| paper | d4-journals-geb-2024-0088 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | unclear |
The paper’s named results concern properties and characterizations of the stable correspondence; none asserts an algorithm, computational hardness, or another qualifying computational result. Stable Mass Matching is a coherent adjacent computational extension, but the paper does not establish it or connect it to a computational result, so no anchor qualifies.
fails bit a — no named computational result to mirror
Theorem 1 establishes own-side singles monotonicity, while Theorems 2 and 3 characterize the stable solution; the other numbered results concern compatibility and restricted variants of these axiomatic properties. None supplies a computational result to mirror.
The strongest positive case is a high multiplicity version of finding a stable matching, but there is an important limit: this paper contains no numbered computational result—no theorem about polynomial time, hardness, or parameterized complexity. Its numbered results are axiomatic. So there is no qualifying computational anchor, and I would not present one as if the paper had established an algorithmic result.
The closest conceptual hook is Theorem 1, proved in this paper. It says that the stable solution is own-side singles monotonic: if an agent \(h\) is unmatched in every stable matching before a change, and the change only raises \(h\)’s position in the preferences of agents on the other side, then for every stable matching \(\mu\) before the change, there is a stable matching \(\nu\) afterward such that every agent on \(h\)’s side other than \(h\) weakly prefers their assignment in \(\mu\) to their assignment in \(\nu\), under their unchanged preferences. That is a result about a property of the stable correspondence, not about computing it. Theorem 2 and Theorem 3 likewise characterize the stable solution by axioms; they are not computational anchors.
The best unanchored computational mirror I can offer is Stable Mass Matching. A large residency or hiring market is a plausible regime: many applicants and one-position employers, with applicants grouped by standardized credentials and preferences, and employers grouped by common priorities over applicant types. There might be thousands of agents but only tens or hundreds of distinct types. A type must capture the full information used by the model; the aggregation is credible only where identity-level distinctions do not affect preferences or stability.
An instance consists of applicant types \(I\), employer types \(J\), rational masses \(\mu_i,\nu_j\), and strict rankings by each type over the opposite-side types and being unmatched. The decision variables are matched masses \(x_{ij}\) and unmatched masses \(u_i,v_j\), satisfying
\[
\sum_{j\in J}x_{ij}+u_i=\mu_i,\qquad
\sum_{i\in I}x_{ij}+v_j=\nu_j.
\]
Mass is a share of the population: in an atomless cohort, each individual still receives at most one partner, while a type’s members can be divided among different partner types. There is no scalar objective; the task is to output a stable mass assignment. Require individual rationality, so \(x_{ij}=0\) whenever either type finds the other unacceptable. For every mutually acceptable pair \(i,j\), define
\[
A_{ij}=u_i+\sum_{j':\,j\succ_i j'}x_{ij'},\qquad
B_{ij}=v_j+\sum_{i':\,i\succ_j i'}x_{i'j}.
\]
Stability requires \(A_{ij}=0\) or \(B_{ij}=0\): there cannot be positive mass on both sides who would prefer to match with each other over their current assignment. A solution is the rational array \((x,u,v)\) satisfying these conditions. The input is described by the support sizes and binary bit length of the masses, rather than by expanding each cohort into individuals.
I would expect this problem to be tractable by a deferred-acceptance procedure that proposes and holds rational masses by type. That is a conjecture about the mirror, not a result of this paper. It would make the market’s population size less important than its number of types. The paper’s stable solution gives the authors a recognizable object to compute, though they may reasonably say this is an algorithmic extension of their subject rather than a mirror of their contribution. A useful follow-up would be to establish the exact dictionary between integer-mass instances and finite matching markets, and ask whether a mass analogue of Theorem 1 holds. The latter is axiomatic and outside ChoCo’s computational scope.
The weakest point is that the paper ranks named individuals, while the proposed model ranks types. If identity-specific rankings affect blocking, aggregating them changes the problem. The mirror is persuasive only in a market regime where cohort membership preserves everything relevant to preferences and stability. That makes this a plausible, bounded positive case—not an anchor-backed continuization of the paper’s stated results.
The strongest case is that this paper has no computational result to continuize. Theorem 1 says something about how the stable correspondence responds to a preference change; Theorems 2 and 3 characterize that correspondence through axioms. None asks for an algorithm, an optimization, or a complexity classification. A type-level version of those claims would still be axiomatic, which ChoCo puts outside scope.
Stable Mass Matching is the best rescue, but it changes the paper’s question. The paper takes the set of stable matchings as the object of study; it asks which solutions satisfy singles monotonicity and which axioms characterize stability. Computing a stable mass allocation is a coherent adjacent problem, but it is not a computational version of any result here. A formal dictionary between repeated finite agents and rational masses could help validate that new model; it would not turn the paper’s axioms into computational results. To make the monotonicity question computational, one would need to add an objective or task—such as minimizing the cost of a preference change—that the paper does not supply.
So I would not treat this paper as a ChoCo mirror. The limit of that conclusion is the word “any”: stable mass matching could be worthwhile as a separate project. The case against is strongest at the paper-to-program boundary, not as a proof that no useful continuous matching problem can be posed.
The adversarial triple: the proponent anchors on up to three named results; the opponent sees that case and must defeat every anchor; the judge decides which case convinced it. These are the pipeline’s own outputs, generated by tools/triple_run.py — no human edited them. The paper’s own text is not reproduced here beyond the quoted statement above.