arXiv subjects

Subhadip Mitra

Publications and source records attributed to Subhadip Mitra.

At least 19 recordsLinked to original sources

arXiv2026

Extracting Work from Discrete Quantum Polytropic Processes

We establish an upper bound on extractable work for time-dependent, non-Markovian quantum heat engines operating with finite baths. This bound analytically isolates the distinct thermodynamic penalties arising from system-bath correlations, bath non-equilibrium, and residual interaction energy. Evaluating this framework operationally via a quantum polytropic cavity-optomechanical cycle, we demonstrate that maximal efficiency requires quasi-static operation to successfully harvest coherent, non-Markovian system-bath resonances. Conversely, optimising for maximum power enforces a strict finite-time regime. Under realistic hardware constraints, this acceleration necessitates larger discrete operational steps, where we expect Trotterisation errors to manifest as physical noise. Such noise would irreversibly suppress delicate quantum memory effects, forcing a collapse to the memoryless Markovian Otto limit. Coupled with the permanent energetic tax of switching finite-bath interactions, our results indicate that the exploitation of quantum memory resources and finite-power operation belong to different operational regimes.

quant-ph↗

arXiv2026

Closing the Activation-Cone Blind Spot: Response-Time Probing and Unified Defense

Inference-time safety methods for large language models have proliferated, yet no systematic comparison exists. We evaluate five defense paradigms (no defense, static steering, CAST, AlphaSteer, probe-gated) across seven instruction-tuned models (7-31B) and five attack types (GCG, AutoDAN, DeepInception, prefilling, intent laundering). Our central finding: prompt-time activation defenses are structurally blind to prefilling attacks. AlphaSteer achieves 0% attack success on GCG, AutoDAN, and intent laundering but 50% on prefilling. We prove a corollary: any defense that gates intervention on a single layer's activation alignment with a benign reference (cone, subspace, or null-space) is blind to attacks that craft activations to lie inside that reference, whether checked at prompt time or per token. As its constructive contrapositive we introduce response-time probing: a linear probe on the model's hidden state at the first generated tokens, with AUROC 0.97-1.00 across all seven models. Combined with a halt, it cuts prefilling attack success to 0/40 on every model with 0% benign false positives, outperforming Llama Guard 3. Cross-template generalisation depends on probe depth, so we scope the claim to the canonical prefilling-template family. Composing the response-halt with AlphaSteer's null-space steering gives an orthogonal split (the halt catches prefilling, AlphaSteer catches semantic attacks), reaching defense success 0.983 on Mistral and 0.994 on Llama and dominating both components. We further show MMLU fails to capture steering's true utility cost, which appears as behavioral hedging rather than factual loss, and that diverse negative training sets cut probe false positives from 80-100% to near zero. Code, attacks, per-sample results, and the judge prompt are released.

cs.CR↗

arXiv2026

Testing varying coupling constants through multi-Higgs production at the LHC

We propose the One Scalar Theory (1ST), a minimalist framework where a single real singlet scalar field mediates the dynamical generation of the Higgs self-coupling and the top Yukawa coupling. Unlike generic portal models, the 1ST removes parametric freedom by locking production and decay modes to a single fundamental scale $\Lambda_0$, rendering the framework highly predictive with unique experimental signals. We demonstrate that the collider phenomenology is partitioned by the $2m_t$ kinematic threshold into di-Higgs and di-top resonance regimes. By recasting current ATLAS data, we set lower bounds on $\Lambda_0$ at the TeV scale and show that the High-Luminosity LHC will probe this scale up to $4$ TeV, providing a definitive test for the dynamical origin of the electroweak sector.

hep-ph↗

arXiv2026

Quality-Diversity Evolution for Discovering Diverse Vulnerabilities in LLM Safety

