arXiv Science⌕ Search

arXiv subjects

Sayan Ghosh

Publications and source records attributed to Sayan Ghosh.

At least 19 recordsLinked to original sources

Time-changed Compound Poisson Process

In this paper, we introduce a time-changed compound Poisson process (TCCPP) obtained by subordinating the classical compound Poisson process (CPP) with a subordinator and an inverse subordinator, which are defined via distinct Bernstein functions. We study the governing equations and distributional properties of a TCCPP including its moments, dependence structure and limiting behaviour. We derive distributions of the integrated CPP and a linear combination of TCCPPs. We also obtain the law of iterated logarithm, the central limit theorem and weighted sum representation for a TCCPP. Finally, we provide some characterizations of a TCCPP in terms of martingales, a univariate multinomial process, a triangular array process and waiting time distribution. The proposed framework generalizes several existing fractional stochastic processes and offers greater flexibility for modelling complex count data.

math.PR↗

Coupled Magnetic and Lattice Excitations in Multiferroic Ba3HoRu2O9: Insights from Neutron Scattering, Raman Spectroscopy, and Machine-Learned Phonon Calculations

Inelastic neutron scattering (INS) spectra of correlated 4d-4f multiferroic oxides are difficult to interpret because magnetic, crystal-field, and lattice excitations frequently overlap in energy. The hexagonal 6H perovskite Ba3HoRu2O9 provides an ideal platform to investigate this interplay owing to the coexistence of Ru2O9 molecular units and localized Ho3+ moments. To identify the contributions from these different excitations, we combine INS with linear spin-wave calculations, crystal field analysis, Raman spectroscopy, and machine-learned force field (MLFF) phonon calculations. A magnetic excitation below 6.2 meV, observed only in the low-temperature magnetic phase, is reproduced by linear spin-wave theory with intradimer Ru-Ru and Ru-Ho exchange of comparable size and a Ho single-ion anisotropy, and is therefore consistent with a collective spin-wave excitation of the coupled Ru-Ho magnetic network. At higher energies, broad excitations centered near 20, 39, 70, and 90 meV are observed that persist far above the magnetic ordering temperatures. The magnetic part of the 39 meV excitation, isolated by subtracting the Bose-corrected high-temperature data, follows the Ho3+ magnetic form factor. Crystal-field calculations based on the Stevens formalism place a group of strong Ho3+ transitions in this energy range, while Raman spectroscopy and MLFF phonon calculations identify optical phonons with comparable energies. Together, these complementary results show that the broad INS feature near 39 meV arises from overlapping Ho3+ crystal-field excitations and lattice vibrations. These findings establish a microscopic framework for understanding the interplay between spin, crystal-field, and lattice degrees of freedom in this multiferroic 4d-4f compound

cond-mat.str-el↗

Long-range model for electronic pseudospin-nuclear-spin coupling and two-stage ordering in TmVO$_4$

We study an infinite-range coupled electronic-pseudospin and nuclear-spin model for ferroquadrupolar and nuclear-spin ordering in TmVO$_4$ in external magnetic and strain fields. This material is an experimental realization of a transverse-field Ising model, where the Ising degree of freedom is quadrupolar and non-magnetic, but a transverse component is magnetic and couples both to external magnetic fields and to the nuclear spins via a hyperfine coupling. At zero external magnetic field, the electronic and nuclear degrees of freedom undergo well-separated two-stage ordering, accompanied by the release of their respective entropies. A transverse magnetic field polarizes the electronic orbital moments and, via the hyperfine coupling, the nuclear spins. The quadrupolar ordering temperature is gradually reduced to zero. However, the nuclear transition is replaced by a crossover at nonzero magnetic field. Quantum fluctuations are enhanced near the phase transitions and lead to peaks in the magnetic susceptibility. The spectral functions reveal a softening of a low-energy mode near the quantum critical point, consistent with the closing of the excitation gap and its reopening in the disordered phase. These features provide direct dynamical signatures of the field-driven quantum critical phenomena.

