arXiv subjects

Shi Pu

Publications and source records attributed to Shi Pu.

At least 19 recordsLinked to original sources

arXiv2026

Probing the tetrahedral $α$ clusters in relativistic $^{16}$O + $^{16}$O collisions

Relativistic $^{16}$O +$^{16}$O collisions provide a valuable opportunity to study both the quark-gluon plasma formed in small systems and the intrinsic structure of $^{16}$O. Recent implementations of \textit{ab initio} nuclear configurations in heavy-ion simulations have produced different predictions for cluster-sensitive observables, making it difficult to identify the origin of possible $α$-cluster signals. In this work, we introduce a controlled sampling scheme that varies the compactness of $α$-cluster-induced multi-nucleon correlations while keeping the one-body density distribution of $^{16}$O fixed. This allows us to separate effects driven by the tetrahedral one-body density from those driven by genuine multi-nucleon correlations. We show that the normalized ratios ${\rm Norm}(v_{2}\{2\}/v_{2}\{4\})$ and ${\rm Norm}(v_{2}\{2\}/v_{3}\{2\})$, together with their initial-state eccentricity counterparts, provide complementary constraints on initial-condition model dependence and cluster compactness. Hybrid hydrodynamic simulations and comparisons with recent LHC measurements further clarify the extent to which the initial-state signals survive final-state evolution. Our results provide a framework for using relativistic light-ion collisions to constrain cluster correlations in $^{16}$O.

nucl-th↗

arXiv2026

Optimal Network Dependence in Distributed Stochastic Optimization via Tree Routing

Communication is a central bottleneck in distributed optimization, but its effect is often summarized by the spectral gap of a chosen mixing matrix. Since this gap depends on link weights as well as topology, it can obscure the intrinsic effect of the network. We clarify this distinction by relating graph diameter to the inverse spectral gap. We establish a universal one-sided bound that is tight up to constant factors, while constructions across several graph families show that no topology-only converse or universal two-sided scaling law exists. This motivates a class-based view of network optimality: diameter-dependent guarantees can be assessed uniformly over graph classes, whereas spectral-gap optimality requires both the graph--matrix pair class and the admissible algorithm class to be specified. Based on the resulting diameter-based minimax benchmark, we introduce Tree-Routed Gradient Tracking (Tree-RGT), a distributed method for stochastic nonconvex optimization. The method pipelines model dissemination and gradient aggregation over a rooted shortest-path spanning tree. Each iteration uses one round of one-hop communication and one stochastic-gradient evaluation per agent, without inner consensus or multi-gossip. Under smoothness and unbiased bounded-variance oracles, Tree-RGT achieves uniformly optimal network dependence over connected graphs of prescribed size and diameter. It recovers centralized stochastic scaling after $\mathcal{O}(nD_{\mathcal{G}}^2)$ transient iterations, where $n$ is the number of agents and $D_{\mathcal{G}}$ is the graph diameter. This network-dependent transient-iteration bound improves upon or matches those reported for representative distributed methods. Thus, topology-aware routing attains diameter-optimal guarantees without relying on a prescribed mixing matrix.

math.OC↗

arXiv2026

Local Spin Polarization in Anisotropic Gubser Flow: Suppression Mechanism and Formulation Dependence

We analytically study the longitudinal spin polarization in relativistic heavy-ion collisions using a perturbed Gubser flow solution. In the large-system-size limit, we derive analytical expression of the local spin polarization along the beam direction. In our treatment, the contributions from thermal vorticity and thermal shear are of comparable magnitudes. The thermal vorticity yields the polarization with a sign opposite to that observed experimentally, while the thermal shear counteracts this effect, helping recover the desired sign. We find that the choice of the reference unit vector aligned with the fluid velocity gives the experimentally observed sign only at low transverse momenta, whereas another formulation with the unit vector fixed along the laboratory time direction yields the desired sign for a wide range of transverse momenta. Notably, a recent formulation with the unit vector normal to the freeze-out hypersurface exhibits an exact cancellation between contributions from thermal vorticity and thermal shear at leading order in the large-system-size limit. We identify a general cancellation pattern with acceleration dominance, which is manifest particularly in the latter two formulations. Thus, the total polarization originates from non-acceleration effects, which need not be substantial even when the elliptic flow is finite, as clearly demonstrated in our analytical results. For comparison, we also discuss the spin polarization in the Hubble flow with rotation.

