{
  "schemaVersion": "1.6",
  "slug": "subset-sum-support-entropy-tails",
  "title": "Fixed-size subset sums: sharp support, Rényi entropy and tail bounds",
  "shortTitle": "Random subset sums: how concentrated can they be?",
  "url": "https://evidencepress.org/releases/subset-sum-support-entropy-tails/",
  "oneLine": "If n numbers add to zero, at least 2·C(n−2, k−1) of the k-subsets have a nonzero sum, and only (1, −1, 0, …, 0) attains it, proving a 2018 conjecture on Johnson graphs; the paper also finds the exact minimum Rényi entropy of a random subset sum and proves the hypergeometric tail bound from n = 4k.",
  "abstract": "Choose k of n given real numbers uniformly at random, without replacement, and add them. How concentrated can such a fixed-size subset sum be? This candidate answers three classical forms of the question exactly. Zero sums: if the numbers add up to zero and are not all zero, at least 2·C(n−2, k−1) of the k-subsets have a nonzero sum whenever k ≥ 5 and n ≥ 2k + 1, with equality only for (1, −1, 0, …, 0) up to scaling and reordering. This is the smallest support of an eigenfunction in the first nontrivial eigenspace of the Johnson graph, and proves the 2018 conjecture of Mogilnykh, Vorob’ev and Valyuzhenich, who had checked it by computer for n ≤ 600; a formula with all equality cases is given for every n and k, including one small case, (1, 1, −1, −1, 0) at n = 5, k = 2, that earlier classifications missed. Entropy: the minimum Rényi entropy of the sum is found for every order, answering a question of Ai, Kuželka and Pelekis (2026). One outlier is optimal except at half-sampling (n = 2k), where two outliers win above an explicit order α_k (α_2 = 2, about 1.386k for large k), so the universal Bernoulli lower bound holds exactly for orders up to 2. Tails: for two-valued populations the probability that the sum is at least its mean is at least k/n whenever n ≥ 4k, extending the range n ≥ 8k of Ai and Pelekis (2026) with a 2026 lemma of Ai, Karagiannis, Kuželka and Pelekis; so the Aydinian–Blinovsky conjecture would imply the Manickam–Miklós–Singhi conjecture for all n ≥ 4k. This does not prove that conjecture. Exact analogues over Abelian groups and a small-ball inequality are included. Limits: real numbers only for the zero-sum theorem, first Johnson eigenspace only, tail problem below 4k open. AI-assisted, unrefereed candidate; no external peer review, independent reproduction or formal verification is claimed.",
  "datePublished": "2026-11-15",
  "dateModified": "2026-11-15",
  "version": "0.1.0-candidate",
  "doi": "10.5281/zenodo.24241302",
  "doiUrl": "https://doi.org/10.5281/zenodo.24241302",
  "conceptDoi": "10.5281/zenodo.24241301",
  "pdfUrl": "https://github.com/ipitchford/subset-sum-support-entropy-tails/releases/download/v0.1.0-candidate/subset-sum-support-entropy-tails-0.1.0-candidate.pdf",
  "altPdfUrl": null,
  "altPdfLabel": null,
  "zenodoUrl": "https://zenodo.org/records/24241302",
  "repoUrl": "https://github.com/ipitchford/subset-sum-support-entropy-tails",
  "releaseUrl": "https://github.com/ipitchford/subset-sum-support-entropy-tails/releases/tag/v0.1.0-candidate",
  "markdownUrl": "https://evidencepress.org/releases/subset-sum-support-entropy-tails/index.md",
  "bibtexUrl": "https://evidencepress.org/releases/subset-sum-support-entropy-tails/cite.bib",
  "audioUrl": "https://evidencepress.org/assets/audio/subset-sum-support-entropy-tails.mp3",
  "imageUrl": "https://evidencepress.org/assets/og/subset-sum-support-entropy-tails.png?v=a7a92e5e3b",
  "coverArtUrl": "https://evidencepress.org/assets/art/subset-sum-support-entropy-tails.svg?v=4d62231695",
  "media": [
    {
      "type": "audio",
      "url": "https://evidencepress.org/assets/audio/subset-sum-support-entropy-tails.mp3",
      "name": "Plain-English audio briefing",
      "description": "Communication aid; not additional research evidence.",
      "transcriptUrl": "https://evidencepress.org/assets/audio/subset-sum-support-entropy-tails.txt"
    }
  ],
  "authors": [
    "Anonymous"
  ],
  "license": "CC0-1.0",
  "status": "unrefereed-candidate",
  "verification": {
    "peerReviewed": false,
    "independentlyReproduced": false,
    "formallyVerified": false,
    "internallyReplayed": true,
    "detail": "Unrefereed AI-assisted mathematics candidate. Theorems 1.1–1.6 and their lemmas and corollaries have complete written proofs. Internal AI-assisted review in five reviewer roles and an adversarial review by a second AI model, clean replays, one revision round (new introduction, filled proof steps, the tail theorem from n = 4k and a small-ball inequality added), then a re-review, a proof audit and a second-model audit of the new proofs; two wrong sentences, an attribution gap, an over-stated introductory sentence and a misreported source were corrected, and no numbered result changed. Two attributions were checked against the cited sources at publication. External specialist review, independent reproduction and formal verification have not been established."