cond-mat.str-el↗

Mixed moment and joint non-vanishing of $\mathrm{GL}(3)\times\mathrm{GL}(1)$ and $\mathrm{GL}(1)$ $L$-functions

In this article, we study the mixed-moment problem for the product of central values of two $L$-functions twisted by a Dirichlet character of large conductor $q$. Precisely, for a fixed $\mathrm{SL}_3(\mathbb{Z})$ Hecke-Maass cusp form $π$, we asymptotically evaluate the first moment, with $\textit{power-saving}$ error term for the product of central $L$-values $L(1/2, π\timesχ)\overline{L(1/2, χ)}$ averaged over the $\textit{full family}$ $\{χ\,(\mathrm{mod}\,{q}): Q\leq q\leq 2Q\}$ of primitive Dirichlet characters modulo $q$. And as a consequence, we also established the simultaneous non-vanishing of the aforementioned central $L$-values.

math.NT↗

Enumeration-based Triangle Counting on GPUs

Subgraph enumeration in partitioned graphs across distributed-memory systems is beset with challenges in balancing data movement with the local computation. Especially, counting the number of triangles induced by the closed wedges (i.e., length-3 cycles) is a measure of connectedness in a graph, but often limited by the significant search space caused by open wedges or length-2 paths (usually greater than the number of edges in a graph), which must be inspected to assess whether a particular wedge is closed. To mitigate triangle counting on massive graphs, distributed-memory partitioned-graph approaches are viable (i.e., assessing closed wedges independently within a partition), but load imbalance due to arbitrary communication can overshadow any gains from enhanced concurrency, owing to the inherent lack of structure in regular graphs. In this paper, we discuss a distributed-memory exact graph triangle counting implementation, and assess its performance over billion-edge real-world graphs on contemporary scale-out and scale-up based NVIDIA GPU platforms (considering NVIDIA ``Ampere'' and ``Hopper'' architectures), depicting a scalability spectrum of multiple orders of magnitude. Comparisons with existing state-of-the-art approaches demonstrates up to 10$\times$ improvements relative to distributed-memory CPU baseline, 2-100$\times$ over single-GPU vendor-optimized NVIDIA RAPIDS cuGraph and 10-60\% over multi-GPU $G^2$Miner, in terms of end-to-end execution times.

cs.DC↗

Sample, Align, Synthesize: Graph-Based Response Synthesis with ConGrs

Language models can be sampled multiple times to access the distribution underlying their responses, but existing methods cannot efficiently synthesize rich epistemic signals across different long-form responses. We introduce Consensus Graphs (ConGrs), a flexible DAG-based data structure that represents shared information, as well as semantic variation in a set of sampled LM responses to the same prompt. We construct ConGrs using a light-weight lexical sequence alignment algorithm from bioinformatics, supplemented by the targeted usage of a secondary LM judge. Further, we design task-dependent decoding methods to synthesize a single, final response from our ConGr data structure. Our experiments show that synthesizing responses from ConGrs improves factual precision on two biography generation tasks by up to 31% over an average response and reduces reliance on LM judges by more than 80% compared to other methods. We also use ConGrs for three refusal-based tasks requiring abstention on unanswerable queries and find that abstention rate is increased by up to 56%. We apply our approach to the MATH and AIME reasoning tasks and find an improvement over self-verification and majority vote baselines by up to 6 points of accuracy. We show that ConGrs provide a flexible method for capturing variation in LM responses and using the epistemic signals provided by response variation to synthesize more effective responses.

cs.CL↗

ContextWeaver: Selective and Dependency-Structured Memory Construction for LLM Agents