nucl-th↗

arXiv2026

CLVisc Agent for autonomous relativistic hydrodynamics studies

We enable large language model (LLM) agents to autonomously perform end-to-end hydrodynamic simulations of the quark-gluon plasma evolution and calculation of final hadron spectra in relativistic heavy-ion collisions. We design a meta skill that allows an agent to explore a project's source code, craft a specialized skill, and iteratively refine it. Applying this meta skill to the (3+1)D viscous hydrodynamic code CLVisc, the agent builds a CLVisc skill encoding its operational knowledge and then independently executes full scientific workflows: designing parameter scans, running simulations, comparing ensemble results, and producing publication-ready figures. Crucially, the agent draws on literature-informed heavy-ion physics to select physically meaningful observables and interpret outcomes without explicit instruction. We demonstrate the pipeline in two scenarios: temperature-dependent shear viscosity over entropy density $η/s$, and nuclear-structure effects in O+O collisions at $\sqrt{s_{\mathrm{NN}}} = 5.36$~TeV using four \textit{ab initio} descriptions of $^{16}$O. In both, the agent plans, executes, and analyzes autonomously, devising new initial-state observables to explain final observations and extract qualitative knowledge. The meta skill is agnostic to code versions and Monte Carlo generators, promising future multi-agent systems in high-energy nuclear physics.

nucl-th↗

arXiv2026

In-in worldline formalism in pair creating fields

An in-in framework under Schwinger pair creating fields in strong-field quantum electrodynamics is formulated using in-out propagators in coordinate space, that have first-quantized or worldline representation. The framework is derived to all orders in the background field coupling from both the Bogoliubov coefficient method and Schwinger-Keldysh closed-time path formalism. In-out matrix elements in pair creating fields are readily handled using first-quantized methods, and the approach we develop serves to facilitate the evaluation of in-in observables in pair creating backgrounds. We find that in-in augmentations to the in-out partition function and or propagator amount to the insertion of a non-local interaction term that sandwiches a function that receives contributions from singularities and critical points in complex Schwinger propertime. Furthermore, we show the resummation of the in-in partition function leading to vacuum non-persistence that en-route gives an exact first-quantized definition of creating $N$-pairs.

hep-th↗

arXiv2026

Global polarization of $Λ$, $Ξ^{-}$, and $Ω^{-}$ hyperons in Au+Au collisions at RHIC BES-II energies

We investigate the global spin polarization of $Λ$ hyperons and the multi-strange hyperons $Ξ^{-}$ and $Ω^{-}$ in Au+Au collisions across the RHIC Beam Energy Scan II (BES-II) energy range, $\sqrt{s_{NN}}=7.7$--$27$ GeV. The polarization is computed using the modified Cooper--Frye formula, which includes contributions from thermal vorticity, the thermal shear tensor, and the gradient of the baryon chemical potential, combined with the (3+1)-dimensional viscous hydrodynamic framework CLVisc with SMASH initial conditions. We present the global polarization as a function of collision energy, centrality, transverse momentum, and rapidity. We find that the global polarization of $Ω^{-}$ is systematically larger than those of $Λ$ and $Ξ^{-}$ because of its larger spin quantum number, but it remains below the central value of the recent STAR measurement. This discrepancy may suggest that additional mechanisms, such as spin correlations among strange quarks inside the $Ω^{-}$, could contribute to the observed $Ω^{-}$ polarization. We also find that the global-polarization splitting between hyperons and anti-hyperons increases toward lower collision energies and is dominated by the chemical-potential-gradient contribution.

nucl-th↗

arXiv2026

Achieving Linear Speedup for Composite Federated Learning

This paper proposes FedNMap, a normal map-based method for composite federated learning, where the objective consists of a smooth loss and a possibly nonsmooth regularizer. FedNMap leverages a normal map-based update scheme to handle the nonsmooth term and incorporates a local correction strategy to mitigate the impact of data heterogeneity across clients. Under standard assumptions, including smooth local losses, weak convexity of the regularizer, and bounded stochastic gradient variance, FedNMap achieves linear speedup with respect to both the number of clients and the number of local updates for nonconvex losses, both with and without the Polyak-Łojasiewicz condition. To the best of our knowledge, this is the first algorithm establishing linear speedup for nonconvex composite federated learning. Numerical experiments corroborate our theoretical findings and demonstrate the linear speedup of FedNMap.

cs.LG↗

arXiv2026

Late-Time Relaxation from Landau Singularities

Nonlinear hydrodynamic interactions can change the relaxation of fluctuations from exponential to power-law decay at late times. Schwinger-Keldysh effective field theory provides a standard framework for describing such fluctuation effects, where the nonlinear late-time behavior is encoded in loop corrections. Extracting this behavior requires identifying the singularities of loop integrals, whose structure becomes increasingly intricate beyond simple models. We apply Landau singularity analysis to two-point functions in effective field theories and determine the singularities induced by nonlinear interactions without performing the loop integrations explicitly. From these frequency-space singularities, we extract nonlinear relaxation modes that control the late-time behavior. When gapless modes are present, these modes produce power-law decay at late times. Our results give a systematic singularity-based description of nonlinear late-time relaxation in a broad class of macroscopic effective theories.

