| paper | d4-journals-jet-2025-0075 |
| authors | — |
| venue | ? |
| filed under | unclassified |
| judged by | gpt-6-luna / xhigh (triple__luna__xhigh__c6r1) |
| judge confidence | high |
| authors would recognise it | yes |
P→P discrete result → its continuous mirror
Answer: Polynomial-time solvable, with exact rational output, using at most \(m\) linear programs. · discrete side: polynomial · adjudication: proved
full proof & adversarial review →
Corollary 2
statement extracted from the paper’s text layer
Given \(m\) charities and a finite support of types \(t\), each with mass \(\mu_t\), nonempty approval set \(A_t\), and per-person contribution \(b_t\), compute the unique per-capita EDR distribution \(\delta\). Find allocations \(x_{t,a}\ge 0\), zero for \(a\notin A_t\), such that \(\sum_{a\in A_t}x_{t,a}=\mu_t b_t\) and \(\delta_a=\sum_t x_{t,a}\); among feasible allocations, \(\delta\) lexicographically maximizes its charity amounts sorted in increasing order. Equivalently, it maximizes \(\sum_t\mu_t b_t\log(\min_{a\in A_t}\delta_a)\). This asks for the high-multiplicity EDR outcome given the paper’s finite-agent characterization, not for equilibrium in a nonatomic best-response game.
Use complete binary-Leontief types \(t=(A_t,b_t)\) with rational masses \(\mu_t\) summing to \(1\). The variables \(x_{t,a}\) allocate each type’s mass-weighted supply \(\mu_t b_t\) among approved charities; their sums give per-capita charity amounts \(\delta_a\). The objective is leximin over \((\delta_a)_{a\in A}\), equivalently type-weighted Nash welfare. Aggregation and splitting among identical donors preserve the feasible set and outcome, so the paper’s at-most-\(m\)-LP algorithm yields polynomial time in \(m\), support size \(\tau\), and rational input bit length \(L\).
The mirror covers computation of binary-weight EDR, including Corollary 2 and its leximin characterization in Proposition 5. It leaves general Leontief weights, strategic properties, monotonicity results, and spending dynamics aside.
My strongest mirror is a narrow computational one: compute the paper’s equilibrium distribution when a very large donor population is described by a small number of complete types. The paper’s divisible charities already make the *outcome* continuous; the mirror makes the *population* continuous too.
The best regime is a large donation or municipal funding program with a fixed menu of contribution levels and approval portfolios. Millions of participants might each contribute one of a few standard amounts and approve one of a small number of charity sets. A type \(t\) is the pair \((A_t,b_t)\): the donor’s approved charities and contribution per unit of population mass. Let \(\mu_t\) be the fraction of the population of that type, with \(\sum_t\mu_t=1\). Then \(n\) can be enormous while the number \(\tau\) of types with positive mass remains modest. Grouping donors this way preserves the paper’s notion of indistinguishability: type members have the same preferences and contribution.
Lead problem — Continuous Binary-Leontief EDR. Given \(m\) charities and the supported types \((\mu_t,A_t,b_t)\), find the equilibrium charity distribution \(\delta\), measured per unit of population. A solution includes a witness allocation \(x_{t,a}\): \(x_{t,a}\ge 0\), it is zero when \(a\notin A_t\), each type allocates exactly its mass-weighted contribution, \(\sum_{a\in A_t}x_{t,a}=\mu_t b_t\), and each charity receives \(\delta_a=\sum_t x_{t,a}\). Among all such decomposable allocations, the required \(\delta\) is the one whose charity amounts, sorted in increasing order, are lexicographically maximal. The distribution is unique, though its witness decomposition need not be. Equivalently, it maximizes the type-weighted Nash welfare \(\sum_t \mu_t b_t\log(\min_{a\in A_t}\delta_a)\).
The anchor is Corollary 2, proved in this paper: with binary weights, the equilibrium distribution can be computed by solving at most \(m\) linear programs. Its basis is Proposition 5, also proved here: for binary weights, the equilibrium distribution is exactly the leximin-maximal charity allocation among decomposable distributions. The continuous problem above is the type-aggregated form of that underlying allocation problem, not a claim about the LP encoding itself. Aggregating individual budgets and allocations by type gives the constraints just stated; conversely, an aggregate allocation can be split across identical donors. Thus the corollary’s LP approach applies with \(\tau\) type rows rather than \(n\) individual rows. With rational input, this gives a polynomial-time algorithm in \(m,\tau\), and the input bit length \(L\), using standard polynomial-time LP algorithms. The paper proves the stated LP bound; the support-sensitive running time is the direct high-multiplicity formulation.
The direction I expect is tractable. The paper’s structural characterization survives aggregation, and the algorithm depends on the charity count and supported types, not on the number of donors represented by those types. This is a plausible question for the authors: their agents already have divisible budgets, and the mirror retains their binary Leontief preferences, budget contributions, feasible allocations, and equilibrium distribution. The paper’s other results—especially its strategic claims—are outside this mirror’s scope; I am claiming only the computation of the binary-weight EDR allocation.
No named hardness result in the paper competes with this algorithmic anchor. That makes Corollary 2 the natural lead, though it also limits what this case establishes: it is a tractability result, not a map of the paper’s complexity landscape. A useful next question is whether exact computation remains polynomial in \(m,\tau,L\) for rational, nonbinary Leontief weights. Another is whether rounded type masses give controlled approximation guarantees for the allocation.
The weakest point is that the paper already treats money and allocations as divisible, and compressing repeated identical donors is a fairly direct aggregation. The added value is therefore modest: the mirror asks for computation from a succinct population description and fractions, rather than an explicit list of donors. I would present this as a credible, focused high-multiplicity version of their allocation problem, not as a broad claim that the paper opens a new complexity landscape.
The strongest case against is that this mirror may be too immediate to justify much programme time. In the binary case, the paper already reduces the problem to leximin allocation subject to linear decomposability constraints. If a type \(t\) has approval set \(A_t\), mass \(\mu_t\), and per-person contribution \(b_t\), its row in that program simply has supply \(\mu_t b_t\). The type allocation \(x_{t,a}\) can be split among identical donors, and individual allocations can be summed back into type allocations. So the proposed algorithm is the paper’s existing LP with repeated rows combined; it needs no new continuous-optimization machinery.
That objection reaches all three anchors. Corollary 2 already supplies the LP algorithm, with type aggregation giving a runtime polynomial in \(m\), \(\tau\), and the rational input length. Proposition 5 already identifies the objective and feasible set after aggregation. And the Nash-welfare characterization in Theorem 1 aggregates exactly as well: identical types contribute the combined weight \(\mu_t b_t\). None seems to leave a substantive obstacle for the proposed mirror.
The best version of the negative case, then, is about marginal value: the binary mirror appears to restate a result whose structure and algorithm are already in the paper, rather than opening a difficult new computational frontier. But that case is weak under ChoCo’s standard. The continuous population is still the object being modeled, and the support-sensitive input is a legitimate high-multiplicity regime. A straightforward or already-solved answer does not make its question unworthy; existing high-multiplicity work is support for the regime, not a collision. Nor does the fact that the charity allocation is divisible mean the population has already been continuized.
I cannot honestly defeat this anchor. A large donation program with repeated approval portfolios and contribution levels is a plausible setting, and the type aggregation preserves the problem exactly. The negative case can question how much new algorithmic insight this particular mirror would produce, but it cannot establish that no worthwhile continuous mirror exists.
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.