Determinants and Cramer's Rule Advanced. This lesson goes beyond core Algebra II. You can skip it.

Solving the general two-variable system

Watch the pattern appear on real numbers before meeting the letters. Solve

To clear $y$, multiply the first equation by $-1$ (the $y$-coefficient of the second equation) and the second equation by $1$ (the $y$-coefficient of the first). That is the same move you are about to see done with letters:

Subtract the second line from the first. The $y$ terms cancel, and $x$ is left with the coefficient $-5$:

Where did that $-5$ come from? From the four coefficients $4, 1, 1, -1$, combined as $4 \times (-1) - 1 \times 1$. Hold on to that combination. It is about to reappear, unchanged, when the same two steps are repeated with letters instead of numbers.

Take the most general system of two linear equations in two unknowns. Write the coefficients as letters so the result works for every system at once:

Solve it by elimination, with one thing to watch for. The multipliers about to appear, $d$ and $b$, are letters standing for coefficients that could themselves be $0$ for some particular system, unlike the nonzero multiplier the row operations of the last lesson required. That is not a problem here, because at this stage all that is needed is a one-way fact: multiply both sides of a true equation by any number, zero included, and the result is still a true equation. So whatever the actual solution turns out to be, it must satisfy the combination about to be built, whether or not $d$ or $b$ happens to be zero. Whether the reverse holds, whether the values this process produces actually solve the system, is checked separately below, and only for the case where the process can be completed, $D \ne 0$. To clear $y$, multiply the first equation by $d$ and the second by $b$, so the $y$ terms match:

Subtract the second line from the first. The $bdy$ terms cancel, and $x$ is left with the coefficient $ad - bc$:

Check it against the numbers you already found: with $a = 4$, $b = 1$, $c = 1$, $d = -1$, this coefficient is exactly the $-5$ you computed by hand above.

Now do the mirror image to clear $x$. Multiply the first equation by $c$ and the second by $a$, letting the same one-way reasoning cover $c$ and $a$ whether or not either happens to be zero, then subtract the first from the second, and the same coefficient $ad - bc$ reappears in front of $y$:

Look at what these two results share. The quantity $ad - bc$ sits in front of both $x$ and $y$, which means it becomes the denominator the moment you divide to solve:

One combination of the four coefficients controls both answers. If it is not zero, you can divide and the system has a clean solution. If it is zero, the division is illegal, and that tells you the two equations no longer pin down one single point. A number that important should not stay nameless.

The determinant and its bar notation

The shared quantity $ad - bc$ is the determinant of the four coefficients, arranged in the square grid in which they appear:

The straight bars are the determinant symbol. The rule they stand for is short: multiply down the main diagonal ($a$ times $d$), then subtract the product up the other diagonal ($b$ times $c$). We write the coefficient determinant of our system as $D$, so $D = ad - bc$.

Cramer’s rule for a two-variable system

Look again at the answers we derived,

and notice that every numerator is itself a determinant. The denominator is $D$ in both. The numerator for $x$ is $ed - bf$, which is the determinant you get from $D$ by a single column replacement. You take $D$ and replace its first column (the $x$-column, $a$ and $c$) with the constants $e$ and $f$. The numerator for $y$ is $af - ce$, the determinant from replacing the second column (the $y$-column, $b$ and $d$) with the same constants. Name these column-replacement determinants $D_x$ and $D_y$:

With that naming, the whole solution collapses into a pattern worth memorizing. This is Cramer’s rule:

The determinant test for a unique solution

The derivation quietly answered a deeper question than “what is the solution.” It told us exactly when a solution exists and is unique. Everything hinged on whether we were allowed to divide by $D$, so $D$ answers one specific question: does this system have exactly one solution, or not?

This is the tidy end of the one-none-infinitely-many trichotomy you met at the start of the chapter. The determinant is a single number that detects which case you are in:

• $D \ne 0$: the two lines cross at one point, so there is exactly one solution.
• $D = 0$: for two variables, the two lines are parallel (no solution) or the same line (infinitely many). The determinant cannot tell those two apart by itself, only that the system is degenerate. This two-line picture is specific to two variables; with three variables the same $D = 0$ test still correctly signals “no solution or infinitely many,” but the geometry has more shapes than “parallel or the same plane,” since three planes can also fail to share a common point without any two of them being parallel.

