pith. sign in

arxiv: 2605.30220 · v1 · pith:UZPRLELUnew · submitted 2026-05-28 · 💻 cs.LG

TriSearch: Learning to Optimize Triangulations via Bistellar Flips

Yiran Wang , Guido Mont\'ufar This is my paper

Pith reviewed 2026-06-29 08:50 UTC · model grok-4.3

classification 💻 cs.LG
keywords triangulation optimizationbistellar flipsreinforcement learningpolytopesreflexive polytopesCalabi-Yau threefoldsflip graph traversalmachine learning for geometry
0
0 comments X

The pith

TriSearch trains a reinforcement learning policy to select bistellar flips that optimize triangulations of polytopes, using a circuit-supported action representation that generalizes from small instances to much larger search spaces.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents TriSearch as a reinforcement learning method that improves triangulations of polytopes by repeatedly choosing bistellar flips. Its core technical move is to represent each feasible flip through its supporting circuit and the local subtriangulation it produces, so that a policy can score actions from local geometric and combinatorial data alone. This design removes the need to enumerate the full triangulation space and creates an interface that works across dimensions. When trained on small polytopes the resulting policy transfers directly to larger ones whose flip graphs are exponentially bigger, yielding stronger metric scores in three dimensions and more distinct fine regular star triangulations of reflexive four-polytopes than existing sampling procedures under the same budget.

Core claim

TriSearch is a reinforcement learning framework for optimizing objectives over triangulations of a polytope via bistellar flips. The key idea is a circuit-supported subtriangulation action representation: feasible flips are encoded by their supporting circuit and realized local subtriangulation, enabling a learned policy to rank them using local geometric and combinatorial features. This yields a dimension-agnostic interface and enables efficient traversal of the flip graph without explicit enumeration of the full triangulation space. Instantiated in 3D and 4D, TriSearch generalizes zero-shot from small training instances to larger polytopes with exponentially larger search spaces.

What carries the argument

The circuit-supported subtriangulation action representation, which encodes each feasible bistellar flip by its supporting circuit together with the local subtriangulation it realizes so that a policy can rank flips from local features.

If this is right

  • The method achieves top performance on metric objectives for triangulations in three dimensions.
  • In four dimensions it discovers more distinct Fine, Regular, Star triangulations of reflexive polytopes than existing samplers under a fixed budget.
  • Training on small instances produces a policy that applies directly to polytopes whose triangulation spaces are exponentially larger.
  • The same action representation supplies a dimension-agnostic interface for traversing flip graphs of polytopes.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same local-feature policy could be tested on polytopes whose triangulations correspond to Calabi-Yau threefolds with prescribed Hodge numbers or other topological invariants.
  • If local geometric features suffice for ranking flips, similar representations might transfer to other combinatorial objects whose connectivity is described by circuits, such as matroid bases or oriented matroids.
  • The observed zero-shot scaling suggests that the learned ranking function captures regularities in the flip graph that are independent of overall dimension and volume.
  • An immediate next measurement would be the fraction of all possible fine regular star triangulations recovered by the policy as polytope size increases.

Load-bearing premise

The circuit-supported subtriangulation action representation allows a learned policy to effectively rank feasible flips using only local geometric and combinatorial features, enabling generalization without full enumeration of the triangulation space.

What would settle it

If a policy trained on small polytopes is applied to larger polytopes and produces no improvement in metric objectives or fewer distinct fine regular star triangulations than existing samplers under the same computational budget, the generalization claim is falsified.

Figures

Figures reproduced from arXiv: 2605.30220 by Guido Mont\'ufar, Yiran Wang.