  },
  "assurance": [
    {
      "dimension": "availability",
      "label": "Availability and archiving",
      "question": "Is the evidence package publicly retrievable from an archive under a persistent identifier?",
      "state": "passed",
      "evidenceUrl": "https://zenodo.org/records/24241302",
      "note": "The versioned research package and paper are publicly archived on GitHub and Zenodo. Provider asset parity was checked against frozen local files; availability does not establish correctness or peer review."
    },
    {
      "dimension": "internalReplay",
      "label": "Internal replay",
      "question": "Does the producer’s own pipeline reproduce the stated result from the archived package?",
      "state": "passed",
      "note": "All five scripts were rerun from fresh copies in normal mode and under python -O; their outputs matched the shipped files (verification.json apart from the recorded Python version and run time). A weakened support check makes verify.py fail under -O. GitHub CI reruns all five scripts and the negative control on Linux."
    },
    {
      "dimension": "independentRerun",
      "label": "Independent rerun",
      "question": "Has someone else run the supplied implementation and obtained the stated result?",
      "state": "not-assessed",
      "note": "No unaffiliated rerun has been established."
    },
    {
      "dimension": "independentReimplementation",
      "label": "Independent reimplementation",
      "question": "Has someone else reached the result from an independent implementation?",
      "state": "not-assessed",
      "note": "No unaffiliated reimplementation; internal reviewers recomputed the Z_2, Z_3 and Z_5 counterexamples and the tail cases with their own code."
    },
    {
      "dimension": "formalVerification",
      "label": "Formal verification",
      "question": "Is a formalised statement machine-checked, and over which trusted base?",
      "state": "not-assessed",
      "note": "No proof-assistant formalisation."
    },
    {
      "dimension": "specialistReview",
      "label": "Specialist review",
      "question": "Has a domain specialist assessed the argument?",
      "state": "not-assessed",
      "note": "AI-assisted internal and adversarial review is not external specialist review."
    },
    {
      "dimension": "editorialPeerReview",
      "label": "Editorial peer review",
      "question": "Has a journal or venue run peer review to a decision?",
      "state": "not-assessed",
      "note": "No external editorial peer review."
    },
    {
      "dimension": "dataEnvironmentReproducibility",
      "label": "Data and environment reproducibility",
      "question": "Are data and computational environment pinned well enough to rebuild?",
      "state": "passed",
      "note": "Standard-library Python (3.10 or later); every decision uses exact integers or fractions and the random sanity runs use fixed seeds, so the outputs are platform-independent. Not a locked container."
    }
  ],
  "provenance": {
    "aiGenerated": true,
    "aiAssisted": true,
    "generatedBy": [
      "AI-assisted mathematical research",
      "AI-assisted internal review (five reviewer roles), adversarial review and audit by a second AI model, re-review, proof audit, clean replays",
      "Claude Code: publication packaging and release"
    ],
    "humanRole": "Publication authority only; no human research contribution inferred.",
    "reviewStatus": "Internal AI-assisted and adversarial review, not external peer review."
