A single number that captures everything geometric about a square matrix: how much it stretches volumes, whether it flips orientation, and whether the transformation it encodes can be undone. The determinant is the most compressed summary of a linear map.
10 min readPrereqs: Matrices, Linear TransformationsUpdated 2026·05·17
What you'll leave with
The determinant of a matrix is the signed volume-scaling factor of the linear transformation it represents.
For 2×2: $ad - bc$. For 3×3: cofactor expansion with alternating signs.
Sign tells you orientation — positive preserves it, negative flips it.
$\det = 0$ exactly when the matrix is singular (non-invertible) and the columns are linearly dependent.
1. The geometric meaning
Before the formula, the picture. Take a 2×2 matrix and read its columns as two vectors in the plane:
$$ A = \begin{pmatrix} a & b \\ c & d \end{pmatrix}, \quad \mathbf{u} = \begin{pmatrix} a \\ c \end{pmatrix}, \quad \mathbf{v} = \begin{pmatrix} b \\ d \end{pmatrix} $$
Those two vectors span a parallelogram. The determinant of $A$ is the signed area of that parallelogram. Not "related to" the area — literally the area, with a sign attached.
Push this up a dimension and the same idea works. For a 3×3 matrix, the columns are three vectors in space, and they span a parallelepiped — a skewed box. The determinant is its signed volume. Push it up further to $n \times n$, and the determinant is the signed $n$-dimensional volume of the parallelotope spanned by the columns.
Now read the matrix as a linear transformation $T(\mathbf{x}) = A\mathbf{x}$. The columns of $A$ are the images of the standard basis vectors under $T$. The unit square (in 2-D) or unit cube (in 3-D) has area or volume equal to $1$, and after $T$ acts on it, the result has area or volume equal to $|\det A|$. So:
Determinant (geometric definition)
$\det A$ is the signed factor by which the linear transformation $A$ scales $n$-dimensional volume. Apply $A$ to any region of volume $V$; the image has volume $|\det A| \cdot V$.
Why this matters
Every algebraic property of the determinant — the alternating signs, the multiplicativity $\det(AB) = \det A \cdot \det B$, the fact that swapping rows flips its sign — falls out of this single geometric idea. You can derive the formulas by remembering what they're for.
2. Computing the 2×2 determinant
This is the case to know cold. For
$$ A = \begin{pmatrix} a & b \\ c & d \end{pmatrix} $$
the determinant is
$$ \det A = ad - bc. $$
"Diagonal product minus anti-diagonal product." Memorize the shape; the formula reads off the page.
A quick check: the identity matrix should leave area alone, so its determinant should be $1$.
For 3×3, the formula gets longer but the structure is the same: it's the signed volume of the parallelepiped spanned by the three columns. The standard way to compute it is cofactor expansion along the first row.
Write the matrix as
$$ A = \begin{pmatrix} a & b & c \\ d & e & f \\ g & h & i \end{pmatrix}. $$
For each entry in the top row, cross out its row and column, take the determinant of the 2×2 block that remains, multiply by the entry, and attach the alternating sign $+, -, +$:
$$ \det A = a \begin{vmatrix} e & f \\ h & i \end{vmatrix} - b \begin{vmatrix} d & f \\ g & i \end{vmatrix} + c \begin{vmatrix} d & e \\ g & h \end{vmatrix} $$
Expanding the 2×2 determinants gives the explicit form:
That alternating pattern — $+, -, +$ — is the most failure-prone piece. The middle term is subtracted; forget that minus sign and you'll silently get the wrong answer.
Going further
For general $n \times n$, the determinant is defined recursively by cofactor expansion along any row or column (with the same alternating-sign pattern, called the checkerboard of signs). For anything bigger than 3×3, no one computes by hand — row reduction or LU decomposition is far faster. The formula matters; the procedure scales poorly.
4. Sign tells you orientation
The magnitude $|\det A|$ is the volume scaling factor. The sign tells you something separate: whether the transformation preserves or reverses orientation.
"Orientation" is the handedness of your coordinate system. In 2-D it's clockwise vs. counterclockwise — go around the unit square $\mathbf{e}_1 \to \mathbf{e}_2$ and you're going counterclockwise. In 3-D it's right-hand vs. left-hand: curl your right hand's fingers from $\mathbf{e}_1$ to $\mathbf{e}_2$ and your thumb points along $\mathbf{e}_3$.
$\det A > 0$: $A$ preserves orientation. Rotations, scalings by positive factors, and shears all fall here.
$\det A < 0$: $A$ reverses orientation. Reflections always do this.
The cleanest example: reflection across the $y$-axis sends $(x, y) \mapsto (-x, y)$. Its matrix is
Area is preserved (the magnitude is $1$), but the plane is flipped — the unit square goes from being traced counterclockwise to clockwise. That sign flip is the determinant telling you a reflection happened.
Same magnitude, but the positive sign tells you orientation is preserved — as it should be for a rotation.
5. The zero determinant
The most useful single fact about determinants:
A square matrix is invertible if and only if its determinant is nonzero.
