Recursively fine-tuning language models on their own ground-truth-pruned reasoning traces improves GSM8K Pass@1, but the claimed GPT-4o-beating result rests on a nonstandard 500-question test subset.
Manifestation of edge-bulk incompatibility in fractional quantum Hall platform
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
The edges of a two-dimensional topological phase of matter serve as a platform underlying its low-energy dynamics. The topology of the bulk phase dictates the structure of the gapless modes. Proximitizing boundary modes to another boundary, may lead to gap opening at the edge. Subsequently, one may engineer different segments of the boundary with intrinsic incompatibility of gap-generating mechanisms ("edge-edge incompatibility"), facilitating the generation of topological excitations, e.g. Majorana zero modes (MZMs). Here we address the possibility of bulk-edge incompatibility, whereby the intrinsic bulk gap competes with a gap generated via boundary modes. Specifically, we consider two $\nu = 2/3$ fractional quantum Hall phases whose shared boundary modes are gapped out via disorder-generated tunneling across the boundary. A neutral superconducting phase, made up of neutral edge modes, is stabilized over a broad range of interaction parameters. This phase cannot coexist with the bulk gap. The resulting edge-bulk incompatibility gives rise to the emergence of MZMs (in the neutral sector). We propose an experimental setup to verify both the neutral superconductivity phase and the emergent MZMs.
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Think, Prune, Train, Improve: Scaling Reasoning without Scaling Models
Recursively fine-tuning language models on their own ground-truth-pruned reasoning traces improves GSM8K Pass@1, but the claimed GPT-4o-beating result rests on a nonstandard 500-question test subset.