefficients of particular variables.

The bar is a visual divider, not an extra variable or an equation operation. In each row, read the entries before the bar from left to right as the coefficients of \(x_1,\dots,x_n\). Read the entry after the bar as the right-hand side. If a coefficient is zero, its term may be omitted from the written equation, but its column still has its assigned variable. A row of zeros before the bar also has a meaning: it has coefficient zero for every variable.

Reading Each Row in Order

The check in a reading problem is to re-encode the equations: list their coefficients in the declared variable order, then place each right-hand side in the final column. That check catches a misplaced coefficient even when the written equations look plausible. It does not involve finding values of the variables.

The Row-by-Row Rule

Proof. Let row \(i\) of the augmented matrix be \((a_{i1},\dots,a_{in}\mid b_i)\). By the fixed variable order, its first entry multiplies \(x_1\), its second entry multiplies \(x_2\), and, for every \(j\), its \(j\)th entry multiplies \(x_j\). The entry after the bar is \(b_i\), so the equation represented by this row is exactly \(a_{i1}x_1+\cdots+a_{in}x_n=b_i\). This holds for each \(i=1,\dots,m\), giving the stated system. Conversely, in an equation already written in this form, the coefficient of each specified variable \(x_j\) is fixed as \(a_{ij}\), and the right-hand side is fixed as \(b_i\). Placing those numbers in the prescribed order recovers exactly one row \((a_{i1},\dots,a_{in}\mid b_i)\). Doing this for all equations recovers exactly one augmented matrix. \(\square\)

The uniqueness in this statement depends on the fixed variable order and on writing equations in the stated form. The same equation can be rearranged into another algebraically equivalent form, but reading an augmented matrix is a matter of interpreting the entries as they are displayed—not rearranging or solving the equations.

Rows with Only Zero Coefficients

A row with all zero coefficients must be read together with its final entry: the row represents \(0=b_i\). The reading rule itself does not decide the status of the entire system; it only records the equation encoded by that row. In particular, do not discard a row while translating, or silently replace its final entry with zero.

Variable Order and the Constants Column

The same numerical array can describe different equations if the variables are assigned to its coefficient columns in a different order. The constants column has a different role: it stays to the right of the bar and is not reassigned to a variable when the variable order changes.

Proof. In row \(i\), the variable terms are \(\sum_{j=1}^{n}a_{ij}x_j\). Since \(y_k=x_{\pi(k)}\), the term involving \(x_{\pi(k)}\) becomes \(a_{i,\pi(k)}y_k\). Rewriting the terms in the order \(y_1,\dots,y_n\) gives \(\sum_{k=1}^{n}a_{i,\pi(k)}y_k\). The right-hand side of the equation is still \(b_i\), because changing the names and order of the variables does not change that side. Thus the coefficient columns are reordered according to \(\pi\), while the constants column is unchanged. \(\square\)

For instance, in a row with variable order \(x,y,z\), the coefficients \((1,0,-2)\) mean \(x-2z\). If the declared order is instead \(z,y,x\), the same equation has coefficient list \((-2,0,1)\). The right-hand-side constant remains the same. The matrix entries must always be interpreted with the variable order that accompanies the matrix.

Common Mistakes

Mistake: treating the constants column as another variable coefficient. From the row \((2,-1\mid5)\), a wrong reading is \(2x-y+5=0\), as if \(5\) were a coefficient. The correction is to respect the bar: \(2\) and \(-1\) multiply the ordered variables \(x\) and \(y\), while \(5\) is the right-hand side. The equation is \(2x-y=5\).

Mistake: skipping a zero coefficient and shifting later entries. In the order \(x,y,z\), the row \((1,0,-2\mid4)\) does not mean \(x-2y=4\). The second column still belongs to \(y\), so the zero says there is no \(y\)-term; the third column belongs to \(z\). The correct equation is \(x-2z=4\).

Mistake: dropping a zero row during transcription. A row \((0,0,0\mid0)\) translates to \(0=0\). It is part of the system as recorded, even if it contributes no variable terms. Reading every row, including this one, keeps the equation count and the matrix-to-system translation accurate.