Large language model (LLM) agents often struggle in long-context interactions. As the history grows, sliding windows and prompt compression may drop earlier information that later steps rely on. Retrieval-based memory systems surface relevant content but still overlook the causal and logical dependencies between steps. We introduce ContextWeaver, a selective and dependency-structured memory framework that organizes an agent's interaction trace into a graph of reasoning steps and selects the relevant context for future actions. ContextWeaver supports (1) dependency-based construction and traversal that link each step to the earlier steps it relies on; (2) compact dependency summaries of the reasoning paths; and (3) a lightweight validation layer that incorporates execution feedback. On five software-engineering and terminal benchmarks with a fixed budget per task, ContextWeaver improves the resolve rate over a sliding-window baseline on all five, with a significant gain when pooled across benchmarks, while using 13-18% fewer agent steps. It also performs favorably against alternative context-management approaches, with the largest gains on long-horizon repository tasks. On SWE-Bench Pro, ContextWeaver raises the resolve rate from 66.5% to 72.6%, with significant improvements across all three model backbones. The improvement also holds when the sliding window's budget is raised to cover the summary cost.

cs.CL↗

Budgeted KV-Cache Repair for Evolving Contexts

Language-model agents revise their working context as information becomes obsolete, but deleting text does not erase its influence on cached computation. Later key-value (KV) states remain conditioned on the deleted content, while recomputing the entire retained suffix can be costly. We introduce budgeted KV-cache repair: recovering task performance by recomputing a user-chosen fraction of the retained suffix. Comparing repair allocations under identical relevance scores and token budgets, we find that refreshing coherent regions can substantially outperform refreshing isolated tokens. This finding motivates Semantic-Unit Refresh (SUR), which allocates the budget to query-relevant text regions using inexpensive embedding scores, without training or an auxiliary transformer forward. To execute these repairs efficiently, a layer-packed executor processes selected states together at each layer while preserving full causal cache access. Across five models, SUR achieves strong recovery in controlled deletion tasks while recomputing only 2-5% of the suffix. On four natural multi-hop QA datasets, it achieves near-full-recompute exact match at a 5% budget with lower post-edit latency. The executor accelerates fragmented refresh by up to 16.1x over interval-wise execution.

cs.AI↗

Generalized Space Time Fractional Skellam Process

This paper introduces the Generalized Space-Time Fractional Skellam Process (GSTFSP) and the Generalized Space Fractional Skellam Process (GSFSP). We investigate their distributional properties including the probability generating function (p.g.f.), probability mass function (p.m.f.), fractional moments, mean, variance, and covariance. The governing state differential equations for these processes are derived, and their increment processes are examined. We establish recurrence relations for the state probabilities of GSFSP and related processes. Furthermore, we obtain the transition probabilities, $n^{th}$-arrival times, and first passage times of these processes. The asymptotic behavior of the tail probabilities is analyzed, and limiting distributions as well as infinite divisibility of GSTFSP and GSFSP are studied. We provide the weighted sum representations for these processes and derive their characterizations. Also, we establish the martingale characterization for GSTFSP, GSFSP and related processes. In addition, we introduce the running average processes of GSFSP and its special cases, and obtain their compound Poisson representations. Finally, the p.m.f. of GSTFSP and simulated sample paths for GSTFSP and GSFSP are plotted.

math.PR↗

Dispersive-phonon-driven room-temperature Ni1+-Ni2+ polaron hopping in spin-charge coupled rutile niobate