  },
  "attributionBoundary": null,
  "problem": null,
  "orderedProductAverage": null,
  "corrections": [],
  "publicCorrections": [],
  "publicCorrectionIndexes": null,
  "pageStructureVersion": "2.0",
  "pageStructureVariant": "canonical-theorem",
  "pageStructureWaivers": [],
  "recordMaturity": "current",
  "metadataProvenance": "Prepared from the final paper (revision of 14 November 2026 with the version and DOI added to the title-page date line, 19 pages), its five scripts and recorded outputs and the frozen release archive. The versioned GitHub and Zenodo research archives were checked against the frozen release assets on 15 November 2026.",
  "grandfatheredAtSchemaVersion": null,
  "keywords": [
    "subset sums",
    "sampling without replacement",
    "Johnson graph",
    "minimum support",
    "eigenfunctions",
    "Rényi entropy",
    "majorization",
    "hypergeometric distribution",
    "Manickam–Miklós–Singhi conjecture",
    "Boolean slice"
  ],
  "keyResults": [
    "Theorem 1.1 (proof): for a nonzero zero-sum vector, at least 2·C(n−2, k−1) of the k-subsets have a nonzero sum when k ≥ 5 and n ≥ 2k + 1, with equality only for (1, −1, 0, …, 0) up to scaling and reordering (Conjecture 1 of Mogilnykh, Vorob’ev and Valyuzhenich, 2018); the complete formula and all minimisers for every n and k, including (1, 1, −1, −1, 0) at (n, k) = (5, 2).",
    "Theorems 1.2 and 1.3 (proofs): the law of a random k-subset sum is majorized by Ber(k/n) away from half-sampling and by U or Q_k at n = 2k; hence the exact minimum Rényi entropy for every order, with the crossover order α_k at half-sampling (α_2 = 2, α_k ≈ (2k − ½) ln 2).",
    "Corollary 1.4 (proof): the universal Bernoulli bound H_α(X) ≥ H_α(Ber(k/n)) holds for all populations if and only if 0 < α ≤ 2.",
    "Theorem 1.5 (proof): P(Hyp(n, i, k) ≥ ik/n) ≥ k/n for n ≥ 4k, with a single outlier the unique minimiser for n > 4k; Corollary 7.2: the Aydinian–Blinovsky conjecture would imply the Manickam–Miklós–Singhi conjecture for n ≥ 4k.",
    "Theorem 1.6 (proof): the exact minimum support of degree-one functions on Boolean slices over every nontrivial Abelian group, and an entropy dichotomy decided by elements of order two."
  ],
  "reviews": [],
  "evidencePackage": "A 19-page paper with a plain introduction (the question, what was known, notation, main results, what is new and what is not, roadmap) and complete written proofs of every result; no step of a proof is computational. Five standard-library Python scripts check the statements exactly on finite ranges, each in seconds: the support formula and minimisers on all zero-sum vectors with entries in {−2, …, 2} for n ≤ 10 and a complete certificate for n ≤ 7; hypergeometric atoms, the parity majorization chain and the entropy equality statements; the integer-mean lemma for n ≤ 300; the tail theorem for k ≤ 40 and 4k ≤ n ≤ 10k (930,660 cases); and the counterexamples over Z_2, Z_3 and Z_5. Clean replay from fresh copies in normal and python -O mode reproduced every output, and a weakened check makes the scripts fail.",
  "openProblems": [],
  "mathObjects": [],
  "relatedWorks": [
    {
      "citation": "Mogilnykh, Vorob’ev and Valyuzhenich (2018), MMS-type problems for Johnson scheme. Siberian Electronic Mathematical Reports. Conjecture 1, proved here (Theorem 1.1); their Theorem 1 reduces the problem to two classes of vectors and their Section 5 compares the candidates by computer for n ≤ 600.",
      "url": "https://doi.org/10.33048/semi.2018.15.137",
      "doi": "10.33048/semi.2018.15.137"
    },
    {
