14.0CLJun 1
PlanarBench: Evaluating LLM Spatial Reasoning via Planar Graph DrawingOleksandr Nikitin
PlanarBench tests whether LLMs can draw planar graphs as ASCII art given only an edge list -- a spatial reasoning task that resists memorization because edge order, edge orientation, and node labels are all permutable. We evaluate 91 models on the 199 simplest non-isomorphic connected planar graphs (2 - 7 vertices). Edge count is the dominant difficulty predictor ($r = -0.85$) -- a finding not reported in prior LLM graph benchmarks, which use only node count as the difficulty axis.
CRJun 26, 2013
Information-Theoretic Security for the MassesOleksandr Nikitin
We combine interactive zero-knowledge protocols and weak physical layer randomness properties to construct a protocol which allows bootstrapping an IT-secure and PF-secure channel from a memorizable shared secret. The protocol also tolerates failures of its components, still preserving most of its security properties, which makes it accessible to regular users.