Egyptian fractions: counting, extremal denominators, and the limits of finite checks
Working notebook synthesis, not a publication-reviewed article. Quotes reproduce recorded AI reviews; their status and scope are not independent verification.
An Egyptian-fraction representation writes a rational number as a sum of reciprocals of positive integers. This notebook asks several different questions about those representations. The useful result is a map of what each calculation actually checks, which assumptions it needs, and which earlier claims were corrected.
Evidence and reproduction status
This guide summarizes existing public notebook material. It establishes no new original result. Source statements, reported finite checks and in-campaign reviews have different roles; AI agreement is not mathematical validation. The underlying computational checks have not been rerun for this guide. Private artifacts are not exposed; a recorded evidence reference is not a public reproduction package.
Start with the objective and the domain
Counting representations, maximizing the largest denominator, maximizing the full product or a prefix product, and maximizing the least common multiple are different problems. Ordered denominators need not be distinct. Integer and real domains also differ. A witness for one objective cannot silently establish an extremum for another.
Recorded AI review: Egyptian largest denominator
The stated approximately 1.1e27 largest denominator for k=9 is contradicted by an explicit nine-term distinct unit-fraction representation summing exactly to 1 whose final denominator is 12864938683278671740537145998360961546653259485195806. This verifies the numeric error without certifying the separate general maximal-denominator theorem.
Open exact review in notebook · review:egyptian-largest-denominator
Recorded AI review: Egyptian lcm objective
Unit-fraction tuples(2,3,10,15) and(2,3,7,46,483) have lcms30 and966, respectively. They invalidate identifying the lcm objective with denominator values24/924; witness arithmetic alone does not establish lcm maximality.
Open exact review in notebook · review:egyptian-lcm-objective-1945
Recorded AI review: Egyptian cp base dependency
CP distinctness is needed by the m3/m4 bases; sortedness alone is insufficient. No E1 counterexample or novelty claim.
Open exact review in notebook · review:egyptian-CP-base-dependency-0230
Recorded AI review: Simmons hypothesis counterexamples
Without final sorting,(3,3/2) sums reciprocals to1 but product9/2 exceeds4. Without integer prefix,(2,5/2,10) sums reciprocals to1 but product50 exceeds36. These verify necessity of hypotheses, not the proposed ordering-invariant proof. New source-gap allegations and reconstruction remain under independent review.
Open exact review in notebook · review:simmons-hypothesis-counterexamples-0630
A failed step is not a failed final bound
The notebook records a distinct-denominator counterexample to an intermediate inequality in a survey sketch. Later notes retract the stronger allegation that the survey bound itself is wrong: they describe another route using a remainder bound and a uniform counting theorem. Soundararajan’s statement permits repeated denominators. Matching that domain and the counting theorem’s quantifiers matters more than comparing two printed constants alone.
Source context: Approximating 1 from below using n Egyptian fractions — https://arxiv.org/pdf/math/0502247v1; The number of solutions of the Erdős-Straus Equation and sums of k unit fractions (arXiv v1 PDF) — https://arxiv.org/pdf/1805.02945v1; Egyptian fractions (NAW 5/23 nr 4, Dec 2022, p.237) — https://www.nieuwarchief.nl/serie5/pdf/naw5-2022-23-4-237.pdf.
Recorded AI review: Egyptian survey step exact
Distinct denominators(2,3,11,14,231) sum to1. Their remainder inverses y3=66/5 and y4=231 violate y4<=y3(y3+1)=4686/25. The survey sketch step is false even with distinct denominators; its final asymptotic bound can still follow by another argument.
Open exact review in notebook · review:egyptian-survey-step-exact-1125
Recorded AI review: Egyptian error direction retracted
The literal ElPl Cor3 and survey exponents differ after converting bases1.59791 and1.26408, but a new prefix-count argument suggests the stronger survey bound follows from ElPl Thm2. Retract the earlier assertion that the survey is mathematically wrong. Independent proof and original2018 lineage review are running.
Open exact review in notebook · review:egyptian-error-direction-retracted-1050
Recorded AI review: Egyptian prefix closure
Soundararajan2005 primary explicitly allows b1<=...<=bk, proving the missing remainder bound with repeated denominators. Together with the two reviewed prefix arguments and ElPl Thm2 uniformity in m,n, this supports the stronger survey asymptotic bound. Preserve the retraction of the earlier survey-wrong claim and distinguish the false intermediate sketch step.
Open exact review in notebook · review:egyptian-prefix-closure-1125
What the finite calculations buy us
The recorded five-term enumeration and exact rational checks are useful base cases and tests of proposed bounds. They can expose an incorrect value or an omitted assumption. They do not extend themselves to every length. The sampler note likewise demonstrates bias in a particular retry procedure; it is not a validation of large-scale enumeration performance.
Recorded AI review: Egyptian e1 bases executed
14 four-term and147 five-term unit partitions exhaustively enumerated; E1 prefixes uniquely(2,3,7)and(2,3,7,43). Not an all-length proof.
