lemmastandard

Moore determinant and cofactor properties

Shows Moore cofactor C(A)=(det⁡A)A−1C(A)=(\det A)A^{-1}, nonnegative, degree m−1m-1
linear-algebra facts for cofactor and Dirichlet estimates
Lemma·lem:quaternionic-matrices·§3.2 The Moore cofactor

Statement (paper)

The Moore determinant is a homogeneous polynomial of degree mm. Its cofactor C(A)C(A) is the Hermitian matrix defined by C(A):B=ddtdet⁡(A+tB)∣t=0. C(A):B=\left.\frac{d}{dt}\det(A+tB)\right|_{t=0}. It is polynomial of degree m−1m-1, and C(A)=(det⁡A)A−1(A>0). C(A)=(\det A) A^{-1}\quad (A>0). If A≥0A\ge0, then C(A)≥0C(A)\ge0; if also AA is singular, C(A)C(A) kills the image of AA. On positive definite matrices log⁡det⁡\log\det is concave and increasing in the matrix order. For a unit vector v∈Hmv\in\HH^m let A∣v⊥A|_{v^\perp} denote the compression of the form to its quaternionic orthogonal complement. For every Hermitian AA and real tt, v∗C(A)v=det⁡(A∣v⊥),det⁡(A+tvv∗)=det⁡A+t v∗C(A)v. v^*C(A)v=\det(A|_{v^\perp}),\qquad \det(A+t vv^*)=\det A+t\,v^*C(A)v. The determinant on the zero-dimensional space is one. Finally, for a Hermitian M=(ℓc∗cD)M=\bigl(\begin{smallmatrix}\ell&c^*\\ c&D\end{smallmatrix}\bigr) with D>0D>0, put σ=ℓ−c∗D−1c\sigma=\ell-c^*D^{-1}c and let v0v_0 be the first coordinate vector. Then det⁡M=(det⁡D)σ,M−σv0v0∗≥0. \det M=(\det D)\sigma,\qquad M-\sigma v_0v_0^*\ge0 .

In plain English

For quaternionic Hermitian m×mm\times m matrices, the Moore determinant is homogeneous of degree mm, and its cofactor C(A)C(A), defined as the derivative of det⁡(A+tB)\det(A+tB) in direction BB, is a degree m−1m-1 polynomial equal to (det⁡A)A−1(\det A)A^{-1} for A>0A>0. Moreover C(A)≥0C(A)\ge0 for A≥0A\ge0, log⁡det⁡\log\det is concave and monotone on positive definite matrices, v∗C(A)v=det⁡(A∣v⊥)v^*C(A)v=\det(A|_{v^\perp}) for unit vv, and a Schur complement formula gives det⁡M=(det⁡D)σ\det M=(\det D)\sigma with M−σv0v0∗≥0M-\sigma v_0v_0^*\ge0.

Why it's here

This lemma collects the standard linear-algebra facts about quaternionic determinants that the analytic arguments need: the cofactor plays the role of the linearized determinant, and the compression and Schur formulas let one compute or bound it along a direction. Later results such as the cofactor divergence identity, the energy balance, the Dirichlet problem estimates and the double normal derivative bound all rely on these facts.
Classified standard:
Standard linear algebra of quaternionic Hermitian matrices via the Moore determinant: cofactor formula, log-concavity, compression and Schur complement identities (Moore 1922; see Aslaksen 1996 'Quaternionic determinants', and Alesker's work on quaternionic Monge–Ampère).

Sources

PaperOpen
Section
§3.2 The Moore cofactor
Page
—
Label
lem:quaternionic-matrices
Line
sections/calculus.tex:99
Leannot aligned

Phase 2 aligns this node to a Lean declaration.