Source author record

Peijie Sun

Peijie Sun appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

cond-mat.str-el Information Retrieval cond-mat.mtrl-sci Artificial Intelligence Computation and Language cond-mat.mes-hall Cryptography and Security Machine Learning Social and Information Networks

Catalog footprint

What is connected

19works

9topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

PRISM: Personalized Recommendation via Information Synergy Module

Multimodal sequential recommendation (MSR) leverages diverse item modalities to improve recommendation accuracy, while achieving effective and adaptive fusion remains challenging. Existing MSR models often overlook synergistic information that emerges only through modality combinations. Moreover, they typically assume a fixed importance for different modality interactions across users. To address these limitations, we propose \textbf{P}ersonalized \textbf{R}ecommend-ation via \textbf{I}nformation \textbf{S}ynergy \textbf{M}odule (PRISM), a plug-and-play framework for sequential recommendation (SR). PRISM explicitly decomposes multimodal information into unique, redundant, and synergistic components through an Interaction Expert Layer and dynamically weights them via an Adaptive Fusion Layer guided by user preferences. This information-theoretic design enables fine-grained disentanglement and personalized fusion of multimodal signals. Extensive experiments on four datasets and three SR backbones demonstrate its effectiveness and versatility. The code is available at https://github.com/YutongLi2024/PRISM.

preprint2026arXiv

Safety Anchor: Defending Harmful Fine-tuning via Geometric Bottlenecks

The safety alignment of Large Language Models (LLMs) remains vulnerable to Harmful Fine-tuning (HFT). While existing defenses impose constraints on parameters, gradients, or internal representations, we observe that they can be effectively circumvented under persistent HFT. Our analysis traces this failure to the inherent redundancy of the high-dimensional parameter space: attackers exploit optimization trajectories that are orthogonal to defense constraints to restore harmful capabilities while deceptively adhering to safety restrictions. To address this, we propose Safety Bottleneck Regularization (SBR). SBR shifts the defensive focus from the redundant parameter space to the unembedding layer, which serves as a geometric bottleneck. By anchoring the final hidden states of harmful queries to those of the safety-aligned model, SBR enables the model to maintain safe responses even under persistent HFT. Extensive experiments confirm SBR's effectiveness, demonstrating that utilizing just a single safety anchor is sufficient to reduce the Harmful Score to $<$10 while preserving competitive performance on benign downstream tasks.

preprint2023arXiv

Dipolar Spin Liquid Ending with Quantum Critical Point in a Gd-based Triangular Magnet

By performing experiment and model studies on a triangular-lattice dipolar magnet KBaGd(BO$_3$)$_2$ (KBGB), we find the highly frustrated magnet with a planar anisotropy hosts a strongly fluctuating dipolar spin liquid (DSL), which originates from the intriguing interplay between dipolar and Heisenberg interactions. The DSL constitutes an extended regime in the field-temperature phase diagram, which gets lowered in temperature as field increases and eventually ends with an unconventional quantum critical point (QCP) at $B_c\simeq 0.75$~T. Based on dipolar Heisenberg model calculations, we identify the DSL as a Berezinskii-Kosterlitz-Thouless (BKT) phase with emergent U(1) symmetry. Due to the tremendous entropy accumulation that can be related to the strong BKT and quantum fluctuations, unprecedented magnetic cooling effects are observed in the DSL regime and particularly near the QCP, making KBGB a superior dipolar coolant to commercial Gd-based refrigerants. We establish the phase diagram for triangular-lattice dipolar quantum magnets where emergent symmetry plays an essential role, and provide a basis and opens an avenue for their applications in sub-Kelvin refrigeration.