Understanding how lattice dynamics mediate polaron hopping is essential for designing multifunctional correlated oxides. Here, we demonstrate room-temperature dispersive phonon excitations and elucidate the Ni1+-Ni2+ polaron-hopping mechanism and the presence of rare spin-charge-phonon coupling even in a magnetically short-range-ordered state in rutile niobate, a rare room-temperature magnetodielectric system. We reveal room-temperature dispersive phonon excitations using inelastic neutron scattering (INS), complemented by machine-learning-based phonon calculations, to establish the microscopic origin of the polaron-hopping mechanism. Experimental evidence of dispersive phonon-driven polaron hopping is scarce. INS measurements show significant dispersive phonon excitations at 21, 33, and 47 meV, implying collective lattice dynamics that enable delocalized polaron propagation via coupled charge-spin-phonon interactions. Dispersive phonons couple to charge carriers and promote correlated NiO6 lattice distortions, facilitating delocalized polaron hopping. Low-energy magnetic excitations at 4 and 8 meV indicate the presence of local short-range magnetic correlations or spin-orbit-coupling-induced anisotropy in deformed NiO6 octahedra, which are thoroughly discussed. Machine-learning phonon calculations replicate the experimentally observed phonon excitations and demonstrate lattice instability, which is compatible with dynamic local distortions caused by polaron production. These findings provide microscopic evidence for a coupled charge-spin-phonon mechanism that mediates polaron hopping in rutile oxide systems.

cond-mat.mtrl-sci↗

Basis Adaptive Algorithm for Quantum Many-Body Systems on Quantum Computers

We introduce a Basis Adaptive (BA) algorithm for hybrid quantum-classical simulation of correlated quantum many-body systems. Starting from a small set of physically motivated bitstrings, the algorithm iteratively applies a single-step first-order Trotterized circuit on a quantum processor, filters the sampled configurations by enforcing $U(1)$ spin conservation and lattice reflection symmetry, and classically diagonalizes the Hamiltonian in the resulting reduced Hilbert space. This design avoids the variational optimization overhead of VQE, the deep coherent circuits required by QPE, and the symmetry-violating subspaces that arise in SKQD. The ground-state energy error is bounded analytically by $\sqrt{8}\,\|H\|\left(1-\sqrt{α_{D_T}}\right)^{1/2}$, where $α_{D_T}$ is the probability weight captured by the $D_T$ sampled basis states. This bound connects algorithm performance directly to ground-state sparsity and explains the observed accuracy hierarchy across different phases. Benchmarked on the spin-$1/2$ Heisenberg XXZ chain (up to $N=62$ qubits on the IBM Heron processor), the algorithm achieves a $3.5\%$ energy error in the gapped Neel phase ($Δ=2.0$) and below $0.5\%$ at the ferromagnetic boundary ($Δ=-1.0$). The accuracy degrades to $28.7\%$ in the strongly quasi-long-range-ordered regime ($Δ=0.5$). Spin-spin correlation functions are reproduced across all regimes, confirming that symmetry-filtered real-time sampling provides a practical and noise-resilient pathway to ground-state properties on near-term quantum hardware.

cond-mat.str-el↗

Spin-Chain Incipient Magnetocaloric Effect and Rare-Earth Controlled Switching in the Haldane-Chain System, R2BaNiO5

We have experimentally investigated the magnetocaloric effect (MCE) of a prototype spin-frustrated one-dimensional spin-chain system, the famous Haldane-chain system, R2BaNiO5 (R = Nd, Gd, Er, Dy). The significant MCE is observed far above long-range ordering, even in the paramagnetic region, which is attributed to the change in magnetic entropy due to short-range spin correlation arising from (low-dimensional) magnetic frustration. Such a spin-chain incipient MCE above long-range ordering is rarely reported. Interestingly, multiple magnetocaloric switching from conventional to inverse MCE (and vice versa) are observed below long-range magnetic ordering, as a function of temperature and magnetic field, for the R = Nd, Dy, and Er members. However, such MCE switching is absent in the Gd member, which is an S-state atom (orbital moment L = 0). Our systematic investigation of this series demonstrates that the interplay between crystal-electric field (CEF), strong spin-orbit coupling (SOC) and rare earth anisotropy of R-ions play an important role in spin reorientation, leading to multiple MCE switching due to intriguing changes in magnetic and lattice entropy. The maximum change of entropy for Er, Gd, Dy and Nd is 7.8, 6.8, 4.0 and 1.0 J Kg-1 K-1 respectively. Our study presents a pathway for tuning MCE switching and the MCE effect over large temperature regions in d-f coupled spin-frustrated and spin-chain oxide systems.