Figure 1
Figure 1. Figure 1: An illustration of bistellar flips in 3D. The optimal triangulation problem can be viewed as search over G(P). Exact enumeration algorithms are reliable on small instances, but they scale poorly with the size of the flip graph. We formulate the search over the tri￾angulation graph as a Markov Decision Process (MDP) M = {T (P), F(·), Rf , H, γ}. The state space T (P) is the vertex set of G(P), and the actio… view at source ↗
Figure 2
Figure 2. Figure 2: Architecture and training loop of TriSearch. The shared encoder [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Triangulation optimization under a 500-flip budget. It shows the mean relative gap and standard deviation over evaluation polytopes. The black dashed line is exact optimality in 3D, and the gold dashed line is 109 TOPCOM enumeration used as an external 4D baseline. reaches an average gap of 0.16% in 3D and 1.43% in 4D, against 8.97% and 9.09% for the next-best baseline NLS, where the detailed aggregated re… view at source ↗
Figure 4
Figure 4. Figure 4: CY threefold sampling experiments. (a) Success rate of locating a nearby FRST within a [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
read the original abstract

We introduce TriSearch, a reinforcement learning framework for optimizing objectives over triangulations of a polytope via bistellar flips. The key idea is a circuit-supported subtriangulation action representation: feasible flips are encoded by their supporting circuit and realized local subtriangulation, enabling a learned policy to rank them using local geometric and combinatorial features. This yields a dimension-agnostic interface and enables efficient traversal of the flip graph without explicit enumeration of the full triangulation space. Instantiated in 3D and 4D, TriSearch generalizes zero-shot from small training instances to larger polytopes with exponentially larger search spaces. It achieves top performance on metric objectives in 3D and, in 4D, discovers more distinct Fine, Regular, Star triangulations of reflexive polytopes, corresponding to Calabi-Yau threefolds, than existing samplers under a fixed budget.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces TriSearch, a reinforcement learning framework for optimizing objectives over triangulations of a polytope via bistellar flips. The key technical contribution is a circuit-supported subtriangulation action representation that encodes feasible flips by their supporting circuit and realized local subtriangulation, allowing a policy to rank actions using only local geometric and combinatorial features. This yields a dimension-agnostic interface. The method is instantiated in 3D and 4D, with claims of zero-shot generalization from small training polytopes to larger ones (with exponentially larger search spaces), top performance on metric objectives in 3D, and discovery of more distinct Fine/Regular/Star triangulations of reflexive polytopes (corresponding to Calabi-Yau threefolds) than existing samplers under fixed budget in 4D.

Significance. If the empirical claims on zero-shot generalization and outperformance hold with proper verification, the work would be significant for computational algebraic geometry: it offers a scalable, learned alternative to exhaustive or random sampling of triangulation flip graphs, which grow factorially with dimension and number of points. The local-feature inductive bias could accelerate enumeration tasks relevant to string theory and mirror symmetry.

major comments (3)
  1. [Abstract and §4] Abstract and §4 (Experiments): performance and generalization claims are stated without accompanying quantitative results, error bars, training hyperparameters, instance sizes, or statistical tests in the provided text; the central empirical claims cannot be assessed without these details in the experiments section.
  2. [§3.2] §3.2 (Action Representation): the claim that the circuit-supported subtriangulation encoding enables effective ranking 'using only local geometric and combinatorial features' is load-bearing for the zero-shot generalization result; the manuscript must demonstrate (via ablation or feature analysis) that no global polytope information leaks into the local state representation.
  3. [§4.3] §4.3 (4D Results): the statement that TriSearch 'discovers more distinct Fine, Regular, Star triangulations ... than existing samplers under a fixed budget' requires explicit reporting of the budget (number of flips or wall-clock time), the exact baselines, and the number of distinct triangulations found per method with variance across runs.
minor comments (2)
  1. Define all acronyms (e.g., FRS triangulations) at first use and ensure consistent notation for circuits and subtriangulations across sections.
  2. Add a table or figure summarizing the polytope sizes (number of points, dimension) used for training versus testing to support the zero-shot claim.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below and will revise the manuscript to strengthen the presentation of experimental results and supporting analyses.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (Experiments): performance and generalization claims are stated without accompanying quantitative results, error bars, training hyperparameters, instance sizes, or statistical tests in the provided text; the central empirical claims cannot be assessed without these details in the experiments section.

