arXiv subjects

Song Liu

Publications and source records attributed to Song Liu.

At least 19 recordsLinked to original sources

arXiv2026

Magnons in α-Fe2O3 microdisks: from thermal to coherent excitation

Hematite is a prototypical antiferromagnetic insulator for spintronic and magnonic applications, yet controlled studies of coherent magnons in thin films remain challenging due to strain sensitivity and spatial inhomogeneity. Here, we investigate magnon excitations in lithographically patterned hematite microdisks, and obtain reproducible magnon spectra from multiple individual microdisks. Brillouin light scattering reveals bulk-like field-dependent thermal magnon resonances with an ultralow effective damping parameter on the order of ~10-5. Coherent magnons are further excited using a stripline antenna adjacent to the microdisk, revealing a power-dependent frequency blue shift accompanied by a nonlinear evolution of the coherent response under strong microwave driving. Field-dependent analysis indicates a power-dependent increase of the effective magnetic anisotropy field, suggesting a tunable nonlinear response of antiferromagnetic magnons in laterally confined microdisks. These results establish patterned hematite thin films as a platform for exploring coherent and nonlinear antiferromagnetic magnon dynamics.

physics.app-ph↗

arXiv2026

InterQ-RL: Interaction-Aware Correction of Execution Mismatch in Quantized Agentic RL

Multi-turn agent reinforcement learning (RL) relies on long closed-loop rollouts, making generation and KV-cache costs a major bottleneck. Quantized rollout reduces these costs, but its low-precision executor also serves as the behavior policy and can induce execution mismatch with the learner. Existing quantized-RL methods are designed primarily for single-turn generation. They align the learner forward with quantized rollout or correct the current-token rollout-learner ratio, without modeling the policy history that produces each agent state. Even after quantization alignment, residual rollout-learner mismatch accumulates along the state-reaching policy history, so a current action may look locally aligned despite a mismatched history. We introduce InterQ-RL (Interaction-aware Quantized Reinforcement Learning) to correct this post-alignment mismatch. It applies bounded execution-prefix correction over the state-reaching policy history, allocates correction strength across action segments by effective support under a fixed displacement budget, and requires no additional model forward pass. We evaluate dense and mixture-of-experts (MoE) backbones from 4B to 30B on ALFWorld, WebShop, and ScienceWorld, with all quantized arms sharing the same rollout with 8-bit weights, activations, and KV cache (W8A8-KV8). Across four environment-backbone settings, InterQ-RL achieves the strongest quantized endpoint result, matching BF16 to within 0.7 points and exceeding it in two settings. It matches or exceeds the learning speed of alignment alone while avoiding its late performance regression.

cs.LG↗

arXiv2026

Blow-up, critical norm inflation, and global well-posedness for the Geng--Xue system

We study finite-time blow-up, critical norm inflation, and global well-posedness for the Geng--Xue system, an integrable two-component extension of the Novikov equation. For initial velocities $(u_0,v_0)\in(H^s)^2$, $s>3/2$, with second momentum $n_0=v_0-v_{0,xx}\in L^1_{\rm uloc}$, we derive explicit blow-up conditions and upper bounds on the lifespan. The blow-up proof uses a cancellation in the equation for $nu_x$ to close the convolution estimates without an a priori bound on $\|u_x\|_{L^\infty}$. Combined with a continuation criterion involving only $u$, these estimates yield critical Besov norm inflation around the initial data in $(H^{3/2})^2$. For initial momenta in $(H^σ)^2$, $σ>-1/2$, each with a fixed sign, we prove global well-posedness. The blow-up and norm-inflation results also extend to the three-component Degasperis--Procesi system.

math.AP↗

arXiv2026

Steady-state cooperative emission of interlayer excitons in atomically thin light-emitting diodes

When an ensemble of emitters is confined within an optical wavelength, its radiation can become correlated, leading to cooperative emission. In two-dimensional (2D) transition metal dichalcogenide (TMD) electron-hole bilayers, strong light-matter coupling enables the formation of interlayer excitons (IXs), long-lived dipolar excitations that provide an ideal setting for emergent collective phenomena. Here, we engineer an atomically thin light-emitting diode (LED) that generates electroluminescence (EL) from electrically injected IXs that are confined near the sample edges by an edge-enhanced fringing electric field. By measuring Hanbury-Brown-Twiss photon-photon correlations, we find pronounced signatures of collective decay in the form of large photon bunching followed by finite delay antibunching and bunching revivals. These emitter correlation signatures disappear when free charge carriers are doped to form a degenerate Fermi sea but are enhanced by the introduction of localized electrons. The results are consistent with a many-body quantum master equation that models the steady-state superradiant photon statistics of incoherently injected IXs. Our work establishes steady-state cooperative emission in a 2D TMD LED as a platform for studying strong light-matter interactions in electrically driven, non-equilibrium phases of Bose-Fermi mixtures.