Current approaches to LLM adversarial testing suffer from coverage gaps: manual red-teaming does not scale, LLM-as-attacker methods exhibit mode collapse, and gradient-based approaches produce uninterpretable gibberish. We introduce a quality-diversity evolutionary framework that operates at the semantic level, evolving interpretable attack strategies rather than token sequences. Using MAP-Elites, we maintain a diverse archive of attacks across behavioral dimensions (strategy type, encoding method, prompt length). In experiments across GPT-4o-mini, Claude 3.5 Sonnet, Gemini 2.0 Flash, and an open-weight coding model (Devstral-small-2), we discover distinct vulnerability profiles: GPT-4o-mini is vulnerable to hypothetical and multi-turn framing combined with ROT13 encoding (fitness 0.8), Gemini to direct attacks with ROT13 and multi-turn with Leetspeak (0.8), while Claude shows uniformly ambiguous responses across all strategies (max 0.4). The semantic representation produces interpretable attacks that reveal systematic, model-specific weaknesses, providing actionable insights for improving LLM safety and a reproducible baseline for evaluating future frontier models. Code and experiment artifacts are released at https://github.com/bassrehab/red-queen.

cs.CR↗

arXiv2026

Cross-Generational Transfer of Adversarial Attacks Reveals Non-Monotonic Safety Alignment in LLMs

Safety alignment in LLMs does not improve monotonically across model generations. Studying four generations of Google's Gemma family (7B-31B) with quality-diversity evolution (MAP-Elites) as an automated red-teaming probe, we find that Gemma 3 (12B) exhibits 68.7% +/- 5.7% attack success rate (ASR; mean +/- std, 3 seeds), significantly higher than its predecessor Gemma 2 (45.5% +/- 7.2%; p = 0.030, paired bootstrap) and its successor Gemma 4 (33.9% +/- 1.8%). Replaying evolved attack archives across generations reveals that attacks from other generations transfer to Gemma 3 at 44-46% but only 14-18% to Gemma 4, indicating that Gemma 4's safety gains generalize beyond the attack distributions evolved against earlier generations. Under our 8B judge, copyright and cybercrime vulnerabilities register at near-100% across all generations, though a second-judge audit (Section 6) suggests the copyright result is sensitive to judge choice. Misinformation ASR jumps from 29% to 99% between Gemma 2 and Gemma 3 and remains elevated at 77% in Gemma 4, indicating the regression was not fully addressed. These patterns are invisible to static benchmarks and emerge only through adaptive, longitudinal probing. All experiments use 3 random seeds with a unified self-hosted judge; code and artifacts are available at https://github.com/bassrehab/red-queen.

cs.CR↗

arXiv2026

Photon-initiated enhancements in the pair production of highly charged coloured particles

Strong interaction is typically assumed to dominate the pair production of heavy coloured resonances at the LHC. However, mixed QCD-QED contributions from gluon-photon ($g\gamma$) initial states become critical for highly charged states. This contribution scales with the square of their electric charges and maximises for particles in the fundamental colour representation. We study this effect for leptoquarks, which are colour-triplet bosons. We demonstrate that tree-level mixed QCD-QED contributions enhance their pair-production rates by up to $\sim 33\%$ for a charge-$5/3$ state, rivalling the size of next-to-leading-order QCD corrections. The asymmetric colour flow of $g\gamma$ fusion affects the radiation pattern, altering jet multiplicities and angular distributions. By recasting the latest ATLAS $\mu\mu jj$ search data, we find that these often-overlooked QED effects systematically strengthen mass exclusion limits, establishing a necessary precision standard for bounding highly charged coloured states.

hep-ph↗

arXiv2026

Cross-Platform Fused MoE Dispatch in Triton: Portable Expert Routing Without CUDA

Mixture-of-Experts (MoE) architectures power the majority of frontier large language models, but their inference is bottlenecked by irregular memory access patterns and expert routing overhead. Existing optimized MoE kernels (Megablocks, Tutel, FasterMoE) are implemented in CUDA and locked to NVIDIA hardware. We present TritonMoE, a fused MoE dispatch kernel written entirely in OpenAI Triton that performs the complete forward pass -- router scoring, token permutation, expert GEMMs, and weighted output combination -- using only portable Triton primitives. Our key optimization is a fused gate+up GEMM kernel that computes both SwiGLU projections from shared L2-cached input tiles with in-register SiLU activation, eliminating 35% of global memory traffic. On an NVIDIA A100, TritonMoE achieves 89-131% of the throughput of the CUDA-optimized Megablocks at inference batch sizes (<= 512 tokens) across Mixtral-8x7B, DeepSeek-V3, and Qwen2-MoE configurations. All 162 correctness tests pass on both NVIDIA A100 and AMD MI300X with zero code changes, validating cross-platform portability. We additionally characterize sensitivity to routing imbalance under Zipfian-skewed expert assignments and identify the regime -- 64+ experts under extreme skew -- where our fixed-tile scheduling underperforms Megablocks' block-sparse layout, motivating dynamic block-to-expert assignment as future work. Code is available at https://github.com/bassrehab/triton-kernels.