    Authors: We agree that the submitted version of the experiments section and abstract did not include sufficient quantitative detail for independent assessment. In the revised manuscript we will expand §4 with tables containing concrete performance metrics (including means and standard deviations across runs), full lists of training hyperparameters, exact polytope instance sizes for training and zero-shot evaluation, and appropriate statistical comparisons. The abstract will be updated to reference these quantitative results. revision: yes

  2. Referee: [§3.2] §3.2 (Action Representation): the claim that the circuit-supported subtriangulation encoding enables effective ranking 'using only local geometric and combinatorial features' is load-bearing for the zero-shot generalization result; the manuscript must demonstrate (via ablation or feature analysis) that no global polytope information leaks into the local state representation.

    Authors: We acknowledge that an explicit demonstration is needed to support the load-bearing claim. The revised manuscript will include a new ablation study that trains the policy on the local representation while systematically masking or adding global polytope features; performance differences will be reported to confirm that the observed zero-shot generalization relies only on the local circuit-supported subtriangulation encoding. revision: yes

  3. Referee: [§4.3] §4.3 (4D Results): the statement that TriSearch 'discovers more distinct Fine, Regular, Star triangulations ... than existing samplers under a fixed budget' requires explicit reporting of the budget (number of flips or wall-clock time), the exact baselines, and the number of distinct triangulations found per method with variance across runs.

    Authors: We will revise §4.3 to report the precise computational budget (both flip count and wall-clock time), name the exact baseline samplers, and tabulate the number of distinct Fine/Regular/Star triangulations discovered by each method together with run-to-run variance or standard deviation. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper presents TriSearch as an empirical RL method whose performance claims (zero-shot generalization, discovery counts) rest on experimental results rather than any derivation chain. No equations, fitted parameters renamed as predictions, or load-bearing self-citations appear in the abstract or described structure; the action representation is introduced as an inductive bias enabling traversal, not as a self-referential definition. The work is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no explicit free parameters, axioms, or invented entities; all details on modeling choices, feature definitions, and reward functions are absent.

pith-pipeline@v0.9.1-grok · 5677 in / 1072 out tokens · 27039 ms · 2026-06-29T08:50:42.678087+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

48 extracted references · 15 canonical work pages · 9 internal anchors

  1. [1]

    Iste, 2007

    Pascal Jean Frey and Paul-Louis George.Mesh generation: application to finite elements. Iste, 2007

    2007

  2. [2]

    Mesh generation and optimal triangulation

    Marshall Bern and David Eppstein. Mesh generation and optimal triangulation. InComputing in Euclidean geometry, pages 47–123. World Scientific, 1995

    1995

  3. [3]

    Delaunay refinement algorithms for triangular mesh generation

    Jonathan Richard Shewchuk. Delaunay refinement algorithms for triangular mesh generation. Computational Geometry, 22(1–3):21–74, 2002

    2002

  4. [4]

    Dual Polyhedra and Mirror Symmetry for Calabi-Yau Hypersurfaces in Toric Varieties

    Victor V Batyrev. Dual polyhedra and mirror symmetry for calabi-yau hypersurfaces in toric varieties.arXiv preprint alg-geom/9310003, 1993

    work page internal anchor Pith review Pith/arXiv arXiv 1993

  5. [5]

    Complete classification of reflexive polyhedra in four dimensions

    Maximilian Kreuzer and Harald Skarke. Complete classification of reflexive polyhedra in four dimensions.arXiv preprint hep-th/0002240, 2000

    work page internal anchor Pith review Pith/arXiv arXiv 2000

  6. [6]

    Tetrahedrizing point sets in three dimensions

    Herbert Edelsbrunner, Franco P Preparata, and Douglas Brent West. Tetrahedrizing point sets in three dimensions. InInternational Symposium on Symbolic and Algebraic Computation, pages 315–331. Springer, 1988

    1988

  7. [7]

    De Loera, and Jürgen Richter-Gebert

    Alexander Below, Jesús A. De Loera, and Jürgen Richter-Gebert. The complexity of finding small triangulations of convex 3-polytopes.Journal of Algorithms, 50(2):134–167, 2004

    2004

  8. [8]

    Minimum-weight triangulation is NP-hard.Journal of the ACM (JACM), 55(2):1–29, 2008

