arXiv subjects

Sourav Karmakar

Publications and source records attributed to Sourav Karmakar.

12 recordsLinked to original sources

arXiv2026

Interplay of dimerization and quasiperiodicity in the superconducting proximity effect of a one-dimensional hybrid ring

Recent studies of the superconducting proximity effect in quasicrystalline and topological systems have opened up a new research direction for exploring how quasiperiodicity and topology influence proximity induced superconductivity. In this work, we investigate spatial variation of the proximity induced pairing amplitude in a hybrid ring composed of a spin singlet superconductor and a normal region described by three different lattice models using the self-consistent Bogoliubov-de Gennes formalism. We first consider the normal region described by the diagonal Aubry-André-Harper (AAH) model. Increasing the quasiperiodic potential enhances spatial fluctuations in the induced order parameter, while progressively suppressing its magnitude in the normal region. Beyond the localization transition, proximity-induced pairing is strongly diminished due to the localized nature of the underlying electronic states. The normal region is then modeled by the Su-Schrieffer-Heeger (SSH) chain to investigate the effect of hopping dimerization. Weak dimerization introduces oscillatory modulations in the induced pairing that extend deep into the normal region, whereas strong dimerization confines these oscillations and significantly reduces the penetration of superconducting correlations. Finally, we study the combined SSH-AAH model to explore the interplay between the dimerized hopping and quasiperiodicity. The results show that the SSH dimerization determines the oscillatory behavior and penetration of the induced pairing, while the AAH potential enhances spatial inhomogeneity and further reduces its magnitude. Together, these two effects provide a versatile means of controlling proximity-induced superconductivity in quasiperiodic hybrid systems.

cond-mat.supr-con↗

arXiv2026

Proof-Carrying Cognition: Closing the Verification Gap with Reality-Settled Reward

Frontier gains in language-model reasoning come from reinforcement learning on reasoning traces and are concentrated in domains with a cheap, sound verifier. We argue the field's binding constraint is the verification gap: no scalable, incorruptible reward for reasoning outside formal domains. We make four contributions. (1) Theory: in a joint-Gaussian model of best-of-N selection, verifier-gold correlation rho is the exact exchange rate between test-time compute and capability, and an unsound verifier pays a polynomial penalty N^(1/rho^2); a margin-free copula form predicts realized soundness of real LLM judges to 4% median error. (2) Demonstration: in program-synthesis testbeds with executable ground truth, including a pre-registered scaled replication, unsound verifiers lose Soundness-under-Pressure as optimization grows (0.94 to 0.32 at N=4096) while a sound verifier improves monotonically; reality-anchored settlement beats a frozen verifier under i.i.d. and adversarial pressure, driving the hacking gap from ~0.27 to ~0; soundness scales log-linearly with settled labels, with on-policy settlement ~10x more label-efficient than random labeling. With real LLM judges and unit-test execution as gold, a weak judge loses soundness under best-of-N (p<0.001), a stronger judge is more robust, and selection alone manufactures +0.53 hacking gaps from honest samples. Under real GRPO training, a frozen reward model traces the full overoptimization curve (executed reward collapses 90%) while the same model refit on a 10% settlement stream preserves 6x the executed reward. (3) Paradigm: proof-carrying cognition, where reasoning steps are typed probabilistic claims priced by a self-built world model trained only on held-out reality and settled by proper scoring rules. (4) Benchmark: we specify Soundness-under-Pressure as the headline metric for a reality-settled reasoning benchmark.

cs.AI↗

arXiv2026

Bosonic Condensed Phase Real-time Dynamics from Ring Polymer Molecular Dynamics

We present a method for approximating real-time correlation functions and quantum transport coefficients of bosonic condensed phases. The direct evaluation of quantum real-time correlation functions in the path integral formulation is impossible due to a severe dynamical sign problem. Evaluating imaginary-time correlation functions and inverting them using analytic continuation is famously ill-posed. Ring polymer molecular dynamics (RPMD) provides an alternative approach resulting in accurate approximations for real-time correlation functions of condensed phases but neglects exchange effects. In this work, we develop a bosonic RPMD method, which is exact in the harmonic, short time, and high temperature limits. A critical enabling observation is that the Kubo-transformed correlation function for bosons is real only when the correlated observables are symmetric under exchange. We benchmark the method on harmonic and anharmonic model systems and then use it to directly obtain the Lieb-Liniger gas density-density real-time correlation functions at various temperatures and momentum transfers. This work enables the direct simulation of finite-temperature, real-time dynamics of large bosonic condensed phases using RPMD for the first time.

physics.chem-ph↗

arXiv2026

Are Large Language Models Truly Smarter Than Humans?