cond-mat.mes-hall↗

arXiv2026

Finite-time blow-up for the Geng--Xue system with an application to the three-component Degasperis--Procesi system

We prove finite-time blow-up for the Geng--Xue system with initial data satisfying $u_0\in W^{p,q}(\R)$ and $n_0\in L^{r'}(\R)$ at Sobolev regularity $s\geq2$, where $n=v-v_{xx}$. A cancellation in the balance law for $nu_x$ yields closed convolution estimates that do not depend on an a priori uniform bound for $u_x$. These estimates control $v$ along a characteristic and give an explicit condition on the initial slope, together with an upper bound on the lifespan. The proof does not use conservation laws. For the three-component Degasperis--Procesi system, the same argument applies to $nu_x+ρ^2-\barρ^2$, where $\barρ\in[0,1]$ is a constant density background. We also prove local well-posedness for this system in $H^{s-1}(\R)\times H^s(\R)\times H^s(\R)$ for $s>\frac32$, and a blow-up criterion in $B^1_{2,1}(\R)\times B^2_{2,1}(\R)\times B^2_{2,1}(\R)$.

math.AP↗

arXiv2026

Self-sufficient Independent Component Analysis for Demixing Flows

We study the problem of learning disentangled signals from data using non-linear Independent Component Analysis (ICA). Motivated by advances in self-supervised learning, we propose to learn self-sufficient signals: Given the remaining values of a recovered signal, observing other signals should not change the conditional distribution of its missing value. We formulate this problem as the minimization of a conditional KL divergence. Our algorithm is prior-free and likelihood-free in the sense that it prescribes neither parametric source densities nor an observation likelihood. To tackle the KL divergence minimization problem, we propose a sequential algorithm that learns a de-mixing flow model at each iteration, and prove local descent of the total correlation for its idealized Wasserstein-gradient-flow variant with exact velocities and a population projection condition. This approach completely avoids the unstable adversarial training, a common issue in minimizing the KL divergence. Experiments on toy and real-world datasets show the effectiveness of our method.

stat.ML↗

arXiv2026

Density Ratio Estimation with Stein Displacement Fields

Density ratios quantify distribution shift from a probability-mass point of view, whereas displacement fields describe, from a dynamical point of view, how one distribution is transported onto another. Although both offer complementary insights, they are usually estimated separately, and converting one into the other requires post-processing. In this paper, we estimate the density ratio between a target and a base distribution by parametrizing it through a displacement field acting on the base: the log-ratio is modeled as minus the Stein operator of the base applied to the field, up to a normalizing constant. This gives both statistical and dynamical descriptions of the distribution shift through a single convex optimization problem. Iterating this estimate-and-move step gives two inference algorithms: push-forward moves the model and corrects a pretrained sampler without retraining it, whereas pull-back moves the data closer to the base and fits a transformation model one layer at a time. Applications to distribution shift in simulation-based inference and to nonlinear independent component analysis illustrate the benefits and limitations of the approach.

stat.ML↗

arXiv2026

Controlling Dependence in Implicit Generative Models via Spread Mutual Information

Mutual information (MI) provides an objective for suppressing or encouraging statistical dependence in implicit generative models. However, direct MI evaluation is challenging in implicit models due to typically intractable densities. A remedy is estimating the generator gradient from the difference between conditional and marginal scores. This score difference can, in turn, be estimated by differentiating a log density ratio learned through classification. This construction nevertheless faces two difficulties: (i) singular distributions need not admit the required score functions, and (ii) poor overlap can hinder density-ratio estimation. We therefore introduce Spread Mutual Information (SMI), a weighted integral of MI across noise levels obtained by applying a common spreading kernel to the generated variable. Gaussian spreading yields smooth, strictly positive conditional and marginal densities, extending the gradient construction to distributions that may originally be singular. Across a variaty of experiments, SMI consistently achieves effective dependence control among MI-based methods and remains competitive with established task-specific approaches.