The geometric reason is hard to miss once you see it. $\det A = 0$ means $A$ scales volume by zero — it collapses the unit cube to something of lower dimension. In 2-D, the parallelogram flattens into a line segment (or a point). In 3-D, the parallelepiped flattens into a plane (or a line, or a point).
Once you've collapsed a dimension, you can't recover it. Multiple input vectors get mapped to the same output, so no inverse function exists. Algebraically, the columns of $A$ become linearly dependent: one column is a combination of the others, and the matrix is called singular.
This is the test you'll reach for over and over: to decide whether a matrix can be inverted, whether a linear system $A\mathbf{x} = \mathbf{b}$ has a unique solution, whether a set of vectors is linearly independent — compute the determinant and check whether it's zero.
A determinant near zero isn't quite the same
"Nonzero determinant" is a yes/no condition, but in numerical practice a determinant that's tiny often indicates a matrix that's only barely invertible — a so-called ill-conditioned matrix. The condition number, not the determinant, is the right tool for numerical stability questions.
6. Playground: signed area of the parallelogram
Slide the entries of a $2 \times 2$ matrix and watch the parallelogram its columns span. The fill color tracks the sign of the determinant — orange when orientation is preserved, red when it's flipped, fading to nothing as the matrix becomes singular. The dashed unit square is there for scale.
2.01.00.01.0
det = ad − bc2.00
preserves orientation (det > 0)
|det| = area of parallelogram = 2.00
2.0
1.0
0.0
1.0
Copied!
Try it
Start from the identity and slowly drag $d$ down through zero. The parallelogram squashes flat at $d = 0$ (where $\det = 0$) and then re-emerges in red — same shape, opposite orientation. The sign of the determinant is exactly the moment the parallelogram passes through degeneracy.
7. Algebraic properties
A handful of identities turn the determinant from a recipe into a tool. Every one of them is a consequence of the geometric definition — they say something specific about how matrix operations change signed volume.
Multiplicativity
For any two square matrices $A$ and $B$ of the same size:
$$ \det(AB) = \det(A) \cdot \det(B). $$
Geometrically obvious: apply $B$ then $A$, and volume scales by $\det B$ then by $\det A$ — the total scaling is the product. A useful corollary: $\det(A^{-1}) = 1/\det(A)$ whenever $A$ is invertible, because $A A^{-1} = I$ has determinant $1$.
Transpose preserves the determinant
$$ \det(A^\top) = \det(A). $$
This is the symmetric flip-side: cofactor expansion gives the same result along any row or any column, so swapping rows for columns leaves the answer alone. One practical use: any property stated for rows automatically holds for columns too.
Row (and column) operations
The three elementary row operations each affect the determinant in a predictable way. These are how row reduction actually computes determinants in practice.
Operation
Effect on $\det A$
Swap two rows
Multiplies $\det A$ by $-1$
Multiply a row by scalar $k$
Multiplies $\det A$ by $k$
Add a multiple of one row to another
Leaves $\det A$ unchanged
The third one is the workhorse: shears don't change volume. Combined with the swap rule for sign-tracking, you can reduce any matrix to triangular form, and then the determinant is just the product of the diagonal entries — a stunningly faster procedure than cofactor expansion for anything bigger than 3×3.
Scalar multiple of an entire matrix
Scaling every entry of an $n \times n$ matrix by $k$ scales $n$ separate rows by $k$, so:
$$ \det(kA) = k^n \det(A). $$
The exponent $n$ surprises people who pattern-match to $\det(kA) = k\det(A)$ — it is wrong unless $n = 1$.
Triangular matrices
If $A$ is upper- or lower-triangular (all entries above or below the diagonal are zero), then $\det A$ is just the product of the diagonal entries. The off-diagonal entries contribute nothing — every term in the cofactor expansion that involves them ends up multiplied by zero.
Together these say
Row-reduce $A$ to triangular form $U$, tracking how the determinant changed at each step (swaps flip the sign, row-scaling by $k$ multiplies by $k$, adding multiples of rows does nothing). Then $\det A$ is whatever you've accumulated times the product of $U$'s diagonal entries. This is the $O(n^3)$ algorithm computers actually use — cofactor expansion is $O(n!)$ and unusable for large $n$.
8. Cramer's rule
Cramer's rule is a closed-form solution to a square linear system $A\mathbf{x} = \mathbf{b}$ when $\det A \neq 0$. For each unknown $x_i$:
$$ x_i = \frac{\det A_i}{\det A}, $$
where $A_i$ is the matrix obtained by replacing column $i$ of $A$ with the right-hand side $\mathbf{b}$. Every other column stays the same.
A concrete 2×2 example. Solve
$$ \begin{aligned} 2x + 3y &= 8, \\ x - y &= 1. \end{aligned} $$
Cramer's rule is theoretically elegant and great for proofs, small symbolic systems, and seeing exactly how the solution depends on the data. For numerical work it is terrible — solving an $n \times n$ system requires $n+1$ determinants, each $O(n!)$ by cofactor expansion or $O(n^3)$ by row reduction. Gaussian elimination directly solves the same system in a single $O(n^3)$ pass. Use Cramer for insight, not computation.