Public leaderboards increasingly suggest that large language models (LLMs) surpass human experts on benchmarks spanning academic knowledge, law, and programming. Yet most benchmarks are fully public, their questions widely mirrored across the internet, creating systematic risk that models were trained on the very data used to evaluate them. This paper presents three complementary experiments forming a rigorous multi-method contamination audit of six frontier LLMs: GPT-4o, GPT-4o-mini, DeepSeek-R1, DeepSeek-V3, Llama-3.3-70B, and Qwen3-235B. Experiment 1 applies a lexical contamination detection pipeline to 513 MMLU questions across all 57 subjects, finding an overall contamination rate of 13.8% (18.1% in STEM, up to 66.7% in Philosophy) and estimated performance gains of +0.030 to +0.054 accuracy points by category. Experiment 2 applies a paraphrase and indirect-reference diagnostic to 100 MMLU questions, finding accuracy drops by an average of 7.0 percentage points under indirect reference, rising to 19.8 pp in both Law and Ethics. Experiment 3 applies TS-Guessing behavioral probes to all 513 questions and all six models, finding that 72.5% trigger memorization signals far above chance, with DeepSeek-R1 displaying a distributed memorization signature (76.6% partial reconstruction, 0% verbatim recall) that explains its anomalous Experiment 2 profile. All three experiments converge on the same contamination ranking: STEM > Professional > Social Sciences > Humanities.

cs.AI↗

arXiv2026

Persistent Charge and Spin Currents in a Ferromagnetic Hatano-Nelson Ring

We investigate persistent charge and spin currents in a ferromagnetic Hatano-Nelson ring with anti-Hermitian intradimer hopping, where non-reciprocal hopping generates a synthetic magnetic flux and drives a non-Hermitian Aharonov-Bohm effect. The system supports both real and imaginary persistent currents, with ferromagnetic spin splitting enabling all three spin-current components, dictated by the orientation of magnetic moments. The currents are computed using the current operator method within a biorthogonal basis. In parallel, the complex band structure is analyzed to uncover the spectral characteristics. We emphasize how the currents evolve across different topological regimes, and how they are influenced by chemical potential, ferromagnetic ordering, finite size, and disorder. Strikingly, disorder can even amplify spin currents, opening powerful new routes for manipulating spin transport in non-Hermitian systems.

cond-mat.mes-hall↗

arXiv2025

Combining Harmonic Sampling with the Worm Algorithm to Improve the Efficiency of Path Integral Monte Carlo

We propose an improved Path Integral Monte Carlo (PIMC) algorithm called Harmonic PIMC (H-PIMC) and its generalization, Mixed PIMC (M-PIMC). PIMC is a powerful tool for studying quantum condensed phases. However, it often suffers from a low acceptance ratio for solids and dense confined liquids. We develop two sampling schemes especially suited for such problems by dividing the potential into its harmonic and anharmonic contributions. In H-PIMC, we generate the imaginary time paths for the harmonic part of the potential exactly and accept or reject it based on the anharmonic part. In M-PIMC, we restrict the harmonic sampling to the vicinity of local minimum and use standard PIMC otherwise, to optimize efficiency. We benchmark H-PIMC on systems with increasing anharmonicity, improving the acceptance ratio and lowering the auto-correlation time. For weakly to moderately anharmonic systems, at $β\hbar ω=16$, H-PIMC improves the acceptance ratio by a factor of 6-16 and reduces the autocorrelation time by a factor of 7-30. We also find that the method requires a smaller number of imaginary time slices for convergence, which leads to another two- to four-fold acceleration. For strongly anharmonic systems, M-PIMC converges with a similar number of imaginary time slices as standard PIMC, but allows the optimization of the auto-correlation time. We extend M-PIMC to periodic systems and apply it to a sinusoidal potential. Finally, we combine H- and M-PIMC with the worm algorithm, allowing us to obtain similar efficiency gains for systems of indistinguishable particles.

physics.comp-ph↗

arXiv2025

Reentrant localization in a quasiperiodic chain with correlated hopping sequences

Quasiperiodic systems are known to exhibit localization transitions in low dimensions, wherein all electronic states become localized beyond a critical disorder strength. Interestingly, recent studies have uncovered a reentrant localization (RL) phenomenon: upon further increasing the quasiperiodic modulation strength beyond the localization threshold, a subset of previously localized states can become delocalized again within a specific parameter window. While RL transitions have been primarily explored in systems with simple periodic modulations, such as dimerized or long-range hopping integrals, the impact of more intricate or correlated hopping structures on RL behavior remains largely elusive. In this work, we investigate the localization behavior in a one-dimensional lattice featuring staggered, correlated on-site potentials following the Aubry-André-Harper model, along with off-diagonal hopping modulations structured according to quasiperiodic Fibonacci and Bronze Mean sequences. By systematically analyzing the fractal dimension, inverse participation ratio, and normalized participation ratio, we demonstrate the occurrence of RL transitions induced purely by the interplay between quasiperiodic on-site disorder and correlated hopping. We further examine the parameter space to determine the specific regimes that give rise to RL. Our findings highlight the crucial role of underlying structural correlations in governing localization-delocalization transitions in low-dimensional quasiperiodic systems, where the correlated disorder manifests in both diagonal and off-diagonal terms.

cond-mat.dis-nn↗

arXiv2025

Circular currents in a magnetic ring with zero net magnetization in presence of a side-coupled one-dimensional chain