Open exact review in notebook · review:Egyptian-E1-bases-executed-0310
Recorded AI review: Egyptian k5 prefix max correction
Full147length5partitions give maxQ3=maxE=128,attained(2,8,8,8,8); tentative125suggestionrefuted. Fullrows/topthreeproducts saved; noalllengthpromotion.
Open exact review in notebook · review:Egyptian-k5-prefix-max-correction-0330
Recorded AI review: Egyptian k5 third product
All147 sorted five-term unit partitions enumerated with rigorous rational branch bounds; unique third product1190700 at(2,3,7,45,630), second1707552 at(2,3,7,44,924). No all-length inference.
Open exact review in notebook · review:egyptian-k5-third-product-0100
Recorded AI review: Egyptian sampler
Reviewed reference sampler executed at k5: F=72,33 heads, exact mixture variance373990944/6125. Retrying dead branches changes the expected estimate to124416/689 (about180.575), showing bias versus72. No k9 enumeration or performance claim validated.
Open exact review in notebook · review:egyptian-sampler-20261108T0542Z
Why prefix products connect to geometry
The cited Bäuerle work uses unit-fraction partitions in bounds for lattice simplices. That makes prefix optimization relevant beyond a list of fractions. The notebook identifies an existing prefix-product formula and an earlier perturbation argument; it explicitly retracts novelty for those ingredients. These are context and prior-art connections, not a new implication from the campaign to a geometric theorem.
Source context: Sharp volume and multiplicity bounds for Fano simplices — https://arxiv.org/pdf/2308.12719v1; Largest integral simplices with one interior integral point: Solution of Hensley's conjecture and related results — https://arxiv.org/pdf/1309.7967v3.
Recorded AI review: Egyptian prefix bound prior art
Baeuerle2308.12719 Proposition3.10(iii),realPSdomain ofDefinition3.4, already gives length5prefix3maximum128 and general prefix(k-2)formula. No novelty for this bound; second/third-largest product classifications remain separate candidates. Journalversion and fullproof notchecked.
Open exact review in notebook · review:egyptian-prefix-bound-prior-art-0425
Recorded AI review: Akn repair prior art v3
AKN arXiv1309.7967v3 pages18-19 already contains the PS(k) perturbation/second-derivative step. Retract novelty attribution for the campaign re-derivation. Version metadata2014-12-15; typeset front-page date2018-06-18; journal text remains unopened. Full theorem not reverified.
Open exact review in notebook · review:akn-repair-prior-art-v3-0700
What remains conditional
Positive-polynomial certificates support a reduction only under its stated published bounds, base cases and domain. A repaired induction closes a particular arithmetic gap; it does not certify the whole optimizer classification. The later perturbation note targets a reported sufficient bound on a feasible set, not the headline theorem. The Simmons note downgrades an error allegation to omitted or delegated exposition. The global runner-up classification, external verification and novelty remain unpromoted here.
Recorded AI review: Egyptian w polynomial certificates
Five shifted polynomials have strictly positive constant and nonnegative remaining coefficients; four rational polynomial identities and four E1 examples passed exactly. These support the new W/C-star reduction for all permitted a, conditional on published LS/prefix bounds and bases. General source-based proof is under final independent lane review; no novelty or external validation claim.
Open exact review in notebook · review:egyptian-W-polynomial-certificates-0525
Recorded AI review: Egyptian claimv induction repaired
The final referee replaces the unsupported growth sentence by an induction. Coordinator independently checked the r>=5 squaring bound using e<3 and positive polynomial certificates, plus the small-r bases and c+r>=4 domain. This closes that arithmetic gap only; structural optimizer reduction and full published proof dependencies remain separately qualified.
Open exact review in notebook · review:egyptian-ClaimV-induction-repaired-0525
Recorded AI review: Baeuerle perturbation bound counterexample
For feasible y=(3/20 five times,1/10,3/40,3/40),epsilon119/10000 meets the sign-corrected reported bounds yet produces y7=419/4000>y6=1/10. A smaller corrected epsilon preserves tested constraints. The point is not an optimizer: this refutes that bound sufficiency on the feasible set, not the headline theorem or necessarily an optimality-conditioned proof step. Source-context closure is pending.
Open exact review in notebook · review:baeuerle-perturbation-bound-counterexample-0700
Recorded AI review: Simmons gap label downgrade
Downgrade prior real-gap label to omitted/delegated verification in1941 exposition; the b=c=1 reconstruction passed a second same-family hand review.1932 source still unopened. This is not an established mathematical error or independent human/formal proof.
Open exact review in notebook · review:simmons-gap-label-downgrade-0700
Recorded AI review: Egyptian s simmons specialization
LemmaS is Simmons1941 Theorem1 specialized at auxiliary a1=E; supersedes any novelty/absence label for that lemma. Full runner-up prior art remains unresolved.
Open exact review in notebook · review:egyptian-S-simmons-specialization-0100
Why the evidence matters
These directed relationships restate scoped AI review claims. Each link shows its current evidence state; an absent link does not establish that no relationship exists.
- contradicts: The explicit distinct-denominator tuple rejects this intermediate inequality. The later repaired route can still support the final bound.
- depends on: Positive polynomial certificates do not supply their own source hypotheses or base cases. This dependency keeps those obligations visible.