SearcharxivSearch

arXiv subjects

Wojciech Sas

Publications and source records attributed to Wojciech Sas.

2 recordsLinked to original sources

Correlation-driven origin of shallow electron pocket in Co$_{1/3}$TaS$_2$ revealed by ARPES and cluster perturbation theory

We investigate the electronic structure and Fermi surface of Co$_{1/3}$TaS$_2$ using angle-resolved photoemission spectroscopy (ARPES) combined with theoretical modeling beyond standard density functional theory (DFT+U). A shallow electron pocket, the so-called $β$ feature, is observed at the Fermi level near the corner of the superlattice Brillouin zone, representing the first experimental observation of this feature in an intercalated TaS$_2$ compound. Similar pockets have been reported in $X_{1/3}$NbS$_2$ ($X$ = Co, Cr, Ni), where their surface versus bulk origin remains actively debated. Because conventional DFT+U does not capture this feature, we employ cluster perturbation theory (CPT) to incorporate an explicit treatment of strong electron correlations ($U$) on the Co sites. CPT successfully reproduces the $β$ feature, demonstrating its origin from correlation-driven bulk states rather than surface effects. To further substantiate this conclusion, we studied a reduced Co-content sample, Co$_{0.22}$TaS$_2$, where the reduced charge transfer modifies the Co-derived states near the Fermi level. Its electronic structure remains largely similar to that of pristine 2H-TaS$_2$, showing only a minor overall energy shift and lacking the $β$ feature, consistent with disrupted long-range Co ordering and modified orbital character near the Fermi level. We demonstrate that the $β$ feature arises from strong local correlations on the Co sites and requires long-range crystallographic order among intercalated Co atoms to maintain coherence. These results highlight the importance of strong electronic correlations in magnetically intercalated transition-metal dichalcogenides and provide a microscopic understanding of features not captured by conventional DFT+U.

cond-mat.str-el

The ORCA Benchmark: Evaluating Real-World Calculation Accuracy in Large Language Models

We present ORCA (Omni Research on Calculation in AI) Benchmark - a novel benchmark that evaluates large language models (LLMs) on multi-domain, real-life quantitative reasoning using verified outputs from Omni's calculator engine. In 500 natural-language tasks across domains such as finance, physics, health, and statistics, the five state-of-the-art systems (ChatGPT-5, Gemini~2.5~Flash, Claude~Sonnet~4.5, Grok~4, and DeepSeek~V3.2) achieved only $45\text{--}63\,\%$ accuracy, with errors mainly related to rounding ($35\,\%$) and calculation mistakes ($33\,\%$). Results in specific domains indicate strengths in mathematics and engineering, but weaknesses in physics and natural sciences. Correlation analysis ($r \approx 0.40\text{--}0.65$) shows that the models often fail together but differ in the types of errors they make, highlighting their partial complementarity rather than redundancy. Unlike standard math datasets, ORCA evaluates step-by-step reasoning, numerical precision, and domain generalization across real problems from finance, physics, health, and statistics.

cs.AI