cond-mat.str-el↗

State Localization and Selective Charge Filtering Near a Null Point

Null points in synthetically tunable molecular aggregates are predicted to generate flat energy bands analogous to those known in strongly correlated condensed-matter physics. For chemistry, null points provide a powerful design principle for photovoltaic materials with selective charge filtering similar to photosynthesis. However, null points have never been experimentally verified because their defining prediction - state localization with selective electron or hole transfer - has remained unobserved. Here, using a donor-acceptor dyad as a minimal model, we provide the first experimental observation of a null point. Impulsive pump-probe measurements reveal charge separation through a near-instantaneously generated locally excited-charge transfer (LE-CT) intermediate that emerges upon solvent stabilization of CT states. Polarization anisotropy directly reveals state localization and selective charge-filtering, spanning balanced electron-hole transfer to selective hole filtering consistent with synthetic design. A generalized vibronic theory of null points explains these observations and identifies the ideal synthetic parameters for achieving null points which are protected from the vibrational bath.

physics.chem-ph↗

Quasiparticle Dynamics in the 4d-4f Ising-like Double Perovskite Ba2DyRuO6 studied using Neutron Scattering and Machine-Learning Framework

Double perovskites containing 4d--4f interactions provide a platform to study complex magnetic phenomena in correlated systems. Here, we investigate the magnetic ground state and quasiparticle excitations of the fascinating double perovskite system, Ba$_2$DyRuO$_6$, through Time of flight (TOF) neutron diffraction (TOF), inelastic neutron scattering (INS), and theoretical modelling. The compound Ba$_2$DyRuO$_6$ is reported to exhibit a single magnetic transition, in sharp contrast to most of the other rare-earth (R) members in this family, A$_2$RRuO$_6$ (A = Ca/Sr/Ba), which typically show magnetic ordering of the Ru ions, followed by R-ion ordering. Our neutron diffraction results confirm that long-range antiferromagnetic order emerges at $T_\mathrm{N} \approx 47$~K, primarily driven by 4d--4f Ru$^{5+}$--Dy$^{3+}$ exchange interactions, where both Dy and Ru moments start to order simultaneously. The ordered ground state is a collinear antiferromagnet with Ising character, carrying ordered moments of $μ_{\mathrm{Ru}} = 1.6(1)~μ_\mathrm{B}$ and $μ_{\mathrm{Dy}} = 5.1(1)~μ_\mathrm{B}$ at 1.5~K. Low-temperature INS reveals well-defined magnon excitations below 10~meV. SpinW modelling of the INS spectra evidences complex exchange interactions and the presence of magnetic anisotropy, which governs the Ising ground state and accounts for the observed magnon spectrum. Combined INS and Raman spectroscopy reveal crystal-electric-field (CEF) excitations of Dy$^{3+}$ at 46.5 and 71.8~meV in the paramagnetic region. The observed CEF levels are reproduced by point-charge calculations consistent with the $O_h$ symmetry of Dy$^{3+}$. A complementary machine-learning approach is used to analyse the phonon spectrum and compare with INS data. Together, these results clarify the origin of phonon and magnon excitations and their role in the ground-state magnetism of Ba$_2$DyRuO$_6$.

cond-mat.str-el↗

Prominent Signatures of Energy Transfer in Action-Detected Spectra of a Cyanobacterial Photosynthetic Protein