      "citation": "Vorob’ev, Mogilnykh and Valyuzhenich (2018), Minimum supports of eigenfunctions of Johnson graphs. Discrete Mathematics. Minimum supports in every eigenspace for n large in terms of k.",
      "url": "https://doi.org/10.1016/j.disc.2018.04.018",
      "doi": "10.1016/j.disc.2018.04.018"
    },
    {
      "citation": "Sotnikova and Valyuzhenich (2021), Minimum supports of eigenfunctions of graphs: a survey. The Art of Discrete and Applied Mathematics. Survey restating the two-class description (Theorem 5) that misses the (5, 2) case.",
      "url": "https://doi.org/10.26493/2590-9770.1404.61e",
      "doi": "10.26493/2590-9770.1404.61e"
    },
    {
      "citation": "Ai, Kuželka and Pelekis (2026), An entropic analogue of the MMS conjecture. arXiv:2606.30486v1. The entropy problem answered here (Theorem 1.3) and the chain reduction to hypergeometric laws (their Lemma 3).",
      "url": "https://arxiv.org/abs/2606.30486v1",
      "doi": "10.48550/arXiv.2606.30486"
    },
    {
      "citation": "Ai and Pelekis (2026), On lower bounds for hypergeometric tails. Statistics & Probability Letters. The tail bound for n ≥ 8k, extended here to n ≥ 4k (Theorem 1.5).",
      "url": "https://doi.org/10.1016/j.spl.2026.110798",
      "doi": "10.1016/j.spl.2026.110798"
    },
    {
      "citation": "Ai, Karagiannis, Kuželka and Pelekis (2026), On the probability that a discrete IFR random variable exceeds its mean. arXiv:2610.24726v1. The lemma (Theorem 2) and Theorem 3(1) used in the proof of Theorem 1.5.",
      "url": "https://arxiv.org/abs/2610.24726v1",
      "doi": "10.48550/arXiv.2610.24726"
    },
    {
      "citation": "Aydinian and Blinovsky (2012), A remark on the problem of nonnegative k-subset sums. Problems of Information Transmission. The conjecture that two-valued populations are extremal for the MMS problem.",
      "url": "https://doi.org/10.1134/s0032946012040059",
      "doi": "10.1134/s0032946012040059"
    },
    {
      "citation": "Manickam and Singhi (1988), First distribution invariants and EKR theorems. Journal of Combinatorial Theory, Series A. The Manickam–Miklós–Singhi (MMS) conjecture.",
      "url": "https://doi.org/10.1016/0097-3165(88)90077-5",
      "doi": "10.1016/0097-3165(88)90077-5"
    },
    {
      "citation": "Pokrovskiy (2015), A linear bound on the Manickam–Miklós–Singhi conjecture. Journal of Combinatorial Theory, Series A. MMS for n ≥ 10^46·k.",
      "url": "https://doi.org/10.1016/j.jcta.2015.01.007",
      "doi": "10.1016/j.jcta.2015.01.007"
    },
    {
      "citation": "Dumer and Kapralova (2013), Spherically Punctured Biorthogonal Codes. IEEE Transactions on Information Theory. Corollary 10, the binary Krawtchouk extremum behind Lemma 8.2 and the binary case of Theorem 1.6.",
      "url": "https://doi.org/10.1109/tit.2013.2250579",
      "doi": "10.1109/tit.2013.2250579"
    },
    {
      "citation": "Brouwer, Cioabă, Ihringer and McGinnis (2018), The smallest eigenvalues of Hamming graphs, Johnson graphs and other distance-regular graphs with classical parameters. Journal of Combinatorial Theory, Series B. Theorem 1.2 restates and reproves the Krawtchouk extremum.",
      "url": "https://doi.org/10.1016/j.jctb.2018.04.005",
      "doi": "10.1016/j.jctb.2018.04.005"
    },
    {
      "citation": "Amireddy, Behera, Srinivasan and Sudan (2025), A Near-Optimal Polynomial Distance Lemma over Boolean Slices. ICALP 2025, LIPIcs 334. The bound (r − 1)/n for degree-one functions on slices, sharpened to the exact value in Theorem 1.6.",
      "url": "https://doi.org/10.4230/LIPIcs.ICALP.2025.11",
      "doi": "10.4230/LIPIcs.ICALP.2025.11"
    },
    {
