gpt 6 astra ultra attempt on CS-746

Research-attempt provenance

Definitions and mathematical statement

Shared notation.

Standard number systems are $\mathbb N,\mathbb Z,\mathbb Q,\mathbb R,\mathbb C$. Specialized notation and index ranges follow the cited source and the definitions below.

A universal compressor does not know the actual source parameter in the stated source class. Its performance may be measured by redundancy relative to a source-dependent optimal code. A large alphabet has a number of symbols that grows with block or sample size.

Record scope and qualifications.

The catalogue title is retained as metadata. Named certificates, contracts, bounds or frameworks in that title are not asserted as results of the original source extract. The cleaned threshold places delta outside the square root, whereas the primary report writes sqrt(delta log n). Both the universality criterion and precise alphabet regime are incomplete; no repaired inequality is substituted.

Source-record limitation: The cleaned block threshold places delta outside the square root, whereas the primary report p. 692 writes sqrt(delta log n); alphabet growth and the universality criterion are also missing. The supplied inequality is preserved with this discrepancy.

Statement recorded in the supplied source.

For block size $k>\delta\sqrt{\log n}$, can one construct a universal compressor for the large or unknown alphabets arising from the submatrix decomposition? [S2]

Source status.

The supplied catalogue labels this source record open; that metadata was checked against the supplied snapshot on 2026-10-11. The label is preserved for provenance and is not an independent verification of present open status or mathematical correctness. The catalogue literature assessment is dated 2026-08-22.

Short English statement

Seek source compression with large or unknown alphabets in the stated block regime.

Sources

• S1 Ulam AI and the UnsolvedMath Contributors, UnsolvedMath: A Curated Collection of Open Mathematics Problems, record 30005054 (OWR-9790365-003). Catalogue curation and metadata: CC BY 4.0. The source extract is retained; definitions, English summaries and qualifications are editorial additions. https://huggingface.co/datasets/ulamai/UnsolvedMath.
• S2 New Mathematical Techniques in Information Theory $2023$, Oberwolfach Reports, DOI 10.4171/owr/2022/14, pp. 683--707 https://doi.org/10.4171/owr/2022/14.
• S3 Bhatt--Wang--Wang--Wang, universal graph compression https://arxiv.org/abs/2006.02643.
• Source-specific Oberwolfach Reports licences, including CC BY-SA 4.0 where applicable, are retained; see THIRD_PARTY_NOTICES.md.

Preparation and review.

The original source and UnsolvedMath contributors retain attribution for the source extract. Codex (GPT-6; exact serving revision not exposed) assisted with definitions, English summaries, scope notes and formatting. No new human mathematical authorship or independently verified solution is asserted. Permanent source ID 30005054 and code OWR-9790365-003 checked against the complete supplied registry and current repository. Distinct upstream records, including repeated questions and broad or damaged extracts, are retained for complete coverage of the source CS category. A repeated record is not a new independent conjecture. Complete mathematical-equivalence and current-literature review has not been performed. Preparation and final review follow contribution prompts C and D, with source gaps disclosed. Structural and rendering checks do not establish a new theorem or open status.

Literature and context check

The report, pp. 691--692, ties its open large-alphabet problem to submatrix blocks of stochastic-block-model adjacency matrices. The attempt below tests a particular generic universal-prediction route. Its adversarial alphabet family is not claimed to be exactly the report's graph distribution class.

Research attempt: A fixed-alphabet predictor pays repeatedly for an enormous alphabet

Let a known alphabet have size $K$ and let a sequential predictor assign probability $(n_a+1)/(t+K)$ to symbol $a$ after $t$ previous symbols, where $n_a$ is its previous count. These probabilities sum to one because $\sum_a n_a=t$. On a constant sequence of length $N$, its assigned probability is \[ Q(a^N)=\prod_{t=0}^{N-1}\frac{t+1}{K+t} =\frac{N!}{K(K+1)\cdots(K+N-1)}. \] Since $N!\le N^N$ and every denominator factor is at least $K$, its ideal log loss obeys $-\log_2 Q(a^N)\ge N\log_2(K/N)$. For $K$ much larger than $N$, this predictor keeps paying a large cost even after repeatedly observing the same symbol. The formula is an exact likelihood calculation; no unproved coding-regret theorem is used.

