Lemma 10.3.4 — complementary rank identity
ProvedPutermanMDP.Sensitive.complementary_rankblackwell-optimalitymdpp2o-gran-per-chapterp2o-plan-bookp2o-v1sensitive-discount
Let be real matrices. If has full rank and , then
The identity connects the rank of a chain's limiting matrix with the rank of its deviation matrix. It does not assume symmetry or add the reverse product condition .
Preamble
import Mathlib
Formal statement
namespace PutermanMDP.Sensitive
theorem complementary_rank (m : ℕ) (F H : Matrix (Fin m) (Fin m) ℝ)
(hfull : (F + H).rank = m) (hzero : F * H = 0) :
F.rank + H.rank = m := by sorry
end PutermanMDP.Sensitive
Source
Puterman, Markov Decision Processes, Wiley, 1994, p. 517 (PDF 533), Lemma 10.3.4; https://doi.org/10.1002/9780470316887
Read-back
What the Lean code literally says, in plain math · claude-opus-5-5
Let , and let and be real matrices. Assume:
- , so is invertible;
- the product , in that order, is the zero matrix.
Then
Nothing is assumed about the reverse product .
Degenerate cases. If , all three ranks are and the equality reads .
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.