hep-th↗

arXiv2026

Distributed Normal Map-based Stochastic Proximal Gradient Methods over Networks

Consider $n$ agents connected over a network collaborating to minimize the average of their local cost functions combined with a common nonsmooth function. This paper introduces a unified algorithmic framework for solving such a problem through distributed stochastic proximal gradient methods, leveraging the normal map update scheme. Within this framework, we propose two new algorithms, termed Normal Map-based Distributed Stochastic Gradient Tracking (norM-DSGT) and Normal Map-based Exact Diffusion (norM-ED). We demonstrate that both methods can asymptotically achieve comparable convergence rates to the centralized stochastic proximal gradient descent method under a general variance condition on the stochastic gradients. Additionally, the number of iterations required for norM-ED to achieve such a rate (i.e., the transient time) behaves as $\mathcal{O}(n^{3}/(1-λ)^2)$ for minimizing composite objective functions, matching the performance of the non-proximal ED algorithm. Here $1-λ$ denotes the spectral gap of the mixing matrix related to the underlying network topology. To our knowledge, such a convergence result is state-of-the-art for the considered composite problem. Under the same condition, norM-DSGT enjoys a transient time of $\mathcal{O}(\max\{n^3/(1-λ)^2, n/(1-λ)^3\})$, which matches that of the non-proximal DSGT algorithm and norM-ED under the condition $(1-λ)^{-1}=\mathcal{O}(n^{2})$.

math.OC↗

arXiv2026

Weyl anomaly induced transport in hydrodynamics

We show that the Weyl (trace) anomaly gives rise to a new non-dissipative vector current in accelerated relativistic fluids. The anomaly uniquely fixes the second-order transport coefficient governing the coupling between the electromagnetic field and the fluid acceleration. We derive this result by extending hydrodynamic anomaly matching to include the trace anomaly, and independently reproduce it in boundary quantum field theory by treating the Rindler horizon of an accelerated observer as an effective boundary. From the boundary perspective, the electric- and magnetic-field sectors correspond to screening and vacuum magnetization effects near the boundary. In the local rest frame, the electric-field contribution induces an additional charge density, while the magnetic-field contribution generates a transverse current with a Nernst-like, more generally thermomagnetic Hall-like, tensor structure. Our results reveal a new class of anomaly-induced transport governed by the trace anomaly.

hep-th↗

arXiv2026

Nonlinear response of flow harmonics in Gubser flow with participant-reaction planes mismatch

We investigate the nonlinear response of flow harmonics $v_2,v_4$ to initial-state eccentricities $ε_2,ε_4$ within the Gubser-flow framework. By extending the perturbative solutions of Gubser flow, we derive analytic nonlinear response relations connecting the eccentricities $ε_2,ε_4$ to the flow harmonics $v_2,v_4$. Our results reproduce the well-known result $v_4/v_2^2 \to 1/2$ in large transverse momentum $p_T$ limit. Furthermore, we study the effects of a mismatch between the participant and reaction planes. We find that the conventional nonlinear response coefficients acquire an additional factor determined by the participant-plane angles, which is often approximated as statistical noise driven by event-by-event fluctuations. This factor can modify both the strength but even the sign of the effective nonlinear response coefficient, making it sensitive to the initial configuration of the colliding nuclei. Our study provides new analytical insight into the origin of collective phenomena in relativistic heavy-ion collisions.

nucl-th↗

arXiv2026

Global polarization of $Λ$ hyperons and its sensitivity to equations of state in low-energy heavy-ion collisions

Significant global polarization of $Λ$ hyperons along the direction of the orbital angular momentum has been measured in non-central heavy-ion collisions where the equation of state (EOS) of the produced dense matter is expected to change from intermediate to low colliding energies. We study the sensitivity of the global $Λ$ polarization to EOS in heavy-ion collisions within the SMASH transport model. Among the three different EOS we considered, only the hadron resonance gas (HRG) describes the experimental data well at low colliding energies even when it is below the $Λ$ production threshold in nucleon-nucleon collisions. The polarization induced by thermal vorticity as a function of centrality, rapidity, and transverse momentum at $\sqrt{s_{NN}} = 3$ GeV in Au+Au collisions is shown to agree well with the experimental data. Our study also indicates a possible peak in the global $Λ$ polarization around $\sqrt{s_{NN}} = 2.4$ GeV in Au+Au collisions. Furthermore, we find that the rapidity and transverse momentum-dependent helicity polarization induced by thermal vorticity vanishes due to space-reversal symmetry.

nucl-th↗

arXiv2026

Late-time attractors in relativistic spin hydrodynamics in Gubser flow

We investigate the late-time asymptotic solutions and attractor structure of the spin density in minimal causal spin hydrodynamics in Gubser flow. After deriving the differential equation governing the spin density, we obtain its late-time asymptotic solutions and identify both attractors and repellers in the corresponding numerical solutions. We then map these solutions back to flat Minkowski space and find parameter regions where the spin density exhibits a power-law decay. We further show that, when the characteristic length scale of the system is much larger than the proper time, several components of the spin density can decay as slowly as conventional thermodynamic variables in relativistic hydrodynamics. In this regime, the spin density behaves as a hydrodynamic mode governed by the late-time scaling laws of the flow.