For example $K=2^{20}$ and $N=16$ give log loss at least 256 bits. When $K$ and $N$ are known, there is a simple binary prefix code for all length-$N$ sequences that encodes every constant sequence in 21 bits: write a leading 0 followed by the 20-bit symbol index for a constant sequence; otherwise write a leading 1 followed by all $N$ symbol indices. All codewords in each branch have the same length, and the two branches have different first bits, so the code is prefix-free and decodes uniquely. This does not make it universal-optimal; it is a concrete witness that the predictor's repeated alphabet cost is avoidable on this family.

There is still an unavoidable first-symbol cost. The $K$ possible constant sequences must receive distinct prefix codewords. At least one has length at least $\log_2 K$: if all had shorter length than an integer $L=\lceil\log_2 K\rceil$ when applicable, their disjoint binary-tree capacities could not accommodate $K$ words; equivalently prefix-freeness gives $\sum2^{-\ell}\le1$ by the same extension counting at the maximum code length. If every $\ell<\log_2K$, the sum would exceed one. Thus a universal code cannot uniformly attain zero length just because each point-mass source has entropy zero.

For a $k\times k$ binary block the unrestricted block alphabet can have $2^{k^2}$ symbols. The preceding predictor calculation explains a specific large-alphabet obstruction, but the report's graph blocks have dependence and model restrictions. Proving universality for that class requires exploiting their structure or controlling an appropriate redundancy; the constant-sequence example does not establish an impossibility result for the original graph compressor.

Verification and adversarial audit

The predictor is normalized at each step, the likelihood product is exact, and the alternative code handles every sequence with an explicit branch flag. The first-symbol lower bound is for some alphabet symbol, not every symbol. No arbitrary iid alphabet lower bound is silently transferred to the stochastic-block-model family.

Audit corrections.

The alphabet obstruction is separated from the graph-model target. Ideal log loss, actual prefix-code length, and unavoidable model-identification cost are not conflated.

External computation.

None; symbolic arguments and boundary checks only.

Scope and remaining gaps

The generic fixed-alphabet predictor route is falsified as an efficient universal choice on a specified family. The source's block-size regime and structured graph universality remain unresolved.

Integration review.

Codex (GPT-6; exact serving revision not exposed) reviewed the written arguments and expanded layout and source-reference wrappers. Repository metadata and the production web renderer were checked using Python 3.13 and JavaScript with MathJax 3.2.2. This model-assisted review is not independent human proof verification; no PDF compilation or formal proof checking is asserted.

Completion Estimate

COMPLETION: 0% The calculations below the preserved source statement investigate a concrete route, but establish no substantiated new progress toward the entire original target. Known special cases and failed sufficient conditions are not credited as a solution.

References

The original source references and catalogue attribution are preserved in the Sources section above. The following URLs were actually consulted for this attempt; retention of an additional original reference does not claim fresh consultation of it.

• https://ems.press/content/serial-article-files/46950 (consulted 2026-10-11).

Model and contributor disclosure

OpenAI gpt-6-astra, requested reasoning configuration ultra, generated this research attempt on 2026-10-11. The exact serving revision is not exposed. Codex GPT-6 performed orchestration and repository-format review; its exact serving revision is also not exposed. The mathematical route, boundary checks and final adversarial self-audit followed contribution prompts A then B in docs/contribution_prompts.md, after reading CONTRIBUTING.md. Python 3.14.2 assembled the standalone documents; any mathematical computation is specified in Verification. No theorem prover was used. No human mathematical authorship, substantial human proof correction, or independent human proof checking is asserted. Model-assisted review, source retrieval, and structural checks do not establish correctness or novelty. Full private reasoning transcripts are unavailable as reproducibility artifacts; the complete arguments, inputs and outcomes actually used are stated here or in the companion batch audit.

Final status

Unresolved as a new research contribution. This is a concrete route investigation, not a claim of new complete resolution or a current-status certification. A known answer reproduced for context does not become a new contribution. The catalogue's source-status label is preserved; this attempt status does not assert that a source-labelled solved question is open.