stat.ML↗

arXiv2026

Adaptive KV Cache Reuse for Fast Long-Context LLM Serving

In long-context Large Language Model (LLM) inference, the Time-To-First-Token (TTFT) latency incurred by the prefill stage has become the foremost bottleneck limiting interactive performance and deployment cost. KV Cache reuse offers a direct path to reduce redundant prefill, yet traditional prefix caching applies only to strict-prefix scenarios; directly reusing KV Cache in non-prefix settings breaks the cross-chunk global attention relationships and causes significant degradation in generation quality. When reusable KV Cache is offloaded to GPU-external cache pools, I/O overheads across heterogeneous hardware tiers further emerge as a new TTFT bottleneck. Efficient non-prefix KV Cache reuse therefore requires both semantic-consistency recovery and compute-I/O co-optimization. This paper presents CacheTune, a frequency-guided and hardware-aware KV Cache reuse system for long-context LLM serving. CacheTune first identifies, offline, the KV pairs most critical to cross-attention recovery through frequency-domain analysis, and then selectively recomputes only these semantic-critical tokens online while reusing the remaining KVs. To turn this semantic selection into end-to-end latency reduction, CacheTune further combines sparse KV transfer, multi-stream asynchronous overlap, deferred positional-encoding recovery, and hardware-aware adaptive recomputation-ratio tuning to balance computation and data movement across heterogeneous cache pools. Evaluations on mainstream LLMs and long-context tasks show that CacheTune achieves 3.72x-4.86x TTFT speedup and 3.93x-6.21x higher throughput while maintaining generation quality close to full recompute. Even when caches are offloaded to I/O-bound SSD/HDD storage, CacheTune sustains 2.34x-2.36x TTFT speedup through adaptive recomputation.

cs.AR↗

arXiv2026

UMR: Universal Manipulation Representation

General-purpose embodied manipulation hinges on a unified action representation that generalizes across embodiments and scales readily. Yet existing policies rely on embodiment-specific action spaces, making cross-embodiment demonstrations difficult to leverage at scale and limiting transfer to new embodiments and spatial variations. To this end, we introduce Universal Manipulation Representation (UMR), a unified action representation that enables zero-shot skill transfer from human demonstrations to heterogeneous robots. UMR decomposes manipulation into two functionally distinct yet geometrically linked components: embodiment-agnostic World Flow, which describes task-relevant object motion in the world frame, and Ego Trajectory, which represents end-effector motion relative to the current pose. We instantiate UMR as World--Ego Point VLA (WEPVLA), a compact 0.5B-parameter policy that learns in the unified geometric action space through a dual-stream Point Action Adapter and a unified Point Action Expert, with an $SE(3)$ conjugation coupling the two components. To improve data efficiency, we complement UMR with a Data-Efficient Strategy (DES) that diversifies object configurations through stage-aware point-cloud editing while preserving demonstrated contact geometry. In simulation, WEPVLA achieves average success rates of 97.5\% on LIBERO and 85.7\% on the 10-task RLBench benchmark. In real-world experiments, a single policy trained on human demonstrations augmented by DES transfers zero-shot to diverse deployment conditions. With about 10 minutes of collected human demonstrations per task and no robot demonstrations, it achieves 91.7\% average success across six evaluation settings, compared with 60.8\% for HumanEgo. Code and additional materials are available at https://umr-wepvla.github.io/.

cs.RO↗

arXiv2026

MagLearn 2: High-Fidelity, Saturation-Aware, and History-Efficient Sequence-to-sequence Modeling of Transient B-H Behaviour Under PWM Excitation

Accurate transient B-H modeling under pulsewidth-modulated (PWM) excitation is challenging since the magnetic-field response depends on both the instantaneous flux density and the preceding magnetisation trajectory. Although sequence-to-sequence models enable waveform prediction, existing studies have not jointly addressed how much magnetic history is required, how saturation region should be handled, and how trained models can be adapted to unseen materials. To address these gaps, this paper presents MagLearn 2, a condition-aware, short-window LSTM-based sequence-to-sequence framework for reconstructing the future H(t) response from the historical B(t) and H(t) trajectories. A saturation-aware auxiliary module selectively replaces saturation-dominated predictions within the complete waveform forecast. In a representative saturated 3C90 case, the auxiliary module reduces the sequence-level relative RMSE from 67.84% to 9.19%. To understand how much the history samples contributes to the predictions, an occlusion-based sensitivity analysis is conducted to quantifies the contribution of magnetic history and reveals that that the information most relevant to prediction is concentrated near the forecast boundary. In the independent evaluation for MagNet Challenge II, the proposed framework's best-performing model achieved the lowest reported average RMSE to date on the sequence-prediction task at 2.2%.