cs.DC↗

arXiv2026

Kinematic budget of quantum correlations

The diversity of quantum correlations -- discord, entanglement, steering, and Bell nonlocality -- disappears at the kinematic level of observable second moments. By treating state purity as a finite resource, we introduce a local-unitary-invariant budget that splits these moments into local and nonlocal sectors. This maps quantum systems onto compact, two-dimensional manifolds whose topology is governed by purity and time-reversal symmetry. This dimensional reduction reveals a deep structural link: exceeding classical capacity limits requires the activation of intrinsically time-odd generators, providing a dimension-agnostic guarantee of negative partial transpose (NPT) entanglement. For two qubits, this geometry is analytically solvable; a single boundary isolates classical correlations, while nested regions define thresholds for steering and Bell nonlocality, alongside bounds on non-stabiliser magic. Beyond two qubits, dimensional bottlenecks enforce the kinematic limits on correlations. Because this macroscopic representation is completely determined by global and marginal purities, it bypasses the exponential scaling of full-state tomography. Thus, whenever an $n$-partite state's correlations exceed the classical capacity limits, its NPT entanglement is certified by only $n+1$ purity measurements, with sample complexity independent of Hilbert space dimension. By coarse-graining over gauge-like first moments, this geometry acts as a thermodynamic phase diagram, exposing the hierarchy of quantum resources and their dynamic redistribution under decoherence.

quant-ph↗

arXiv2026

Constraint-Aware Execution Planning for Hybrid Space-Ground Compute Workloads

Low Earth orbit (LEO) satellites increasingly carry compute hardware capable of on-board processing, yet each satellite generates roughly two orders of magnitude more data than it can downlink per orbit. This mismatch forces operators to decide, for every workload, which computation runs on-board and which runs on the ground, how intermediate data crosses the space-ground boundary through narrow contact windows, and how to maintain delivery guarantees over noisy channels. We present Constraint-Aware Execution (CAE), a planning system that takes a satellite identifier, a workload expressed as a directed acyclic graph of processing steps, and a set of orbital and resource constraints, and produces a deterministic, physically grounded execution plan. CAE operates in four phases: (1) orbital environment construction via SGP4 propagation with eclipse detection and ground station pass prediction, (2) compute placement using a cost model that compares on-board resource consumption against transfer overhead, (3) transfer insertion with adaptive forward error correction and security overhead modeling, and (4) greedy first-fit scheduling into orbital windows under power, thermal, compute, and communication constraints. We evaluate CAE against five representative workload patterns across satellites in distinct orbital regimes and demonstrate that the system produces feasible plans in under two seconds, correctly exploits onboard data reduction to minimize transfer volume, and adapts FEC and multi-pass allocation to varying channel conditions. CAE is deployed as a production API computing plans for any cataloged satellite using live two-line element data.

cs.DC↗

arXiv2026

Field-Theoretic Memory for AI Agents: Continuous Dynamics for Context Preservation

We present a memory system for AI agents that treats stored information as continuous fields governed by partial differential equations rather than discrete entries in a database. The approach draws from classical field theory: memories diffuse through semantic space, decay thermodynamically based on importance, and interact through field coupling in multi-agent scenarios. We evaluate the system on two established long-context benchmarks: LoCoMo (ACL 2024) with 300-turn conversations across 35 sessions, and LongMemEval (ICLR 2025) testing multi-session reasoning over 500+ turns. On LongMemEval, the field-theoretic approach achieves significant improvements: +116% F1 on multi-session reasoning (p<0.01, d= 3.06), +43.8% on temporal reasoning (p<0.001, d= 9.21), and +27.8% retrieval recall on knowledge updates (p<0.001, d= 5.00). Multi-agent experiments show near-perfect collective intelligence (>99.8%) through field coupling. Code is available at github.com/rotalabs/rotalabs-fieldmem.