    Wolfgang Mulzer and Günter Rote. Minimum-weight triangulation is NP-hard.Journal of the ACM (JACM), 55(2):1–29, 2008

    2008

  9. [9]

    Cytools: a software package for analyzing calabi-yau manifolds.arXiv preprint arXiv:2211.03823, 2022

    Mehmet Demirtas, Andres Rios-Tascon, and Liam McAllister. Cytools: a software package for analyzing calabi-yau manifolds.arXiv preprint arXiv:2211.03823, 2022

    work page arXiv 2022

  10. [10]

    On counting triangulations in d dimensions.Computational Geometry, 3(6):315–325, 1993

    Tamal Krishna Dey. On counting triangulations in d dimensions.Computational Geometry, 3(6):315–325, 1993

    1993

  11. [11]

    Optimality of the delaunay triangulation in rd

    Vadakkedathu T Rajan. Optimality of the delaunay triangulation in rd. InProceedings of the seventh annual symposium on Computational geometry, pages 357–363, 1991

    1991

  12. [12]

    Topcom: Triangulations of point configurations and oriented matroids

    Jörg Rambau. Topcom: Triangulations of point configurations and oriented matroids. In Mathematical software, pages 330–340. World Scientific, 2002

    2002

  13. [13]

    Machine learning for combinatorial optimization: a methodological tour d’horizon.European Journal of Operational Research, 290(2):405–421, 2021

    Yoshua Bengio, Andrea Lodi, and Antoine Prouvost. Machine learning for combinatorial optimization: a methodological tour d’horizon.European Journal of Operational Research, 290(2):405–421, 2021

    2021

  14. [14]

    Reinforcement learning for combinatorial optimization: A survey.Computers & Operations Research, 134:105400, 2021

    Nina Mazyavkina, Sergey Sviridov, Sergei Ivanov, and Evgeny Burnaev. Reinforcement learning for combinatorial optimization: A survey.Computers & Operations Research, 134:105400, 2021

    2021

  15. [15]

    Combinatorial optimization and reasoning with graph neural networks.Journal of Machine Learning Research, 24(130):1–61, 2023

    Quentin Cappart, Didier Chételat, Elias B Khalil, Andrea Lodi, Christopher Morris, and Petar Veliˇckovi´c. Combinatorial optimization and reasoning with graph neural networks.Journal of Machine Learning Research, 24(130):1–61, 2023

    2023

  16. [16]

    Pointer networks.Advances in neural information processing systems, 28, 2015

    Oriol Vinyals, Meire Fortunato, and Navdeep Jaitly. Pointer networks.Advances in neural information processing systems, 28, 2015

    2015

  17. [17]

    Neural Combinatorial Optimization with Reinforcement Learning

    Irwan Bello, Hieu Pham, Quoc V Le, Mohammad Norouzi, and Samy Bengio. Neural combina- torial optimization with reinforcement learning.arXiv preprint arXiv:1611.09940, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  18. [18]

    Attention, Learn to Solve Routing Problems!

    Wouter Kool, Herke Van Hoof, and Max Welling. Attention, learn to solve routing problems! arXiv preprint arXiv:1803.08475, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  19. [19]

    Reinforcement learning for solving the vehicle routing problem

    Mohammadreza Nazari, Afshin Oroojlooy, Lawrence Snyder, and Martin Takác. Reinforcement learning for solving the vehicle routing problem. InAdvances in Neural Information Processing Systems, volume 31, 2018. 11

    2018

  20. [20]

    POMO: Policy optimization with multiple optima for reinforcement learning

    Yeong-Dae Kwon, Jinho Choo, Byoungjip Kim, Iljoo Yoon, Youngjune Gwon, and Seungjai Min. POMO: Policy optimization with multiple optima for reinforcement learning. InAdvances in Neural Information Processing Systems, volume 33, pages 21188–21198, 2020

    2020

  21. [21]

    On the difficulty of triangulating three-dimensional nonconvex polyhedra.Discrete & Computational Geometry, 7(3):227–253, 1992

    Jim Ruppert and Raimund Seidel. On the difficulty of triangulating three-dimensional nonconvex polyhedra.Discrete & Computational Geometry, 7(3):227–253, 1992