eess.SY↗

arXiv2026

Zero Flux: Flow-Based Comparison of High-Dimensional Discrete Distributions

Comparing two high-dimensional discrete distributions has always been a challenging task due to the exponentially growing state space and complex changes in interactions. A recent work suggests comparing distributions through a vector field trained using flow matching between two continuous distributions. The resulting vector field at mid-point vanishes if and only if two distributions identical. However, such a flow-based criterion does not naturally apply to discrete distributions. We extend this principle to the discrete domain and introduce the \emph{Zero Flux} criterion, a discrepancy based on local probability fluxes. Under independent coupling, we show that all local probability fluxes vanish at the midpoint if and only if two distributions are the same. This discrepancy decomposes the joint distributional difference into smaller, local contributions and can be efficiently estimated from samples. We establish finite sample error bounds for our estimator. Experiments on synthetic and real categorical data demonstrate reliable recovery of sparse dependence signals and stable tracking of distribution shifts in high dimensions.

stat.ML↗

arXiv2026

Zero-Flow Two-Sample Tests

Motivated by the success of modern flow-based generative models in modeling complex data, we study two-sample testing through the lens of flow-based methods. We propose the Zero-Flow Two-Sample Test (ZF2ST), built on the zero-flow criterion, which characterizes distributional equality through a time-reversal antisymmetry of a learnable velocity field. We extend this criterion and further develop the Zero-Flow Discrepancy, an identifying discrepancy that controls the Wasserstein distance, and derive a variational representation in terms of a witness function. This representation naturally leads to a witness-based test whose power is governed by the signal-to-noise ratio (SNR), allowing direct power maximization for witness learning. ZF2ST learns the witness on one data split and performs testing on held-out samples, thereby maintaining Type-I error control and admitting a simple asymptotic null distribution. Experimentally, ZF2ST performs competitively across various synthetic and real-world benchmarks, while showing particularly strong performance in distinguishing image distributions from different sources.

cs.LG↗

arXiv2026

Resource-Efficient Speculative Decoding for Long-Context LLM Serving

Speculative decoding reduces sequential Target model calls by verifying multiple tokens from the Draft model in parallel. Yet KV Cache growth limits long-context serving under constrained GPU memory. Offloading KV to CPU memory relieves this pressure. However, existing offloading schemes restore the full KV history before attention and fail to fully exploit the benefits of KV sharing across queries within a verification round. Existing parallel speculative decoding methods also overlook idle GPU compute capacity while memory-bandwidth-bound Target verification waits for historical KV. We present SpecStream, a speculative decoding system that begins verification without waiting for the full KV history to be restored and exploits compute bubbles during KV transfers for concurrent drafting on the same GPUs. It offloads only Target-committed history, keeping candidate rollback local to the GPUs. Each streamed KV chunk serves all queries in the round, with online softmax preserving full attention. Target-priority scheduling controls Draft concurrent execution under resource limits to constrain interference with Target verification. Experiments show that SpecStream maintains task quality close to SGLang speculative decoding while supporting more concurrent requests under limited GPU memory. Across different datasets, it achieves average throughput speedups of 1.41$\times$ and 1.32$\times$ over the offloading baseline for Qwen3 and InternLM2.5, respectively. Compared with parallel speculative decoding on separate Target and Draft GPUs, SpecStream improves output throughput per GPU by an average of 55.4%.

cs.AR↗

arXiv2026

Broadband Purcell Filter for Fast Superconducting Qubit Reset and Readout

Rapid reset and readout of qubit states are essential for quantum error correction, yet accelerating these operations through stronger coupling to a dissipative environment inevitably increases qubit decay via the Purcell effect. Here we present a broadband Purcell filter that decouples the reset and readout paths, enabling both operations to be independently optimized without compromising qubit coherence. The filter employs two engineered notches - an intrinsic notch and a bandstop notch - to provide broadband Purcell protection, together with an additional reset stub that creates a reset mode below the protected band. To enable fast reset while suppressing filter-mediated interactions between qubits, we couple each qubit to a dedicated reset resonator. We experimentally demonstrate Purcell-limited relaxation times exceeding 1 ms across a 1.2 GHz bandwidth, simultaneously with 500 ns readout without a Josephson parametric amplifier and 100 ns reset with 99.6% efficiency. The reset resonator is designed with a deliberate kappa-chi mismatch, which suppresses photon-shot-noise-induced dephasing by a factor of 70 compared to the readout resonator. Our work provides a scalable hardware solution that resolves the traditional trade-off between fast qubit operations and qubit protection, advancing the prospects for fault-tolerant quantum computing.