Be careful with that last point, because it is the easiest thing to overstate. A zero determinant does not mean “no solution.” It means “not a unique solution,” which is either no solution or infinitely many. You need one more look at the constants to decide which.

See the difference on numbers. Pair $3x + y = 6$ with $6x + 2y = 12$: the second equation is the first one doubled, so every point on the line $3x + y = 6$ satisfies both, and every determinant is zero, $D = D_x = D_y = 0$. That is infinitely many solutions. Now pair the same $3x + y = 6$ with $6x + 2y = 9$ instead. The coefficients still double, $6 = 2(3)$ and $2 = 2(1)$, so $D = 0$ again, but the constant does not match that doubling, $9 \ne 2(6)$, so $D_x = 6(2) - 1(9) = 3$, which is not zero. The two lines are parallel and never meet: no solution.

Determinants of three-variable systems

For three unknowns the coefficients form a $3 \times 3$ array, and its determinant is built out of $2 \times 2$ determinants. Expand along the first row: multiply each top entry by the $2 \times 2$ determinant left after crossing out that entry’s own row and column. Then attach the alternating signs to those products in order: plus, minus, plus.

Each small determinant is the minor of the entry in front of it. The middle term carries a minus sign, which is the single most common slip, so watch for it. This method is called cofactor expansion, and it reduces any $3 \times 3$ determinant to three easy $2 \times 2$ ones.

Cramer’s rule for a three-variable system

Cramer’s rule extends to three unknowns with no new ideas, only more columns. The two-variable version came from combining the equations to make one variable’s coefficient match and cancel, then recording only what that combination implies about any solution, the same one-way reasoning as above, valid whether or not a multiplier lands on zero. Isolating one unknown among three runs through that same combining process, just applied twice instead of once, and lands on the identical pattern: each numerator swaps the constants into one column, with $D$ staying the shared denominator throughout. As before, this only pins down what any solution must satisfy; checking that those values actually work, when $D \ne 0$, is the same substitution argument as the two-variable proof, just with one more variable. Writing out that longer elimination and check in full would not teach anything the two-variable case has not already shown, so the pattern is stated here rather than rederived from scratch.

Let $D$ be the coefficient determinant of

Form $D_x$, $D_y$, and $D_z$ by replacing the $x$-, $y$-, or $z$-column with the constant column $d_1, d_2, d_3$. Then, as long as $D \ne 0$,

Cramer’s rule or row reduction?

Cramer’s rule and the row reduction from the previous lesson both handle a square system, the same number of equations as unknowns, so on that common ground, which should you reach for? Row reduction still reaches further: it also handles a system with more equations than unknowns or fewer, the kind Matrices and Systems solved, where Cramer’s rule does not apply at all, since there is no single coefficient determinant to divide by. Restricted to square systems, though, the two have opposite strengths, and the honest answer is that it depends on what you want.

Cramer’s rule is a closed formula: it writes each unknown as one determinant over another, with no back-substitution and no bookkeeping of row swaps. That makes it excellent in two situations. First, when you need only one unknown, you compute just $D$ and that variable’s numerator and stop, no matter which unknown it is. Row reduction cannot promise that: how much extra work it costs depends on where the unknown you want sits in the elimination order. Ask for the last variable eliminated and forward elimination alone hands it to you, no back-substitution needed at all; ask for an early one instead and you must first pin down every variable below it, climbing the whole staircase to reach it. Cramer’s rule pays the same one extra determinant regardless of which unknown you picked. Second, for theory, the formula shows plainly how the answer depends on the numbers in the system, which is why $D \ne 0$ reads off the unique-solution condition at a glance.

Row reduction wins on scale. A cofactor expansion of an $n \times n$ determinant fans out into smaller determinants and the work grows explosively with $n$, and Cramer needs $n + 1$ of these determinants. Row reduction grows far more gently, which is why it, not Cramer, is what people and computers use to solve large systems.

Common mistakes