9. Common pitfalls
The minus sign in the 3×3 expansion
Cofactor expansion is $a(ei - fh) - b(di - fg) + c(dh - eg)$, not $a(ei - fh) + b(di - fg) + c(dh - eg)$. The middle term carries a negative sign from the alternating $+, -, +$ checkerboard. Forgetting it is the most common single error in a 3×3 calculation. (The Rule of Sarrus diagonal trick exists to dodge this, but it only works for 3×3 — don't carry it to 4×4.)
Determinants are only defined for square matrices
A 2×3 matrix has no determinant — full stop. The geometric picture requires equal numbers of input and output dimensions; otherwise "scaling factor of $n$-D volume" isn't well-defined. If someone writes $\det A$ for a non-square $A$, they mean something else (like a determinant of a related square matrix such as $A^\top A$).
Sign tells you orientation, not magnitude
A determinant of $-7$ does not mean the area shrank by a factor of $-7$ — areas can't be negative. It means the matrix scales area by $7$ and flips orientation. Always read $|\det A|$ for size and the sign of $\det A$ for handedness.
$\det(A + B) \neq \det A + \det B$
The determinant is multiplicative: $\det(AB) = \det A \cdot \det B$. But it is not linear in the usual sense — adding two matrices generally has nothing to do with adding their determinants. Don't pattern-match against linearity here.
10. Worked examples
Five short calculations. Try each before opening the solution; the goal is to see whether your steps match the canonical recipe.
Example 1 · Compute the 2×2 determinant of $\begin{pmatrix} 4 & 3 \\ 6 & 2 \end{pmatrix}$
Apply the formula $ad - bc$ with $a = 4$, $b = 3$, $c = 6$, $d = 2$:
$$ \det A = (4)(2) - (3)(6) = 8 - 18 = -10. $$
The columns span a parallelogram of area $10$, and the negative sign tells you that going from column 1 to column 2 reverses the standard (counterclockwise) orientation.
Example 2 · Compute the 3×3 determinant of $\begin{pmatrix} 1 & 2 & 3 \\ 0 & 4 & 5 \\ 1 & 0 & 6 \end{pmatrix}$
Step 1. Expand along the first row with alternating signs $+, -, +$:
Example 3 · Is $\begin{pmatrix} 2 & -3 \\ -4 & 6 \end{pmatrix}$ invertible?
Step 1. Compute the determinant:
$$ \det A = (2)(6) - (-3)(-4) = 12 - 12 = 0. $$
Step 2. Since $\det A = 0$, the matrix is not invertible.
Geometric reason. The second column $(-3, 6)$ is exactly $-\tfrac{3}{2}$ times the first column $(2, -4)$, so the two column vectors lie on the same line through the origin. The "parallelogram" collapses to a line segment of zero area.
Example 4 · Does $\begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}$ describe a reflection?
Step 1. Compute the determinant:
$$ \det A = (1)(-1) - (0)(0) = -1. $$
Step 2. The magnitude is $1$ (area preserved) and the sign is negative (orientation flipped). Both signatures of a reflection.
Step 3. Read the matrix to identify which reflection. The map sends $(x, y) \mapsto (x, -y)$ — points are flipped across the $x$-axis. So yes, it's the reflection across the $x$-axis.
(A determinant of $-1$ is necessary but not sufficient to be a reflection: any orientation-reversing isometry has $\det = -1$, but a non-isometry like $\begin{pmatrix} 2 & 0 \\ 0 & -\tfrac{1}{2} \end{pmatrix}$ also has $\det = -1$ without being a true reflection. Reflections are orthogonal matrices with $\det = -1$.)
Example 5 · Find the area of the parallelogram with vertices $(0,0)$, $(5, 1)$, $(2, 3)$, $(7, 4)$
Step 1. Identify the two edge vectors from the origin:
The positive sign also tells you that, traversed in the order $\mathbf{u} \to \mathbf{v}$, the parallelogram is oriented counterclockwise.
Sources & further reading
The determinant is one of the most-written-about objects in undergraduate mathematics. The sources below approach it from different angles — pick by what you need next.
A wide overview: definitions for $n \times n$, all the major properties, and connections to topics like the characteristic polynomial, exterior algebra, and computational complexity. Best when you want to see where this idea reaches.
Dense, formal reference. Use this when you want the determinant stated in the language professional mathematicians use, including the permutation-sum definition and the full list of algebraic identities.
Step-by-step videos and practice problems for hand-computing 2×2 and 3×3 determinants, plus the properties most often tested in coursework. Best if you want to drill the mechanics until they're automatic.
The visual intuition behind everything on this page — Episode 6, "The determinant," is the canonical explainer for the volume-scaling, orientation-flipping geometric picture. Watch this if any part of the geometric story above felt abstract.
Test your understanding
A quiz that builds from easy to hard. Pick an answer to get instant feedback and a worked explanation. Your progress is saved in this browser — come back anytime to continue.