The Determinant: Volume and Multilinearity
~7 min read · Tier toggle changes depth (T1 plain · T2 standard · T3 rigour)
You should be comfortable with the definition of a linear map between vector spaces and the concept of a basis. Familiarity with the summation notation and the symmetric group \(S_n\) is helpful but not strictly required to follow the geometric intuition.
Why this matters
The determinant is the single scalar that encodes how a linear transformation distorts space.
In physics, we rarely compute determinants by hand for large matrices, yet the concept is foundational. It tells us whether a system of linear equations has a unique solution, whether a set of vectors is linearly independent, and, most importantly for mechanics and quantum theory, how volumes and orientations transform under change of basis. Understanding the determinant as a measure of "signed volume" provides the geometric intuition necessary to grasp the spectral theorem and the behavior of operators in higher dimensions.
Volume in the Abstract
Imagine you have a set of \(n\) vectors in \(n\)-dimensional space. These vectors form the edges of a generalized box, or parallelepiped. The determinant of the matrix formed by these vectors is the "signed volume" of this box. If the vectors are linearly dependent, they collapse into a lower-dimensional subspace, and the volume is zero. If you swap two vectors, the orientation of the box flips, and the sign of the volume changes. If you stretch one vector by a factor of \(k\), the volume scales by \(k\). These three properties—scaling, antisymmetry, and linearity—are not just rules for calculation; they are the defining characteristics of the determinant. It is the unique function that captures how linear maps stretch space while preserving the algebraic structure of the vector space.
Alternating Multilinearity
The determinant is defined by its behavior with respect to its columns (or rows). It is multilinear, meaning it is linear in each column separately. If you replace one column \(\mathbf{a}_i\) with a linear combination \(\alpha \mathbf{u} + \beta \mathbf{v}\), the determinant splits into a sum of two determinants, with the scalar pulled out. This property allows us to expand the determinant using cofactors, a technique that reduces the calculation of an \(n \times n\) determinant to that of \((n-1) \times (n-1)\) determinants.
More crucially, the determinant is alternating. This means that if any two columns are identical, the determinant is zero. Furthermore, swapping any two columns changes the sign of the determinant. This antisymmetry is the algebraic reflection of the geometric fact that a "flat" box has no volume, and that reversing the order of edges reverses the orientation of the space. Together, multilinearity and alternation constrain the determinant so tightly that, once we fix its value on the standard basis vectors, its value on any other set of vectors is uniquely determined.
The Product Law
One of the most powerful properties of the determinant is the product law: \(\det(AB) = \det(A)\det(B)\). This is not a definition but a consequence of the uniqueness of the alternating multilinear form. When you apply a linear map \(A\) followed by \(B\), the total volume scaling is the product of the individual scalings. This property is essential in physics because it ensures that the determinant behaves consistently under composition of transformations. For example, in the change of variables formula for integrals, the Jacobian determinant of two successive changes of variables is the product of the two individual Jacobian determinants, which is the chain rule for volume elements.
The product law also implies that a matrix is invertible if and only if its determinant is non-zero. If \(\det(A) = 0\), then \(A\) maps some non-zero vector to zero, meaning it collapses the space and cannot be reversed. This connects the algebraic property of invertibility directly to the geometric property of non-zero volume scaling.
Connection to Eigenvalues
The determinant plays a central role in the eigenvalue problem. The eigenvalues \(\lambda\) of a matrix \(A\) are the roots of the characteristic polynomial \(\det(A - \lambda I) = 0\). This equation asks for the values of \(\lambda\) for which the operator \(A - \lambda I\) is singular, i.e., for which it collapses the space. The determinant thus provides the bridge between the geometric action of the operator and its algebraic spectrum. In quantum mechanics, the eigenvalues of the Hamiltonian operator correspond to the allowed energy levels, and the determinant of the Hamiltonian minus the energy times the identity operator determines which energies are permitted.
