Maths WorkshopMaths problems with full solutions
Exercise 2120Linear algebra › DeterminantsL1Added 3 Nov 2026

Hadamard's inequality

#Hadamard inequality#determinant#column norms

Statement

Let AA be an n×nn\times n real matrix with columns C1,…,CnC_1,\ldots,C_n. Show that
∣det⁡A∣≤∏i=1n∥Ci∥.|\det A|\le\prod_{i=1}^n\|C_i\|.

Hint

Use Gram-Schmidt or the geometric interpretation of the determinant as a volume.

Show solutionHide solution

Solution

Geometric interpretation. ∣det⁡A∣|\det A| is the volume of the parallelotope spanned by C1,…,CnC_1,\ldots,C_n. This volume equals the product of the lengths of the columns *only* when the columns are orthogonal. In general, the volume is at most the product of lengths (the "base times height" formula, with the height bounded by the corresponding side).

Formal proof by Gram-Schmidt. Apply Gram-Schmidt to C1,…,CnC_1,\ldots,C_n: let q1,…,qnq_1,\ldots,q_n be the orthonormal basis, and rijr_{ij} the coefficients. Then A=QRA=QR with RR upper triangular, and ∣det⁡A∣=∣det⁡Q∣∣det⁡R∣=∏i∣rii∣|\det A|=|\det Q||\det R|=\prod_i|r_{ii}|. But ∣rii∣≤∥Ci∥|r_{ii}|\le\|C_i\| (each riir_{ii} is the length of the component of CiC_i orthogonal to the previous columns, hence bounded by ∥Ci∥\|C_i\|).

Conclusion. ∣det⁡A∣≤∏∥Ci∥|\det A|\le\prod\|C_i\|, with equality iff the columns are orthogonal.