hep-ph↗

arXiv2026

Accelerating Decentralized Optimization via Overlapping Local Steps

Decentralized optimization has emerged as a critical paradigm for distributed learning, enabling scalable training while preserving data privacy through peer-to-peer collaboration. However, existing methods often suffer from communication bottlenecks due to frequent synchronization between nodes. We present Overlapping Local Decentralized SGD (OLDSGD), a novel approach to accelerate decentralized training by computation-communication overlapping, significantly reducing network idle time. With a deliberately designed update, OLDSGD preserves the same average update as Local SGD while avoiding communication-induced stalls. Theoretically, we establish non-asymptotic convergence rates for smooth non-convex objectives, showing that OLDSGD retains the same iteration complexity as standard Local Decentralized SGD while improving per-iteration runtime. Empirical results demonstrate OLDSGD's consistent improvements in wall-clock time convergence under different levels of communication delays. With minimal modifications to existing frameworks, OLDSGD offers a practical solution for faster decentralized learning without sacrificing theoretical guarantees.

cs.LG↗

arXiv2025

FedSUM Family: Efficient Federated Learning Methods under Arbitrary Client Participation

Federated Learning (FL) methods are often designed for specific client participation patterns, limiting their applicability in practical deployments. We introduce the FedSUM family of algorithms, which supports arbitrary client participation without additional assumptions on data heterogeneity. Our framework models participation variability with two delay metrics, the maximum delay $τ_{\max}$ and the average delay $τ_{\text{avg}}$. The FedSUM family comprises three variants: FedSUM-B (basic version), FedSUM (standard version), and FedSUM-CR (communication-reduced version). We provide unified convergence guarantees demonstrating the effectiveness of our approach across diverse participation patterns, thereby broadening the applicability of FL in real-world scenarios.

cs.LG↗

arXiv2025

Distributed Stochastic Momentum Tracking with Local Updates: Achieving Optimal Communication and Iteration Complexities

We propose Local Momentum Tracking (LMT), a novel distributed stochastic gradient method for solving distributed optimization problems over networks. To reduce communication overhead, LMT enables each agent to perform multiple local updates between consecutive communication rounds. Specifically, LMT integrates local updates with the momentum tracking strategy and the Loopless Chebyshev Acceleration (LCA) technique. We demonstrate that LMT achieves linear speedup with respect to the number of local updates as well as the number of agents for minimizing smooth objective functions with and without the Polyak-Łojasiewicz (PL) condition. Notably, with sufficiently many local updates $Q\geq Q^*$, LMT attains the optimal communication complexity. For a moderate number of local updates $Q\in[1,Q^*]$, LMT achieves the optimal iteration complexity. To our knowledge, LMT is the first distributed stochastic gradient method with local updates that enjoys such properties.

math.OC↗

arXiv2025

Local spin polarization of $Λ$ hyperons and its interaction corrections

We have computed the second Fourier sine coefficient of the longitudinal spin polarization, $\langle P_{z} \sin 2(ϕ_{p} - Ψ_{2}) \rangle$, as a function of multiplicity or centrality in Au+Au collisions at $\sqrt{s_{NN}} = 200$ GeV and in $p$+Pb collisions at $\sqrt{s_{NN}} = 8.16$ TeV using the CLVisc hydrodynamic framework. The numerical results successfully describe the data in Au+Au collisions. However, understanding the data in $p$+Pb collisions remains a puzzle. Additionally, we have reported some recent developments in quantum kinetic theory and spin hydrodynamics.

nucl-th↗

arXiv2025

Hydrodynamic effects on spin polarization along the beam direction in Au+Au and p+Pb collisions

We investigate hydrodynamic effects on the spin polarization of $Λ$ hyperons in Au+Au collisions at $\sqrt{s_{NN}} = 200$ GeV and p+Pb collisions at $\sqrt{s_{NN}} = 8.16$ TeV using the CLVisc hydrodynamic framework. We present numerical results for the second Fourier sine coefficient of the longitudinal spin polarization, $\langle P_{z} \sin 2(ϕ_{p} - Ψ_{2}) \rangle$, as a function of multiplicity (centrality) under three equilibrium scenarios: $Λ$ equilibrium, $s$-quark equilibrium, and isothermal equilibrium. We highlight the respective roles of thermal vorticity and the thermal-shear tensor in generating $\langle P_{z} \sin 2(ϕ_{p} - Ψ_{2}) \rangle$ across collision systems and scenarios.

nucl-th↗