preprint2022arXiv

A Review-aware Graph Contrastive Learning Framework for Recommendation

Most modern recommender systems predict users preferences with two components: user and item embedding learning, followed by the user-item interaction modeling. By utilizing the auxiliary review information accompanied with user ratings, many of the existing review-based recommendation models enriched user/item embedding learning ability with historical reviews or better modeled user-item interactions with the help of available user-item target reviews. Though significant progress has been made, we argue that current solutions for review-based recommendation suffer from two drawbacks. First, as review-based recommendation can be naturally formed as a user-item bipartite graph with edge features from corresponding user-item reviews, how to better exploit this unique graph structure for recommendation? Second, while most current models suffer from limited user behaviors, can we exploit the unique self-supervised signals in the review-aware graph to guide two recommendation components better? To this end, in this paper, we propose a novel Review-aware Graph Contrastive Learning (RGCL) framework for review-based recommendation. Specifically, we first construct a review-aware user-item graph with feature-enhanced edges from reviews, where each edge feature is composed of both the user-item rating and the corresponding review semantics. This graph with feature-enhanced edges can help attentively learn each neighbor node weight for user and item representation learning. After that, we design two additional contrastive learning tasks (i.e., Node Discrimination and Edge Discrimination) to provide self-supervised signals for the two components in recommendation process. Finally, extensive experiments over five benchmark datasets demonstrate the superiority of our proposed RGCL compared to the state-of-the-art baselines.

preprint2022arXiv

ProFairRec: Provider Fairness-aware News Recommendation

News recommendation aims to help online news platform users find their preferred news articles. Existing news recommendation methods usually learn models from historical user behaviors on news. However, these behaviors are usually biased on news providers. Models trained on biased user data may capture and even amplify the biases on news providers, and are unfair for some minority news providers. In this paper, we propose a provider fairness-aware news recommendation framework (named ProFairRec), which can learn news recommendation models fair for different news providers from biased user data. The core idea of ProFairRec is to learn provider-fair news representations and provider-fair user representations to achieve provider fairness. To learn provider-fair representations from biased data, we employ provider-biased representations to inherit provider bias from data. Provider-fair and -biased news representations are learned from news content and provider IDs respectively, which are further aggregated to build fair and biased user representations based on user click history. All of these representations are used in model training while only fair representations are used for user-news matching to achieve fair news recommendation. Besides, we propose an adversarial learning task on news provider discrimination to prevent provider-fair news representation from encoding provider bias. We also propose an orthogonal regularization on provider-fair and -biased representations to better reduce provider bias in provider-fair representations. Moreover, ProFairRec is a general framework and can be applied to different news recommendation methods. Extensive experiments on a public dataset verify that our ProFairRec approach can effectively improve the provider fairness of many existing methods and meanwhile maintain their recommendation accuracy.

preprint2021arXiv

DiffNet++: A Neural Influence and Interest Diffusion Network for Social Recommendation

Social recommendation has emerged to leverage social connections among users for predicting users' unknown preferences, which could alleviate the data sparsity issue in collaborative filtering based recommendation. Early approaches relied on utilizing each user's first-order social neighbors' interests for better user modeling and failed to model the social influence diffusion process from the global social network structure. Recently, we propose a preliminary work of a neural influence diffusion network (i.e., DiffNet) for social recommendation (Diffnet), which models the recursive social diffusion process to capture the higher-order relationships for each user. However, we argue that, as users play a central role in both user-user social network and user-item interest network, only modeling the influence diffusion process in the social network would neglect the users' latent collaborative interests in the user-item interest network. In this paper, we propose DiffNet++, an improved algorithm of DiffNet that models the neural influence diffusion and interest diffusion in a unified framework. By reformulating the social recommendation as a heterogeneous graph with social network and interest network as input, DiffNet++ advances DiffNet by injecting these two network information for user embedding learning at the same time. This is achieved by iteratively aggregating each user's embedding from three aspects: the user's previous embedding, the influence aggregation of social neighbors from the social network, and the interest aggregation of item neighbors from the user-item interest network. Furthermore, we design a multi-level attention network that learns how to attentively aggregate user embeddings from these three aspects. Finally, extensive experimental results on two real-world datasets clearly show the effectiveness of our proposed model.

preprint2020arXiv

Nonsaturating magnetoresistance, anomalous Hall effect, and magnetic quantum oscillations in ferromagnetic semimetal PrAlSi

