OctoInformBeta

OpenAI preprint, September 2026 · 2. An entropy inequality with one zero block · Section 2

Gram matrices and block masses: Gram matrices and squared Hilbert-Schmidt masses for complex rectangular blocks

A rectangular matrix with one zero block.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

The entropy inequality with one zero block concerns a rectangular complex matrix partitioned into two block rows and two block columns. Let A∈ℂr×sA\in\mathbb C^{r\times s}, B∈ℂr×uB\in\mathbb C^{r\times u}, and C∈ℂt×sC\in\mathbb C^{t\times s}, where the dimensions are finite. Write X*X^* for the conjugate transpose of a complex matrix XX.

The Definition

Definition 1. The matrix MM with one zero block associated with AA, BB, and CC is the rectangular matrix defined by (1). M=(ABC0)∈ℂ(r+t)×(s+u).M=\begin{pmatrix}A&B\\ C&0\end{pmatrix} \in\mathbb C^{(r+t)\times(s+u)}.(1)✦ For a rectangular complex matrix X∈ℂm×nX\in\mathbb C^{m\times n}, its Gram matrix G(X)G(X) is the square matrix defined by (2). G(X)=XX*∈ℂm×m.G(X)=XX^*\in\mathbb C^{m\times m}.(2)✦ The squared Hilbert-Schmidt mass mass⁡(X)\operatorname{mass}(X) of a rectangular complex matrix X∈ℂm×nX\in\mathbb C^{m\times n} is the real scalar in (3). mass⁡(X)=Re⁡Tr⁡(XX*).\operatorname{mass}(X)=\operatorname{Re}\operatorname{Tr}(XX^*).(3)✦

What Is Checked

The definitions apply to arbitrary finite complex matrices, with compatible block sizes for the construction in (1).