Action-detected two-dimensional electronic spectroscopy (A-2DES) could potentially be a versatile chemical tool with applicability across a range of photophysical observables such as photocurrent, photoionization, or fluorescence. However, a prominent absence of excited state energy/charge transfer dynamics signals in archetypal photosynthetic proteins has suggested severe limitations of A-2DES in probing large aggregates where sensitivity to excited state dynamics is proposed to go down as 1/N, where N is the aggregate size. We report measurements of energy transfer dynamics in a cyanobacterial protein through both conventional and fluorescence 2DES (F-2DES), where the dynamics reported by F-2DES is quite prominent and comparable to that measured by conventional 2DES. Analysis of our experiments combined with coarse-grained simulations of the spectra suggest that the 1/N limit argument, which assumes infinitely fast intra-exciton manifold equilibration, is modified in case of cyanobacterial proteins because of slow annihilation. Our results suggest that action detection may in fact be well-suited to probe exciton diffusion across weakly coupled systems.

physics.chem-ph↗

Non-Vanishing of Cubic Twists of $GL_n(\mathbb{Q})$ $L$-functions

Let $π$ be an irreducible, cuspidal automorphic representation of $GL_n(\mathbb{A}_\mathbb{Q})$ ($n\geq 3$), which is tempered only for $n=3$. Let $s$ be a complex number such that $\Re(s)\notin \left[1/n, 1-1/n\right]$ if $n\neq 4$; $\Re(s)\notin\left[1/5, 4/5\right]$ if $n=4$, then we show that there are infinitely many primitive cubic Dirichlet characters $χ$ such that $L(s,π\times χ)\neq 0$. Similar results were previously known only for primitive Dirichlet characters without any restriction on the order and quadratic Dirichlet characters.

math.NT↗

Intriguing Magnetocaloric Effect in Multiferroic Ba3RRu2O9 (R=Ho, Gd, Tb, Nd) with Strong 4d-4f Correlations

Here we demonstrate the magnetocaloric effect (MCE) of a 4d-4f correlated system, namely Ba3RRu2O9 (R= Ho, Gd, Tb, Nd). The compound Ba3HoRu2O9 antiferromagnetically orders at 50 K where both the Ho and Ru-moments order, followed by another phase transition ~ 10 K. Whereas, the compound Ba3GdRu2O9 and Ba3TbRu2O9 orders at 14.5 and 10.5 K respectively, where the ordering of both R and Ru moments are speculated. Our results reveal robust MCE around low-T magnetic phase transition for all the heavy rare-earth members (Ho, Gd, Tb) in this family. The heavy rare-earth members exhibit an intriguing MCE behavior switching from conventional to non-conventional MCE. Interestingly, the light R-member, Ba3NdRu2O9, orders ferromagnetically below 24 K where Nd-moments order, followed by Ru-ordering below 18 K, exhibits a positive MCE below and above FM-ordering. The compelling MCE are attributed to temperature dependent complex spin-reorientations for different R-members and anisotropy.

cond-mat.str-el↗

NOMAD: Generating Embeddings for Massive Distributed Graphs

Successful machine learning on graphs or networks requires embeddings that not only represent nodes and edges as low-dimensional vectors but also preserve the graph structure. Established methods for generating embeddings require flexible exploration of the entire graph through repeated use of random walks that capture graph structure with samples of nodes and edges. These methods create scalability challenges for massive graphs with millions-to-billions of edges because single-node solutions have inadequate memory and processing capabilities. We present NOMAD, a distributed-memory graph embedding framework using the Message Passing Interface (MPI) for distributed graphs. NOMAD implements proximity-based models proposed in the widely popular LINE (Large-scale Information Network Embedding) algorithm. We propose several practical trade-offs to improve the scalability and communication overheads confronted by irregular and distributed graph embedding methods, catering to massive-scale graphs arising in web and science domains. NOMAD demonstrates median speedups of 10/100x on CPU-based NERSC Perlmutter cluster relative to the popular reference implementations of multi-threaded LINE and node2vec, 35-76x over distributed PBG, and competitive embedding quality relative to LINE, node2vec, and GraphVite, while yielding 12-370x end-to-end speedups on real-world graphs.

cs.LG↗