    1992

  22. [22]

    Learning improvement heuristics for solving routing problems.IEEE Transactions on Neural Networks and Learning Systems, 33(9):5057–5069, 2021

    Yaoxin Wu, Wen Song, Zhiguang Cao, Jie Zhang, and Andrew Lim. Learning improvement heuristics for solving routing problems.IEEE Transactions on Neural Networks and Learning Systems, 33(9):5057–5069, 2021

    2021

  23. [23]

    Learning to perform local rewriting for combinatorial optimization

    Xinyun Chen and Yuandong Tian. Learning to perform local rewriting for combinatorial optimization. InAdvances in Neural Information Processing Systems, volume 32, 2019

    2019

  24. [24]

    Barrett, William R

    Thomas D. Barrett, William R. Clements, Jakob N. Foerster, and Alex I. Lvovsky. Exploratory combinatorial optimization with reinforcement learning. InProceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 3243–3250, 2020

    2020

  25. [25]

    Falkner, Daniela Thyssens, Ahmad Bdeir, and Lars Schmidt-Thieme

    Jonas K. Falkner, Daniela Thyssens, Ahmad Bdeir, and Lars Schmidt-Thieme. Learning to control local search for combinatorial optimization. InJoint European Conference on Machine Learning and Knowledge Discovery in Databases (ECML PKDD), pages 361–376. Springer, 2022

    2022

  26. [26]

    Springer Science & Business Media, 2010

    Jesús De Loera, Jörg Rambau, and Francisco Santos.Triangulations: structures for algorithms and applications. Springer Science & Business Media, 2010

    2010

  27. [27]

    A-discriminants

    Israel M Gelfand, Mikhail M Kapranov, and Andrei V Zelevinsky. A-discriminants. In Discriminants, resultants, and multidimensional determinants, pages 271–296. Birkhäuser Boston, Boston, MA, 1994

    1994

  28. [28]

    Transforming triangulations.Discrete mathematics, 3(4):365–372, 1972

    Charles L Lawson. Transforming triangulations.Discrete mathematics, 3(4):365–372, 1972

    1972

  29. [29]

    Udo Pachner. P.L. homeomorphic manifolds are equivalent by elementary shellings.European journal of Combinatorics, 12(2):129–145, 1991

    1991

  30. [30]

    Transforming calabi- yau constructions: Generating new calabi-yau manifolds with transformers.arXiv preprint arXiv:2507.03732, 2025

    Jacky HT Yip, Charles Arnal, François Charton, and Gary Shiu. Transforming calabi- yau constructions: Generating new calabi-yau manifolds with transformers.arXiv preprint arXiv:2507.03732, 2025

    work page arXiv 2025

  31. [31]

    Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges

    Michael M Bronstein, Joan Bruna, Taco Cohen, and Petar Veliˇckovi´c. Geometric deep learning: Grids, groups, graphs, geodesics, and gauges.arXiv preprint arXiv:2104.13478, 2021

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  32. [32]

    E(n) equivariant graph neural networks

    Vıctor Garcia Satorras, Emiel Hoogeboom, and Max Welling. E(n) equivariant graph neural networks. InInternational conference on machine learning, pages 9323–9332. PMLR, 2021

    2021

  33. [33]

    Equiformer: Equivariant graph attention transformer for 3d atomistic graphs.arXiv preprint arXiv:2206.11990, 2022

    Yi-Lun Liao and Tess Smidt. Equiformer: Equivariant graph attention transformer for 3d atomistic graphs.arXiv preprint arXiv:2206.11990, 2022

    work page arXiv 2022

  34. [34]

    Equiformerv2: Improved equivari- ant transformer for scaling to higher-degree representations.arXiv preprint arXiv:2306.12059, 2023

    Yi-Lun Liao, Brandon Wood, Abhishek Das, and Tess Smidt. Equiformerv2: Improved equivari- ant transformer for scaling to higher-degree representations.arXiv preprint arXiv:2306.12059, 2023

    work page arXiv 2023

  35. [35]