We investigate persistent charge and spin currents in a magnetic quantum ring threaded by an Aharonov-Bohm flux, in the presence of a side-coupled one-dimensional non-magnetic chain. The neighboring magnetic moments in the ring are arranged in an antiparallel configuration. In the absence of the chain, the spin circular current vanishes exactly due to the symmetry between the up and down spin sub-Hamiltonians. Modeling the system within a tight-binding framework, we compute the currents using a second-quantized approach. Both charge and spin currents can be selectively tuned by adjusting the ring-chain coupling strength. Temperature plays a crucial role in modulating the currents, and interestingly, we find that they increase significantly with rising temperature--contrary to conventional expectations.

cond-mat.mes-hall↗

arXiv2021

Arnold web and dynamical tunneling in a four-site Bose-Hubbard model

We present the quantum and classical study of a four-well trapped Bose-Einstein condensate (BEC) modeled by the Bose-Hubbard Hamiltonian. The model used is fashioned as a minimal bipartite system consisting of a trimer coupled weakly to a monomer. Semiclassical insights into the fragmentation dynamics of the condensate is obtained by mapping out the Arnold web of the high dimensional classical phase space. A remarkable one-to-one correspondence between the quantum state space in occupation number representation and the classical Arnold web in action space is observed. The relevance of dynamical tunneling for the dynamics of Fock states near resonance junctions on the Arnold web is highlighted. We also predict the possibility of manipulating the transport within the trimer subsystem due to the weak coupling to the monomer.

nlin.CD↗

arXiv2020

Intramolecular vibrational energy redistribution and the quantum ergodicity transition: a phase space perspective

Intramolecular vibrational energy redistribution (IVR) impacts the dynamics of reactions in a profound way. Theoretical and experimental studies are increasingly indicating that accounting for the finite rate of energy flow is critical for uncovering the correct reaction mechanisms and calculating accurate rates. This requires an explicit understanding of the influence and interplay of the various anharmonic (Fermi) resonances that lead to the coupling of the vibrational modes. In this regard, the local random matrix theory (LRMT) and the related Bose-statistics triangle rule (BSTR) model have emerged as a powerful and predictive quantum theories for IVR. In this Perspective we highlight the close correspondence between LRMT and the classical phase space perspective on IVR, primarily using model Hamiltonians with three degrees of freedom. Our purpose for this is threefold. First, this clearly brings out the extent to which IVR pathways are essentially classical, and hence crucial towards attempts to control IVR. Second, given that LRMT and BSTR are designed to be applicable for large molecules, the exquisite correspondence observed even for small molecules allows for insights into the quantum ergodicity transition. Third, we showcase the power of modern nonlinear dynamics methods in analysing high dimensional phase spaces, thereby extending the deep insights into IVR that were earlier gained for systems with effectively two degrees of freedom. We begin with a brief overview of recent examples where IVR plays an important role and conclude by mentioning the outstanding problems and the potential connections to issues of interest in other fields.

nlin.CD↗

arXiv2020

Stable chaos and delayed onset of statisticality in unimolecular dissociation reactions

Statistical models provide a powerful and useful class of approximations for calculating reaction rates by bypassing the need for detailed, and often difficult, dynamical considerations. Such approaches invariably invoke specific assumptions about the extent of intramolecular vibrational energy flow in the system. However, the nature of the transition to the statistical regime as a function of the molecular parameters is far from being completely understood. Here, we use tools from nonlinear dynamics to study the transition to statisticality in a model unimolecular reaction by explicitly visulaizing the high dimensional classical phase space. We identify generic features in the phase space involving the intersection of two or more independent anharmonic resonances and show that the presence of correlated, but chaotic, intramolecular dynamics near such junctions leads to nonstatisticality. Interestingly, akin to the stability of asteroids in the Solar System, molecules can stay protected from dissociation at the junctions for several picoseconds due to the phenomenon of stable chaos.

physics.chem-ph↗

arXiv2018

Relevance of the resonance junctions on the Arnold web to dynamical tunneling and eigenstate delocalization

In this work we study the competition and correspondence between the classical and quantum routes to intramolecular vibrational energy redistribution (IVR) in a three degrees of freedom model effective Hamiltonian. Specifically, we focus on the classical and the quantum dynamics near the resonance junctions on the Arnold web that are formed by intersection of independent resonances. The regime of interest models the IVR dynamics from highly excited initial states near dissociation thresholds of molecular systems wherein both classical and purely quantum, involving dynamical tunneling, routes to IVR coexist. In the vicinity of a resonance junction classical chaos is inevitably present and hence one expects the quantum IVR pathways to have a strong classical component as well. We show that with increasing resonant coupling strengths the classical component of IVR leads to a transition from coherent dynamical tunneling to incoherent dynamical tunneling. Furthermore, we establish that the quantum IVR dynamics can be predicted based on the structures on the classical Arnold web. In addition, we investigate the nature of the highly excited eigenstates in order to identify the quantum signatures of the multiplicity-2 junctions. For the parameter regimes studies herein, by projecting the eigenstates onto the Arnold web, we find that eigenstates in the vicinity of the junctions are primarily delocalized due to dynamical tunneling.

nlin.CD↗