cs.CL↗

arXiv2026

Spark-LLM-Eval: A Distributed Framework for Statistically Rigorous Large Language Model Evaluation

Evaluating large language models at scale remains a practical bottleneck for many organizations. While existing evaluation frameworks work well for thousands of examples, they struggle when datasets grow to hundreds of thousands or millions of samples. This scale is common when assessing model behavior across diverse domains or conducting comprehensive regression testing. We present Spark-LLM-Eval, a distributed evaluation framework built natively on Apache Spark. The system treats evaluation as a data-parallel problem, partitioningexamplesacrossexecutorsandaggregatingresultswithproperstatistical accounting. Beyond raw throughput, we emphasize statistical rigor: every reported metric includes bootstrap confidence intervals, and model comparisons come with appropriate significance tests (paired t-tests, McNemar's test, or Wilcoxon signed-rank, depending on the metric type). The framework also addresses the cost problem inherent in LLM evaluation through content-addressable response caching backed by Delta Lake, which allows iterating on metric definitions without re-running inference. We describe the system architecture, the statistical methodology, and report benchmark results showing linear scaling with cluster size. The framework and all evaluation code are available as open source.

cs.DC↗

arXiv2025

Vectorlike $\tau$ production through leptoquarks

Numerous phenomenological studies and collider searches have probed for the existence of new physics by looking for signatures of leptoquarks (LQs) or vectorlike leptons (VLLs). We consider a new possibility that can arise in theories with enhanced gauge symmetries: both particles are simultaneously present, and LQ-mediated processes enhance the VLL production at the LHC. We study the effect of non-standard interactions of LQs that contribute to novel production and decay signatures. We obtain the HL-LHC prospects of this framework in the mono-and di-lepton final states, and discuss other potentially relevant channels.

hep-ph↗

arXiv2025

Fresh look at the LHC limits on vector leptoquarks

Vector leptoquarks (vLQs) are popular candidates for searching for physics beyond the Standard Model. In this paper, we present updated exclusion limits on various vLQ species, accounting for all the relevant production mechanisms at the LHC. In particular, we highlight the critical role of indirect production and its interference with the Standard Model Drell-Yan process. This interference can be constructive or destructive, depending on the specific quantum numbers of the vLQ, and significantly impacts the sensitivity of current searches. Furthermore, we demonstrate that including QCD-QED mixed pair production channels leads to a noticeable shift in model-independent mass limits. Additionally, we examine the validity of the full theory with vLQs and corresponding effective operators in the high mass regime. Overall, our analysis yields a substantial improvement in the exclusion limits on vLQs compared to the existing results in the literature.

hep-ph↗

arXiv2025

Tagging fully hadronic exotic decays of the vectorlike $\mathbf{B}$ quark using a graph neural network

Following up on our earlier study in [J. Bardhan et al., Machine learning-enhanced search for a vectorlike singlet B quark decaying to a singlet scalar or pseudoscalar, Phys. Rev. D 107 (2023) 115001; arXiv:2212.02442], we investigate the LHC prospects of pair-produced vectorlike $B$ quarks decaying exotically to a new gauge-singlet (pseudo)scalar field $\Phi$ and a $b$ quark. After the electroweak symmetry breaking, the $\Phi$ decays predominantly to $gg/bb$ final states, leading to a fully hadronic $2b+4j$ or $6b$ signature. Because of the large Standard Model background and the lack of leptonic handles, it is a difficult channel to probe. To overcome the challenge, we employ a hybrid deep learning model containing a graph neural network followed by a deep neural network. We estimate that such a state-of-the-art deep learning analysis pipeline can lead to a performance comparable to that in the semi-leptonic mode, taking the discovery (exclusion) reach up to about $M_B=1.8\:(2.4)$ TeV at HL-LHC when $B$ decays fully exotically, i.e., BR$(B \to b\Phi) = 100\%$.

