AFTD Auto-Formalizing Theoretical Domains

transpose_one_add_diagonal_mul_mul

theoremverified

For any finite index type n with decidable equality, any real n-by-n matrix M, and any vector d : n → ℝ, the transpose of 1 + Matrix.diagonal d * M * Matrix.diagonal d equals 1 + Matrix.diagonal d * M.transpose * Matrix.diagonal d.

Statement

theorem transpose_one_add_diagonal_mul_mul {n : Type*} [Fintype n] [DecidableEq n]
    (M : Matrix n n ℝ) (d : n → ℝ) :
    (1 + Matrix.diagonal d * M * Matrix.diagonal d).transpose =
      1 + Matrix.diagonal d * M.transpose * Matrix.diagonal d
Source
scout/2611-07815-newton-steps-are-faster-for-linear-complementarity-problems.md:529
Verified
07 Nov 2026
Axioms
Classical.choiceQuot.soundpropext

Read back from the Lean

For every finite type $n$ with decidable equality, every $n \times n$ matrix $M$ with real entries, and every function $d : n \to \mathbb{R}$, the transpose of the matrix $(I + D M D)$ equals $I + D M^{\top} D$, where $I$ is the $n \times n$ identity matrix and $D = \text{diagonal}(d)$ is the diagonal matrix whose diagonal entries are given by $d$.

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

The proof, explained

The identity states that transposing I + D * M * D yields I + D * Mᵀ * D, where D is the diagonal matrix with entries d. The proof invokes the simplifier augmented with matrix associativity (`Matrix.mul_assoc`). The simplifier distributes the transpose across addition and products, uses the fact that both the identity and diagonal matrices are symmetric, and closes the goal.

  1. The transpose of I + D * M * D is equal to I + D * Mᵀ * D, where D = Matrix.diagonal d. automation

    The simplifier evaluates the transpose of the sum and products. Transposition distributes over addition and reverses matrix products, transforming (I + (D * M) * D)ᵀ into Iᵀ + Dᵀ * (D * M)ᵀ = Iᵀ + Dᵀ * Mᵀ * Dᵀ. Because the identity matrix and diagonal matrices are symmetric (Iᵀ = I and Dᵀ = D), and matrix multiplication is associative via `Matrix.mul_assoc`, this simplifies directly to I + D * Mᵀ * D.

    line 6

    uses Matrix.mul_assoc

      simp [Matrix.mul_assoc]
Matrix.mul_assoc
Matrix multiplication is associative: (L * M) * N = L * (M * N) for conformable matrices.

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

/-- The transpose of I_n + DMD is I_n + DMᵀD for any diagonal matrix D. -/theorem transpose_one_add_diagonal_mul_mul {n : Type*} [Fintype n] [DecidableEq n]    (M : Matrix n n ℝ) (d : n → ℝ) :    (1 + Matrix.diagonal d * M * Matrix.diagonal d).transpose =      1 + Matrix.diagonal d * M.transpose * Matrix.diagonal d := by  simp [Matrix.mul_assoc]
See it in the graphDiscuss this resultChallenge it