arXiv subjects

Maksymilian Manko

Publications and source records attributed to Maksymilian Manko.

5 recordsLinked to original sources

arXiv2026

New Snake-in-the-Box Records via Snakepit Surgery and Learned Construction

The snake-in-the-box problem asks for a longest induced path in the hypercube graph $Q_n$. We find a length-191 snake in dimension $n=9$, the lowest dimension where the maximum is unknown, improving the previous record of 190 that had stood for 14 years. We also establish new lower bounds in dimensions 10-13. To find these records, we introduce snakepits, collections of disjoint snakes, to expand the search space and open new routes between snakes. This motivates our new Snakepit-in-the-Box benchmark, which seeks maximal edge counts when allowing multiple components. Finally, we introduce Beam Anchor, a search-supervised learned constructor algorithm that finds 100 inequivalent length-190 snakes in dimension 9.

cs.DM↗

arXiv2026

The Two-Hump Problem: Bridging the Difficulty Gap in Mathematical Reinforcement Learning

Mathematical search problems present a unique challenge for Reinforcement Learning (RL) due to vast search spaces and sparse rewards. In previous works, the Andrews-Curtis (AC) conjecture was established as an illustrative example of such problems. In this work, we identify a critical structural barrier in the AC landscape: a "Two-Hump" distribution, where problem instances are either trivially solvable or effectively impossible, with a scarcity of intermediate "hard-but-solvable" instances required for effective learning. We tackle this challenge through two primary avenues: novel data generation techniques to populate the difficulty gap, and significant algorithmic enhancements including the introduction of supermoves and Transformer-based architectures. We demonstrate substantial performance improvements over previous baselines, and release new comprehensive benchmark datasets including AC-19 (125,192 AC-trivial presentations of varying difficulty with length at most 19) and AC-1M (1,136,154 hard AC-trivial presentations of length at most 30), the first large-scale, publicly available datasets of this kind.

cs.LG↗

arXiv2026

Representation theory of non-factorizable ribbon Hopf algebras

In arXiv:2503.19532 new examples of ribbon Hopf algebras based on the construction due to Nenciu were presented. This piece serves as a sequel where we study the representation theory of these new examples of ribbon Hopf algebras. We classify indecomposable projective and simple modules, find the Krull-Schmidt decomposition of the adjoint representation of Nenciu algebras, and prove fusion rules between its components. We also comment on the properties of Müger centres of their representation categories, in particular that they can be non-semisimple. Finally, we consider a new family of ribbon Hopf algebras over fields of prime characteristic $p>2$ in the context of 4-dimensional TQFTs presented in arXiv:2306.03225 that constitute an improvement over examples given therein, although still seemingly falling short of producing powerful invariants of 4-manifolds.

math.QA↗

arXiv2025

Non-factorizable ribbon Hopf Algebras

Building on the work of Nenciu we provide examples of non-factorizable ribbon Hopf algebras, and introduce a stronger notion of non-factorizability. These algebras are designed to provide invariants of $4$-dimensional $2$-handlebodies up to 2-deformations. We prove that some of the invariants derived from these examples are invariants dependent only on the boundary or on the presentation of the fundamental group of the 2-handlebody.

math.QA↗

arXiv2025

An Upper Bound on the Critical Volume in a Class of Toric Sasaki-Einstein Manifolds

We prove the existence of an upper bound on critical volume of a large class of toric Sasaki-Einstein manifolds with respect to the first Chern class of the resolutions of the Gorenstein singularities in the corresponding toric Calabi-Yau varieties. We examine the canonical metrics obtained by the Delzant construction on these varieties and characterise cases when the bound is attained. We comment on computational tools used in the investigation, in particular Neural Networks and the gradient saliency method.

hep-th↗