    Hypergraph neural networks

    Yifan Feng, Haoxuan You, Zizhao Zhang, Rongrong Ji, and Yue Gao. Hypergraph neural networks. InProceedings of the AAAI conference on artificial intelligence, volume 33, pages 3558–3565, 2019

    2019

  36. [36]

    Simplicial neural networks

    Stefania Ebli, Michaël Defferrard, and Gard Spreemann. Simplicial neural networks. In Topological Data Analysis and Beyond workshop at NeurIPS, 2020

    2020

  37. [37]

    Simplicial complex neural networks.IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(1):561–575, 2023

    Hanrui Wu, Andy Yip, Jinyi Long, Jia Zhang, and Michael K Ng. Simplicial complex neural networks.IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(1):561–575, 2023

    2023

  38. [38]

    Mengen konvexer Körper, die einen gemeinsamen Punkt enthalten.Mathematis- che Annalen, 83(1):113–115, 1921

    Johann Radon. Mengen konvexer Körper, die einen gemeinsamen Punkt enthalten.Mathematis- che Annalen, 83(1):113–115, 1921. 12

    1921

  39. [39]

    Non-connected toric Hilbert schemes.Mathematische Annalen, 332(3):645– 665, 2005

    Francisco Santos. Non-connected toric Hilbert schemes.Mathematische Annalen, 332(3):645– 665, 2005

    2005

  40. [40]

    Go-explore: a new approach for hard-exploration problems.arXiv preprint arXiv:1901.10995, 2019

    Adrien Ecoffet, Joost Huizinga, Joel Lehman, Kenneth O Stanley, and Jeff Clune. Go-explore: a new approach for hard-exploration problems.arXiv preprint arXiv:1901.10995, 2019

    work page arXiv 1901

  41. [41]

    Proximal Policy Optimization Algorithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. Proximal policy optimization algorithms.arXiv preprint arXiv:1707.06347, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  42. [42]

    High-Dimensional Continuous Control Using Generalized Advantage Estimation

    John Schulman, Philipp Moritz, Sergey Levine, Michael Jordan, and Pieter Abbeel. High- dimensional continuous control using generalized advantage estimation.arXiv preprint arXiv:1506.02438, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  43. [43]

    Searching for Activation Functions

    Prajit Ramachandran, Barret Zoph, and Quoc V Le. Searching for activation functions.arXiv preprint arXiv:1710.05941, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  44. [44]

    Adam: A Method for Stochastic Optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization.arXiv preprint arXiv:1412.6980, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  45. [45]

    Bounding the Kreuzer-Skarke landscape.Fortschritte der Physik, 68(11-12):2000086, 2020

    Mehmet Demirtas, Liam McAllister, and Andres Rios-Tascon. Bounding the Kreuzer-Skarke landscape.Fortschritte der Physik, 68(11-12):2000086, 2020

    2020

  46. [46]

    Constructing unrooted phylogenetic trees with reinforcement learning.Studia Universitatis Babe¸ s-Bolyai Informatica, 66(1):37–53, 2021

    Panna Lipták and Attila Kiss. Constructing unrooted phylogenetic trees with reinforcement learning.Studia Universitatis Babe¸ s-Bolyai Informatica, 66(1):37–53, 2021

    2021

  47. [47]

    Rhombus tilings: decomposition and space structure

    Frédéric Chavanon and Eric Rémila. Rhombus tilings: decomposition and space structure. Discrete & Computational Geometry, 35(2):329–358, 2006

    2006

  48. [48]

    Medina-Mardones, Lucas Fagan, Bartłomiej Lewandowski, Angus Gruen, Yang Qiu, Piotr Kucharski, Zhenghan Wang, and Sergei Gukov

    Ali Shehper, Anibal M. Medina-Mardones, Lucas Fagan, Bartłomiej Lewandowski, Angus Gruen, Yang Qiu, Piotr Kucharski, Zhenghan Wang, and Sergei Gukov. What makes math problems hard for reinforcement learning: a case study.arXiv preprint arXiv:2408.15332, 2024. 13 A Radon Partitions and Circuit Flips This appendix records the standard construction behind th...

    work page arXiv 2024