Skip to content

Extract a nonzero ordered minor from full column rank #299

Description

@PerAlexandersson

Parent

Prerequisite for #292 and Karlin, Total Positivity, Vol. I, Chapter V, Section 1, Proposition 1.2.

Required theorem

For A : Matrix (Fin n) (Fin m) Real with injective mulVec, and every strictly increasing selected column map cols : Fin q -> Fin m, prove that some strictly increasing rows : Fin q -> Fin n gives a nonzero determinant of A.submatrix rows cols.

Proposed name: Matrix.exists_ordered_minor_ne_zero_of_mulVec_injective.

Required proof route

Use Matrix.mulVec_injective_iff to obtain linear independence of all columns, restrict to the selected columns, and extract a nonzero maximal coordinate/minor. An exterior-power proof is acceptable if it packages the coordinate determinant cleanly; an equivalent rank-versus-minor proof is also acceptable.

Acceptance criteria

  • Uses full column rank/injective mulVec, not the insufficient assumption q <= rank A.
  • Produces a strictly increasing row selector.
  • No sorry, axiom, or theorem-shaped Prop definition.
  • Reusable independently of Gaussian matrices and sign consistency.
  • Full CI is green.

Mathlib currently has the injectivity-to-column-independence bridge and square determinant/invertibility lemmas, but no confirmed rectangular nonzero-minor extraction theorem.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions