arXiv subjects

José H. García

Publications and source records attributed to José H. García.

6 recordsLinked to original sources

arXiv2026

Transferable knowledge graphs with executable learned operators for algorithm design

Procedural knowledge in algorithm design is embedded in source code and rebuilt for each new domain. We introduce Generative Executable Algorithm Knowledge Graphs (GEAKG), a representation in which this knowledge is stored as a generative, executable, transferable graph: typed nodes hold validated operators, edges encode admissible compositions, and learned edge weights record effective sequences. The same engine instantiates the structure across domains by changing only a role ontology (RoleSchema) and a binding. We study GEAKG as a representation mechanism rather than a state-of-the-art optimizer, asking what transfers and when. Layer ablations localize transfer by granularity: within a neural-architecture-search family the learned snapshot transfers across 70 dataset pairs - its weights stay correlated across datasets and one frozen snapshot remains competitive with Regularized Evolution at zero deployment-token cost; across combinatorial domains only the ontology-constrained executable structure transfers, not the learned weights. That structure pays off where target-side search is expensive - a Traveling Salesman snapshot beats an equally untuned from-scratch search on large scheduling instances even at one-fifth its budget - but does not improve on an effective local search where one is cheap, as in assignment and linear ordering. Executable procedural knowledge can thus be acquired offline, compacted, inspected, and reused without runtime language-model calls.

cs.AI↗

arXiv2026

A Fixed-Budget, Cluster-Aware Standard for LLM-as-a-Judge Evaluation: A Multi-Hop RAG Stress Test

Retrieval-augmented generation (RAG) systems are often compared by asking a large language model (LLM) judge which answer is better. For multi-hop RAG, this has become a measurement problem as much as a modeling problem: the same score can reflect retrieval quality, answer length, lexical overlap, or a statistical test that ignores clustered data. We ask what happens when these choices are made explicit. We propose a minimum measurement standard for LLM-as-a-judge comparisons in RAG. The standard fixes the top-100 candidate pool, evidence budget, answer cap, generator, and prompt; it also requires pre-registered hypotheses, cluster-aware inference, an exact cluster sign-flip check when feasible, and second-judge replication. Clustered benchmarks can overstate progress; the field should adopt this standard. We stress-test it with Genetic Algorithm Decoder for Multi-hop Evidence Composition (GADMEC), an evolutionary evidence selector, on 400 multi-hop questions in computer science/machine learning (CS/ML) and Materials Science. The protocol changes the empirical story. A binomial test makes all four semantic-baseline comparisons look significant; cluster-aware inference leaves only one Bonferroni-significant result. BM25 beats pure semantic GADMEC under the same budget, while a lexical-semantic hybrid recovers in CS/ML and narrows the Materials Science gap.

cs.AI↗

arXiv2025

Topologically Driven Spin-Orbit Torque in Dirac Matter

We unveil novel spin-orbit torque mechanisms driven by topological edge states in magnetic graphene-based devices. Within the energy gap, a damping-like torque plateau emerges within the quantum anomalous Hall phase upon breaking particle-hole symmetry, while for energies at the spin-split Dirac points located within the bands, a large damping-like torque develops as a result of a vanishing Fermi contour. Such torques are tunable by the degree of spin-pseudospin entanglement dictated by proximity-induced spin-orbit coupling terms.

cond-mat.mes-hall↗

arXiv2024

Giant Spin-Orbit Torque in Cr-based Janus Transition Metal Dichalcogenides

We report a very large spin-orbit torque (SOT) capability of chromium-based transition metal dichalcogenides (TMD) in their Janus forms CrXTe, with X=S,Se. The structural inversion symmetry breaking, inherent to Janus structures is responsible for a large SOT response generated by giant Rashba splitting, equivalent to that obtained by applying a transverse electric field of $\sim 100 \,\text{V} \,\text{nm}^{-1}$ in non-Janus CrTe\textsubscript{2}, completely out of experimental reach. By performing transport simulations on custom-made Wannier tight-binding models, Janus systems are found to exhibit a SOT performance comparable to the most efficient two-dimensional materials, while allowing for field-free perpendicular magnetization switching owing to their reduced in-plane symmetry. Altogether, our findings evidence that magnetic Janus TMDs stand as suitable candidates for ultimate SOT-MRAM devices.

cond-mat.mtrl-sci↗

arXiv2023

Emerging Spin-Orbit Torques in Low Dimensional Dirac Materials

We report a theoretical description of novel spin-orbit torque components emerging in two-dimensional Dirac materials with broken inversion symmetry. In contrast to usual metallic interfaces where field-like and damping-like torque components are competing, we find that an intrinsic damping-like torque which derives from all Fermi-sea electrons can be simultaneously enhanced along with the field-like component. Additionally, hitherto overlooked torque components unique to Dirac materials, emerge from the coupling between spin and pseudospin degrees of freedom. These torques are found to be resilient to disorder and could enhance the magnetic switching performance of nearby magnets.

cond-mat.mes-hall↗

arXiv2018

Shubnikov-de Haas oscillations in the anomalous Hall conductivity of Chern insulators

The Haldane model on a honeycomb lattice is a paradigmatic example of a system featuring quantized Hall conductivity in the absence of an external magnetic field, that is, a quantum anomalous Hall effect. Recent theoretical work predicted that the anomalous Hall conductivity of massive Dirac fermions can display Shubnikov-de Haas (SdH) oscillations, which could be observed in topological insulators and honeycomb layers with strong spin--orbit coupling. Here, we investigate the electronic transport properties of Chern insulators subject to high magnetic fields by means of accurate spectral expansions of lattice Green's functions. We find that the anomalous component of the Hall conductivity displays visible SdH oscillations at low temperature. \textcolor{black}{The effect is shown to result from the modulation of the next-nearest neighbour flux accumulation due to the Haldane term,} which removes the electron--hole symmetry from the Landau spectrum. To support our numerical findings, we derive a long-wavelength description beyond the linear ('Dirac cone') approximation. Finally, we discuss the dependence of the energy spectra shift for reversed magnetic fields with the topological gap and the lattice bandwidth.

cond-mat.mes-hall↗