Furthermore, the determinant of a matrix is equal to the product of its eigenvalues (counting algebraic multiplicity). This means that the total volume scaling of a transformation is the product of the scaling factors along its principal axes. This insight is crucial for understanding stability in dynamical systems: if any eigenvalue has a magnitude greater than one, the volume of a small region around the fixed point will expand under iteration, indicating instability.
The derivation
Determinant as the Unique Alternating Multilinear Form
On the space of \(n\times n\) matrices over a field \(K\), there is exactly one function \(D:(K^{n})^{n}\to K\) of the ordered columns \((\mathbf{a}_1,\dots,\mathbf{a}_n)\) that is (i) multilinear — linear in each column separately, (ii) alternating — \(D=0\) whenever two columns coincide, and (iii) normalized — \(D(\mathbf{e}_1,\dots,\mathbf{e}_n)=1\) on the standard basis. That unique function is the determinant \(\det\), it equals the Leibniz sum \(\det A=\sum_{\sigma\in S_n}\operatorname{sgn}(\sigma)\prod_{i}A_{\sigma(i)\,i}\), and from the characterization alone it obeys the product law \(\det(AB)=\det(A)\det(B)\).
A common misconception is that the determinant measures the "size" of a matrix in a simple Euclidean sense. In fact, the determinant is sensitive to the *orientation* and *linear independence* of the basis vectors, not just their lengths. A matrix with very large entries can have a determinant of zero if its columns are linearly dependent. Conversely, a matrix with small entries can have a large determinant if the columns are nearly orthogonal and of significant length. Always interpret the determinant in the context of the basis it acts upon. C
Worked example
Problem. Consider the linear transformation \(T: \mathbb{R}^2 \to \mathbb{R}^2\) defined by the matrix \(A = \begin{pmatrix} 2 & 1 \\ 1 & 2 \end{pmatrix}\). Calculate the determinant of \(A\) and interpret its physical meaning in terms of area scaling. Then, find the eigenvalues of \(A\) and verify that their product equals the determinant.
- Compute the determinant using the formula for a \(2 \times 2\) matrix: \(\det(A) = (2)(2) - (1)(1) = 4 - 1 = 3\).
- Interpretation: The transformation \(T\) scales the area of any region in the plane by a factor of 3. Since the determinant is positive, the orientation of the plane is preserved.
- Find the eigenvalues by solving the characteristic equation \(\det(A - \lambda I) = 0\). The matrix \(A - \lambda I\) is \(\begin{pmatrix} 2-\lambda & 1 \\ 1 & 2-\lambda \end{pmatrix}\).
- The determinant is \((2-\lambda)^2 - 1 = 0\). Expanding, we get \(4 - 4\lambda + \lambda^2 - 1 = \lambda^2 - 4\lambda + 3 = 0\).
- Factoring the quadratic, we get \((\lambda - 1)(\lambda - 3) = 0\). Thus, the eigenvalues are \(\lambda_1 = 1\) and \(\lambda_2 = 3\).
- Verify the product: \(\lambda_1 \lambda_2 = 1 \times 3 = 3\), which matches the determinant calculated in step 1.
Units check. The determinant is dimensionless if the matrix entries are dimensionless. If the matrix represents a physical transformation with units, the determinant carries the product of the units of the scaling factors. In this case, the area scaling factor is a pure number, consistent with the dimensionless nature of the matrix entries.
Three quick tests
- What is the determinant of a matrix with two identical columns?
Answer
Zero, because the columns are linearly dependent, and the volume of the parallelepiped is zero. - If \(\det(A) = 5\) and \(\det(B) = 2\), what is \(\det(AB)\)?
Answer
10, by the product law \(\det(AB) = \det(A)\det(B)\). - How does the determinant relate to the eigenvalues of a matrix?
Answer
The determinant is equal to the product of the eigenvalues (counting algebraic multiplicity).
Problem set for PU-104 →
Each derivation in this unit ends with five graded problems with full solutions.
Lecture notes drafted with a local model and checked line by line by the editor; every number in the worked example was recomputed.