hep-ph↗

arXiv2025

Unravelling the non-Markovian spin-boson model and quantum quasi-Otto cycle

We use the spin-boson model to describe the dynamics of a two-level atom interacting with Fabry-P\'erot cavity modes. We solve the Schr\"odinger equation for the system-bath model without the Born-Markov approximation to derive the non-Markovian reduced dynamics of the qubit. We further construct an exact Lindblad-type master equation for it. Similar to the quantum Otto cycle, we construct a non- Markovian quasi-cyclic process based on the atom-cavity interactions, which we call the quasi-Otto cycle. For judicious choices of input state and parameters, the quasi-cycle can be more efficient as a quantum engine than the Otto cycle. We also showed that if the quasi-cycle is repeated multiple times, the efficiency of the quasi-Otto engine asymptotically approaches that of the Otto engine.

quant-ph↗

arXiv2025

Simulating quantum collision models with Hamiltonian simulations using early fault-tolerant quantum computers

We develop randomized quantum algorithms to simulate quantum collision models, also known as repeated interaction schemes, which provide a rich framework to model various open-system dynamics. The underlying technique involves composing time evolutions of the total (system, bath, and interaction) Hamiltonian and intermittent tracing out of the environment degrees of freedom. This results in a unified framework where any near-term Hamiltonian simulation algorithm can be incorporated to implement an arbitrary number of such collisions on early fault-tolerant quantum computers: we do not assume access to specialized oracles such as block encodings and minimize the number of ancilla qubits needed. In particular, using the correspondence between Lindbladian evolution and completely positive trace-preserving maps arising out of memoryless collisions, we provide an end-to-end quantum algorithm for simulating Lindbladian dynamics. For a system of $n$-qubits, we exhaustively compare the circuit depth needed to estimate the expectation value of an observable with respect to the reduced state of the system after time $t$ while employing different near-term Hamiltonian simulation techniques, requiring at most $n+2$ qubits in all. We compare the CNOT gate counts of the various approaches for estimating the Transverse Field Magnetization of a $10$-qubit XX-Heisenberg spin chain under amplitude damping. Finally, we also develop a framework to efficiently simulate an arbitrary number of memory-retaining collisions, i.e., where environments interact, leading to non-Markovian dynamics. Overall, our methods can leverage quantum collision models for both Markovian and non-Markovian dynamics on early fault-tolerant quantum computers, shedding light on the advantages and limitations of simulating open systems dynamics using this framework.

quant-ph↗

arXiv2025

HEP-JEPA: A foundation model for collider physics using joint embedding predictive architecture

We present a transformer architecture-based foundation model for tasks at high-energy particle colliders such as the Large Hadron Collider. We train the model to classify jets using a self-supervised strategy inspired by the Joint Embedding Predictive Architecture. We use the JetClass dataset containing 100M jets of various known particles to pre-train the model with a data-centric approach -- the model uses a fraction of the jet constituents as the context to predict the embeddings of the unseen target constituents. Our pre-trained model fares well with other datasets for standard classification benchmark tasks. We test our model on two additional downstream tasks: top tagging and differentiating light-quark jets from gluon jets. We also evaluate our model with task-specific metrics and baselines and compare it with state-of-the-art models in high-energy physics. Project site: https://hep-jepa.github.io/

cs.LG↗

arXiv2024

TooLQit: Leptoquark Models and Limits

We introduce the leptoquark (LQ) toolkit, TooLQit, which includes leading-order FeynRules models for all types of LQs and a Python-based calculator, named CaLQ, to test if a set of parameter points are allowed by the LHC dilepton searches. The models include electroweak gauge interactions of the LQs and follow a set of intuitive notations. Currently, CaLQ can calculate the LHC limits on LQ ($S_1$ and $U_1$) couplings (one or more simultaneously) for any mass between $1$ and $5$ TeV using a $\chi^2$ method. In this manual for TooLQit, we describe the FeynRules models and discuss the techniques used in CaLQ. We outline the workflow to check parameter spaces of LQ models with an example. We show some illustrative scans for one- and multi-coupling scenarios for the $U_1$ vector LQ. The TooLQit code is available at https://github.com/rsrchtsm/TooLQit

hep-ph↗