quant-ph↗

arXiv2026

SCOPE-OPSD: Fisher-Conditioned Privileged Subspaces for On-Policy Self-Distillation

On-policy self-distillation (OPSD) scores student-generated prefixes with a solution-conditioned self-teacher, yet transfers supervision only through next-token probabilities. We ask whether the aligned final-layer discrepancy offers a useful second channel, and how to test that channel without confusing its geometry with auxiliary strength. SCOPE-OPSD projects the privileged teacher-student residual onto a frozen rank-64 factor estimated from residual covariance and language-model-head Fisher sensitivity. It reuses the forwards already required by OPSD and adds neither rollouts nor inference-time modules. A matched Random control preserves the structured factor's rank and nonzero spectrum and uses per-arm gradient-RMS calibration, isolating the effect of the data-dependent orientation. Across the complete 25/50/75/100-step trajectories for Qwen3-1.7B, 4B, and 8B, Structured is never below Pure OPSD, with strict gains in 11 of the 12 model-checkpoint combinations and an exact tie at 4B step 25. Structured also exceeds matched Random in 10 of the 12 combinations. At step 75 on Qwen3-1.7B, Structured exceeds matched Random by 1.39 Macro Avg@12 points in each of two independent training reruns. A cross-fitted diagnostic also shows 4.40 times greater held-out privileged-gap capture than the matched random orientation. The results support a compact, Fisher-conditioned privileged subspace for short-budget OPSD.

cs.LG↗

arXiv2026

One Demonstration Is Enough for Real-World Robotic Reinforcement Learning

Learning effective robot control policies on physical hardware is challenging due to costly data collection and the difficulty of reward specification. Prior work has incorporated demonstrations into reinforcement learning (RL), yet existing approaches either require large numbers of demonstrations or depend on continuous human intervention during training. To address these limitations, we present AutoSERL, a framework that leverages a single demonstration to fully automate the intervention process in real-world robot RL. The framework includes three complementary mechanisms to accomplish certain tasks: a sliding window intervention mechanism that continuously guides exploration to prevent local optima and unsafe deviations, a safety recovery mechanism that detects and corrects failure states via predefined trajectory recovery points, and an intervention termination criterion that automatically disables guidance once the policy can independently complete the task, preserving its exploration advantage. We evaluate AutoSERL on six contact-intensive manipulation tasks across two robot platforms, spanning insertion, hanging, and hinge-based tasks. AutoSERL consistently outperforms SERL initialized with 20 demonstrations, behavior cloning, and MILES -- a dedicated one-shot imitation learning baseline -- across all tasks while matching HIL-SERL, achieves 100% success rate on insertion tasks, and demonstrates improved robustness to positional variations, all from a single demonstration. Code and videos are available on our project website: https://autoserl.github.io/.

cs.RO↗

arXiv2026

Evidence for Three-component Interlayer Coherent Exciton Condensation

Increasing the number of internal components in a quantum many-body system can host collective orders inaccessible to simpler settings. Quantum Hall bilayers provide a canonical realization of interlayer exciton condensation, yet extending such coherence across three independently addressable electronic fluids has remained elusive. Here we report evidence for three-component interlayer coherent exciton condensation in triple-layer graphene system. Using Rydberg excitons in an adjacent WSe2 monolayer as a layer-sensitive optical probe, we resolve interaction-induced incompressibility at zeroth-Landau-level crossings for all three pairwise layer combinations, establishing top-middle, middle-bottom and top-bottom exciton condensate channels within the same device. Independent control of displacement field and interlayer bias continuously tunes these pairwise states towards a regime where Landau levels from all three layers approach simultaneous degeneracy. At their convergence, incompressibility persists while the exciton energy and spectral weight evolve smoothly between the pairwise limits, suggesting coherent participation of all three layers in a single three-component state. More broadly, the ability to independently control layer potentials and engineer interlayer interactions establishes multilayer graphene as a programmable synthetic dimension for exploring higher-component quantum Hall order and simulating strongly correlated quantum matter.

cond-mat.mes-hall↗