Related papers: Robust Reasoning as a Symmetry-Protected Topologic…
Structured light offers a promising solution for the increasing data demands of modern optical networks, opening up new degrees of freedom that can be leveraged for greater channel capacity and more bits per photon. However, its…
Despite strong medical benchmark accuracy, LLMs can exhibit severe multi-turn sycophancy in clinical dialogue, abandoning initial correct diagnosis under escalating pressure. We propose \textbf{\textsc{Med-Stress}}, a targeted stress test…
We introduce exactly solvable gapless quantum systems in $d$ dimensions that support symmetry protected topological (SPT) edge modes. Our construction leads to long-range entangled, critical points or phases that can be interpreted as…
We study different phases of the one-dimensional bond-alternating spin-$1/2$ Heisenberg model by using the symmetry fractionalization mechanism. We employ the infinite matrix-product state representation of the ground state (through the…
Large Reasoning Models (LRMs) have recently demonstrated impressive performances across diverse domains. However, how the safety of Large Language Models (LLMs) benefits from enhanced reasoning capabilities against jailbreak queries remains…
Large Language Models (LLMs) have achieved impressive performance across diverse natural language processing tasks, but their growing power also amplifies potential risks such as jailbreak attacks that circumvent built-in safety mechanisms.…
Symmetry is one of the cornerstones of modern physics and has profound implications in different areas. In symmetry-protected topological systems, symmetries are responsible for protecting surface states, which are at the heart of the…
We construct an exactly soluble Hamiltonian on the D=3 cubic lattice, whose ground state is a topological phase of bosons protected by time reversal symmetry, i.e a symmetry protected topological (SPT) phase. In this model anyonic…
Large Reasoning Models (LRMs) achieve strong performance by generating long chains of thought (CoT), but often overthink, continuing to reason after a solution has already stabilized and thereby wasting tokens and increasing latency.…
Large language models (LLMs) have shown an impressive ability to perform tasks believed to require thought processes. When the model does not document an explicit thought process, it becomes difficult to understand the processes occurring…
Large Language Models (LLMs) have shown impressive performance across various domains, but their ability to perform molecular reasoning remains underexplored. Existing methods mostly rely on general-purpose prompting, which lacks…
Large Language Models (LLMs) have demonstrated impressive reasoning capabilities, especially when guided by explicit chain-of-thought (CoT) reasoning that verbalizes intermediate steps. While CoT improves both interpretability and accuracy,…
Recent years saw the complete classification of topological band structures, revealing an abundance of topological crystalline insulators. Here we theoretically demonstrate the existence of topological materials beyond this framework,…
Explicit Chain-of-Thought improves the reasoning performance of large language models but often incurs high inference cost due to verbose token-level traces. While recent approaches reduce this overhead via concise prompting or step…
The emergence of Large Reasoning Models (LRMs) introduces a new paradigm of explicit reasoning, enabling remarkable advances yet posing unique risks such as reasoning manipulation and information leakage. To mitigate these risks, current…
Searching for topological insulators/superconductors is a central subject in recent condensed matter physics. As a theoretical aspect, various classification methods of symmetry-protected topological phases have been developed, where the…
Topological phases of matter are often understood and predicted with the help of crystal symmetries, although they don't rely on them to exist. In this chapter we review how topological phases have been recently shown to emerge in amorphous…
Transformers have demonstrated remarkable performance in natural language processing and related domains, as they largely focus on sequential, autoregressive next-token prediction tasks. Yet, they struggle in logical reasoning, not…
Understanding how Large Language Models (LLMs) perform logical reasoning internally remains a fundamental challenge. While prior mechanistic studies focus on identifying taskspecific circuits, they leave open the question of what…
We construct a unified operator framework for quantum holonomies generated from bosonic systems. For a system whose Hamiltonian is bilinear in the creation and annihilation operators, we find a holonomy group determined only by a set of…