      "citation": "Madiman, Wang and Woo (2019), Majorization and Rényi entropy inequalities via Sperner theory. Discrete Mathematics. Majorization and Rényi entropy.",
      "url": "https://doi.org/10.1016/j.disc.2019.03.002",
      "doi": "10.1016/j.disc.2019.03.002"
    },
    {
      "citation": "Mattner and Schulz (2018), On normal approximations to symmetric hypergeometric laws. Transactions of the American Mathematical Society. A different ordering of symmetric hypergeometric laws.",
      "url": "https://doi.org/10.1090/tran/6986",
      "doi": "10.1090/tran/6986"
    },
    {
      "citation": "Levin (2024), A stochastic ordering of multiple hypergeometric laws: Peakedness of category counts about half the population category sizes is symmetric unimodal in the sample size. Sequential Analysis. Peakedness about a fixed centre as the sample size varies; Theorem 4.3 instead compares laws centred at their own moving means.",
      "url": "https://doi.org/10.1080/07474946.2024.2379901",
      "doi": "10.1080/07474946.2024.2379901"
    },
    {
      "citation": "Grosu, Person and Szabó (2014), On the rank of higher inclusion matrices. Journal of the London Mathematical Society. Reports the known incidence-rank results that Corollary 6.3 re-derives.",
      "url": "https://doi.org/10.1112/jlms/jdu029",
      "doi": "10.1112/jlms/jdu029"
    },
    {
      "citation": "Ai, Kuželka and Pelekis (2025), Concentration inequalities for the sum in sampling without replacement: an approach via majorization. arXiv:2503.20473. Adjacent work on majorization in sampling without replacement.",
      "url": "https://arxiv.org/abs/2503.20473",
      "doi": "10.48550/arXiv.2503.20473"
    }
  ],
  "claimAssurance": null,
  "operatingModel": {
    "version": "1.0",
    "workId": "ep-work:subset-sum-support-entropy-tails",
    "attemptIds": [
      "ep-attempt:subset-sum-support-entropy-tails-publication"
    ],
    "aims": [
      "science"
    ],
    "artifactRoles": [
      "research-output",
      "communication"
    ],
    "lineageId": null,
    "accelerationPrimitives": [
      "counterexample-proxy-first",
      "adversarial-controls",
      "assurance-vector",
      "agent-readable-research-object"
    ],
    "decisionObject": {
      "type": "other",
      "description": "Treat the minimum support of the first Johnson eigenspace as 2·C(n−2, k−1) for k ≥ 5, n ≥ 2k + 1 (MVV Conjecture 1), the minimum Rényi entropy of fixed-size subset sums as given by one or two outliers with crossover α_k, and the two-valued hypergeometric tail bound k/n as valid from n = 4k.",
      "scope": "Real populations and the first Johnson eigenspace; the MMS consequence is conditional on the Aydinian–Blinovsky conjecture; the tail problem below 4k remains open."
    },
    "bottleneckTargeted": [
      "assurance",
      "publication"
    ],
    "semanticBridge": {
      "state": "explicit",
      "description": "Section 1 separates what Mogilnykh, Vorob’ev and Valyuzhenich, Ai–Kuželka–Pelekis, Ai–Pelekis, Ai–Karagiannis–Kuželka–Pelekis, Dumer–Kapralova and Amireddy–Behera–Srinivasan–Sudan proved from what is new, in a dedicated list of what is new and what is not; claim_ledger.md gives the prior-work boundary claim by claim.",
      "remainingRisks": [
        "The bounded literature search is not exhaustive priority certification.",
        "Several load-bearing sources are recent arXiv preprints."
      ]
    },
    "humanJudgmentGates": [
      "Publication is authorised; scientific endorsement is not inferred.",
      "External specialist review and independent reproduction remain open."
