RESEARCH / Focused target
Max-minus-min homomesy for rectangle interval-closed rowmotion
RELEASED FOUNDATIONS / PROPOSED CONNECTION
Research connections
All source anchors (1)
- Rectangle rowmotion max-minus-min homomesyReleased anchor
Problem
The question
Elder, Lafrenière, McNicholas, Striker and Welch, Toggling, rowmotion, and homomesy on interval-closed sets, arXiv:2307.08520v2, Conjecture 4.9, states that for every product of two finite chains the number of maximal elements minus the number of minimal elements is 0-mesic under rowmotion on interval-closed sets. Lafrenière, Lewis, McNicholas, Striker and Welch, Interval-closed set rowmotion and homomesy on products of two chains, arXiv:2505.04000v1, restates this target as Conjecture 4.2.
The exact formal target keeps all source quantifiers: positive m,n, every
N and equivalence e : Fin N ≃ Fin m × Fin n satisfying the reverse
linear-extension condition, and every order-convex initial set I, including
the empty set. literalOrbit e I is the finite set of distinct states reached
by iterating the literal successive toggle trace. maxMinusMin is the integer
cardinality difference of global maxima and minima in the actual state. The
conclusion is the sum of this statistic over the actual distinct orbit states.
Motivation
Our foothold
The result gives the exact all-rectangle resolution requested by the original conjecture. It is a uniform symbolic theorem: the legal enumeration is an explicit parameter, the orbit contains actual reachable states once each, and no finite testing range or generic rowmotion law is used.
Gap
Missing bridges
The source conjecture did not supply a proof for all products of two chains.
The formal gap is the zero sum of maxMinusMin over each literal orbit under
the complete reverse extension and interval-closed toggle definitions.
Route
Proposed approach
The proof counts strict rectangular floor corners with the frozen
RectangularCorner.rectangular_corner_card theorem. Coordinate reversal gives
the dual count. The frozen RowmotionEndpointTransport.endpoint_transport
theorem transports the actual endpoint pairs through one literal trace; a
reverse trace recovers the initial state. These facts yield an integer
coboundary for maxMinusMin. The actual finite orbit is closed and the trace
is injective on it, hence a permutation, so the coboundary telescopes.
Falsifier
What would falsify this route
A natural pair of positive dimensions, a legal complete reverse extension, and an order-convex initial set whose actual distinct trace orbit has a nonzero integer sum would refute the theorem. A test using a different toggle, repeated orbit states, a selected finite prefix, or a stronger reversal- commutation premise would test a different statement.
Evidence
Evidence to collect
The theorem result in
D5/S3/Combinatorics/Geometry/RectangleRowmotionHomomesy.lean compiles at the
pinned Lean and Mathlib versions. Its exact imported declaration has the full
telescope above, and its complete kernel axiom closure is
propext, Classical.choice, and Quot.sound. The source has exactly the
two essential definitions literalOrbit and maxMinusMin and one public
result; all endpoint, duality, counting, and orbit arguments are local to
that result. The published source correspondence is arXiv:2307.08520v2
Conjecture 4.9 and arXiv:2505.04000v1 Conjecture 4.2. Preregistration for
this exact external target is issue
11891.
Triage
Scope assessment
theorem; resolution proved for the original all-rectangle max-minus-min
homomesy conjecture. The result is admitted under
admission_basis: open-problem-resolution, with conservative
proof_shape: bind-only and utility: none. The follow-up paper's separate
signed-cardinality result, its explicit orbit descriptions for [2] × [n],
and any broader poset or higher-product statement are outside this result.
What the settlement shows
Proved in this module: let H(S) count pairs (b,a) with b maximal in
lowerClosure S \ S, a maximal in S, and both coordinates of b
strictly smaller than those of a. The rectangular corner identity and its
coordinate-reversed dual, together with endpoint transport, give
maxMinusMin S = H(R(S)) - H(S) in the integers, where
R(S) = trace(e,S,N). The reverse trace recovers every order-convex input,
so R permutes the finite set of distinct reachable states. Summing this
coboundary over that set gives zero. Trace indices beyond N retain the
completed trace state for a fixed input; repeated rowmotion instead reapplies
R to the resulting state.
Proved in this module: this mechanism covers every positive product of two
finite chains, every complete reverse extension, and every interval-closed
initial set, including the empty set. It settles the max-minus-min assertion
of arXiv:2307.08520v2 Conjecture 4.9 and its restatement in
arXiv:2505.04000v1 Conjecture 4.2. For the [2] × [n] cases discussed in the
follow-up, it supplies this same zero-sum statistic without an explicit orbit
classification. The follow-up's separate signed-cardinality result and orbit
descriptions retain their separate statements and proofs.
Outside this module: no extension to general posets or products of more than
two chains is established here. The unrestricted higher-product extension is
false: arXiv:2307.08520v2, Remark 4.10
gives counterexamples [2] × [2] × [5] and [2]^4. No relaxation of the stated
hypotheses, consequences for other source questions, or claims about their
open status follow from this result.
ASSUMED-UNVERIFIED
Unverified assumptions
The source review is bounded to the named arXiv versions and the registered exact target; it does not establish exhaustive worldwide priority or exclude unindexed, unpublished, or later proofs.