| paper | d4-journals-jet-2024-0052 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
The paper’s named results establish equilibrium existence and characterize when it holds, but none states that an object can be computed or that computation is hard. The proposed continuum market is a plausible extension and has a polynomial-size welfare LP under explicit valuation tables, but that computational problem is not anchored in a computational result of this paper. The paper therefore fails bit (a).
fails bit a — no named computational result to mirror
The proposed market concerns equilibrium existence for a society of valuation types; its welfare LP does not capture Theorem 2’s core nonemptiness characterization, Theorem 1’s sufficient condition, or Corollary 1’s two-agent result as computational claims.
The paper has no named computational-complexity result: it proves equilibrium-existence statements, but no theorem that a computational problem is in P, NP-hard, FPT, or similar. So I can’t honestly supply a qualifying anchor under your rule.
The closest results are Theorem 1, proved here, which says that if the core of the associated goods game \(\bar V\) is nonempty, the economy has a competitive equilibrium; and Theorem 2, also proved here, which says that for a convex economy, equilibrium exists if and only if that core is nonempty. Here \(\bar V(N)=MI\), the maximum value of an indivisible allocation, and \(\bar V(T)=\max_j V_j(T)\) for \(T\subsetneq N\). Corollary 1, proved here, says that an economy with two agents and convex reservation values has a competitive equilibrium. These are existence results, not complexity results. The paper’s cited divisible-allocation characterization is due to Bikhchandani and Mamer; it is also not a computational result proved here.
Still, there is a concrete positive-side mirror worth proposing, based on Theorem 2. Call it Continuous Competitive Equilibrium Existence. Let the goods be \(q\) categories, with a supply \(b_i\) of each category per unit of population. The society has a finite support \(\Theta\) of \(\tau\) types, with rational masses \(\mu_\theta>0\) summing to \(1\). A type is a complete reservation-value function \(v_\theta:2^{[q]}\to\mathbb Q_{\ge0}\), including all the preferences relevant to demand; for the closest match to Theorem 2, require each \(v_\theta\) to be monotone and convex in the paper’s sense. Each individual still receives an integral bundle \(S\subseteq[q]\); \(y_{\theta,S}\) is the mass of type \(\theta\) receiving that bundle.
The question is whether there are prices \(p_i\ge0\) and a feasible assignment \(y\) such that
\[
\sum_S y_{\theta,S}=\mu_\theta,\qquad
\sum_{\theta,S:\,i\in S} y_{\theta,S}\le b_i,
\]
and every bundle receiving positive mass maximizes \(v_\theta(S)-\sum_{i\in S}p_i\) for its type, with any good in positive-price supply fully used. A solution is such a price vector and mass assignment. The welfare-maximizing assignment is the linear program
\[
\max_y\ \sum_{\theta,S} y_{\theta,S}v_\theta(S)
\]
subject to those assignment and supply constraints. With valuations listed explicitly for all bundles, the LP and its dual have size polynomial in the input table, and primal-dual optimality supplies an equilibrium. Thus I expect this proposed mirror to be Class A in that explicit representation. Succinct valuation encodings would raise a separate demand or pricing question.
A plausible regime is a large market for standardized indivisible permits or service slots: many buyers share a relatively small number of valuation schedules, and supplies scale with the population. Mass means the fraction of each type assigned each integral bundle. That keeps the paper’s central object—prices supporting agents’ bundle demands—while making the population the continuous object. The article’s Appendix, which divides bundle portions among finite agents, offers a useful mathematical bridge, but that alone is outcome-space continuity and does not count as the population mirror.
My weakest point is also decisive for the requested grading: this is a new computational question inspired by the paper’s equilibrium-existence results, not a mirror anchored in a named computational result of the paper. It also changes the finite economy by scaling category supplies and allowing a type’s mass to split across bundles. That makes it a credible large-market analogue of their equilibrium question, but an author or referee could reasonably say it is a different limit model.
There is no computational result in this paper to mirror. Theorem 1 gives a sufficient equilibrium existence condition; Theorem 2 characterizes existence in convex economies; Corollary 1 gives a two-agent existence result. None states an algorithm or a complexity bound. That is the strongest objection: a computational problem can be built from these results, but it would be a new question inspired by them, not a continuous mirror of a computational result.
The proponent’s Theorem 2 proposal is the most plausible such question. But with \(\tau\) types, \(q\) good categories and valuations listed for every bundle, its variables \(y_{\theta,S}\) form a welfare-maximization LP of size polynomial in the valuation table. Its dual supplies prices, and an optimal assignment gives the type masses receiving each integral bundle. The same formulation works without the paper’s convexity assumption, so the feature Theorem 2 characterizes—the relation between convexity, the core condition and equilibrium existence—does no work in the proposed algorithm. Theorem 1 and Corollary 1 fare no better: they too concern existence, and the mass-assignment LP bypasses their conditions.
There is also a more fundamental problem with this candidate as a mirror. Because a type’s mass can be split across bundles, aggregate assignments are fractional even though each recipient gets an integral bundle. That convexifies the assignment problem: the discrete market’s equilibrium obstruction is replaced by the primal-dual structure of a generic LP. The paper’s Appendix already uses divisible assignments as a comparator; the proposed population model makes that comparator the market itself. This is a continuum-specific degeneration of the phenomenon under study, not a complexity landscape extracted from the paper.
One could choose compact valuation encodings or demand-oracle access to make computation harder. But then the challenge depends on an added representation model, and the paper gives no computational result that would anchor that choice. A large market for standardized permits could make the population regime sensible; it does not solve the anchoring problem. So the negative case is strongest against this paper’s value to ChoCo, though it cannot rule out a worthwhile follow-up that deliberately adds a computational model.
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.