    ],
    "parentLinks": [],
    "assuranceTarget": {
      "dimensions": [
        "specialistReview",
        "independentRerun",
        "formalVerification",
        "priorityAssessment"
      ],
      "nextAction": "Have a specialist in algebraic combinatorics check Lemma 2.2 and Theorem 4.3, and a probabilist check Section 7, and rerun the scripts independently.",
      "claimCeiling": "Written proofs of the support formula, the Rényi optimisation, the tail bound from n = 4k and the group analogues; internal verification is not external assurance."
    },
    "impactClaims": [
      {
        "id": "subset-sum-support-entropy-tails-no-impact",
        "aim": "science",
        "outcome": "An openly inspectable proof of the Mogilnykh–Vorob’ev–Valyuzhenich conjecture, with exact entropy and tail results for fixed-size subset sums.",
        "setting": "AI-assisted mathematical research",
        "status": "NO_IMPACT_EVIDENCE",
        "designClass": "none",
        "comparator": "None registered.",
        "estimand": "No research-speed or scientific-impact effect is estimated.",
        "evidenceRefs": [],
        "registeredDesignRef": null,
        "independentAssessment": null
      }
    ]
  },
  "researchMetrics": {
    "schemaVersion": "1.0",
    "policyUrl": "https://evidencepress.org/api/research-metrics-policy.json",
    "outcomePolicyUrl": "https://evidencepress.org/api/v2/research-metrics-policy.json",
    "workLedgerUrl": "https://evidencepress.org/api/work-ledger.json",
    "claimCeiling": "These records describe scoped workflow inputs and outcomes. They support calibration and reference-class construction, but do not by themselves establish research acceleration, theorem correctness, novelty, independent assurance or impact.",
    "attempts": [
      {
        "attemptId": "ep-attempt:subset-sum-support-entropy-tails-publication",
        "status": "active",
        "resultClass": "pending",
        "metrics": {
          "schemaVersion": "1.0",
          "measurementScope": "publication-only",
          "scopeBoundary": "Only archive packaging, site integration, guarded delivery and canonical readback from registration are prospective. Earlier research, review, revision and extension work are left-censored.",
          "forecast": {
            "frozenAt": "2026-11-15T07:07:27Z",
            "procedureClass": "full-candidate-publication",
            "targetOutcome": "Accurately scoped GitHub, Zenodo and canonical Evidence Press release with explanatory media and machine readback, if all publication gates pass.",
            "expectedActiveMinutes": 90,
            "plausibleLowMinutes": 60,
            "plausibleHighMinutes": 150,
            "expectedUnattendedWaitMinutes": 30,
            "referenceClass": {
              "label": "Full mathematical candidate publication",
              "sampleSize": 0,
              "basis": "Route benchmark planning prior, not empirical calibration."
            },
            "fermiComponents": [
              {
                "component": "Research archive packaging, replay and identities",
                "count": 1,
                "lowMinutesPerUnit": 25,
                "centralMinutesPerUnit": 35,
                "highMinutesPerUnit": 60,
                "basis": "One frozen package with a rating-free public export, clean replay and DOI insertion."
              },
              {
                "component": "Page, metadata and media integration",
                "count": 1,
                "lowMinutesPerUnit": 20,
                "centralMinutesPerUnit": 30,
                "highMinutesPerUnit": 45,
                "basis": "One reader-first release, drafted in parallel with verification."
              },
              {
                "component": "Guarded deployment and canonical readback",
                "count": 1,
                "lowMinutesPerUnit": 15,
                "centralMinutesPerUnit": 25,
                "highMinutesPerUnit": 45,
                "basis": "Mandatory site tests, one merge cycle with ledger carry-forward, zero-cost deployment and public checks."
              }
            ],
            "probabilityPositiveSignal": 0.9,
            "probabilityTargetClosure": 0.85,
            "probabilityHorizonMinutes": 300,
            "assumptions": [
              "The approved scientific package remains valid and configured providers remain available."
            ],
            "reforecastTriggers": [
              "A publication gate reveals a scientific or integrity defect, or a provider outage blocks delivery."
            ],
            "stopRule": "Do not weaken assurance, integrity or cost gates to meet the forecast."
          },
          "outcome": null
        },
        "missingnessReason": null
      }
    ]
  }
}