We report a comprehensive investigation of the structural, magnetic, transport and thermodynamic properties of a single crystal PrAlSi, in comparison to its nonmagnetic analogue LaAlSi. PrAlSi exhibits a ferromagnetic transition at $T_C$ = 17.8 K which, however, is followed by two weak phase transitions at lower temperatures. Based on the combined dc and ac magnetic susceptibility measurements, we propose the two reentrant magnetic phases below $T_C$ to be spin glasses or ferromagnetic cluster glasses. When the magnetic glassy states are suppressed by small field, several remarkable features appear. These include a linear, nonsaturating magnetoresistance as a function of field that is reminiscent of a topological or charge-compensated semimetal, and a large anomalous Hall conductivity amounting to $\sim$2000 $Ω^{-1}$cm$^{-1}$. Specific-heat measurements indicate a non-Kramers doublet ground state and a relatively low crystal electric field splitting of the Pr$^{3+}$ multiplets of less than 100 K. Shubnikov-de Hass oscillations are absent in LaAlSi, whereas they are clearly observed below about 25 K in PrAlSi, with an unusual temperature dependence of the dominating oscillation frequency $F$. It increases from $F$ = 18 T at 25 K to $F$ = 33 T at 2 K, hinting at an emerging Fermi pocket upon cooling into the ordered phase. These results suggest that PrAlSi is a new system where a small Fermi pocket of likely relativistic fermions is strongly coupled to magnetism. Whether hybridization between $f$ and conduction band is also involved remains an intriguing open problem.

preprint2019arXiv

Large transverse thermoelectric figure of merit in a Dirac semimetal

Thermoelectric (TE) conversion in conducting materials is of eminent importance for providing renewable energy and solid-state cooling. Although traditionally, the Seebeck effect plays a key role for the TE figure of merit zST, it encounters fundamental constraints hindering its conversion efficiency. Most notably, there are the charge compensation of electrons and holes that diminishes this effect, and the intertwinement of the corresponding electrical and thermal conductivities through the Wiedemann-Franz (WF) law which makes their independent optimization in zST impossible. Here, we demonstrate that in the Dirac semimetal Cd3As2 the Nernst effect, i.e., the transverse counterpart of the Seebeck effect, can generate a large TE figure of merit zNT. At room temperature, zNT = 0.5 in a small field of 2 T; it significantly surmounts its longitudinal counterpart zST for any field and further increases upon warming. A large Nernst effect is generically expected in topological semimetals, benefiting from both the bipolar transport of compensated electrons and holes and their high mobilities. In this case, heat and charge transport are orthogonal, i.e., not intertwined by the WF law anymore. More importantly, further optimization of zNT by tuning the Fermi level to the Dirac node can be anticipated due to not only the enhanced bipolar transport, but also the anomalous Nernst effect arising from a pronounced Berry curvature. A combination of the former topologically trivial and the latter nontrivial advantages promises to open a new avenue towards high-efficient transverse thermoelectricity.

preprint2019arXiv

Quantum-critical phase out of frustrated magnetism in a strongly correlated metal

Strange-metal phenomena often develop at the border of antiferromagnetic order in strongly correlated metals. It has been well established that they can originate from the fluctuations anchored by the point of continuous quantum phase transition out of the antiferromagnetic order, i.e., a quantum critical point. What has been unclear is how these phenomena can be associated with a potential new phase of matter at zero temperature. Here we show that magnetic frustration of the 4f-local moments in the distorted Kagome intermetallic compound CePdAl gives rise to such a paramagnetic quantum-critical phase. Moreover, we demonstrate that this phase turns into a Fermi liquid through a Mott-like crossover; in a two-dimensional parameter space of pressure and magnetic field, this crossover is linked to a line of zero-temperature 4f-electron localization-delocalization phase transitions at low and moderate pressures. Our discovery motivates a new design principle for strongly correlated metallic states with unconventional excitations that may underlie the development of such effects as high temperature superconductivity.

preprint2016arXiv

Magneto-transport properties of CeRu$_2$Al$_{10}$: Similarities to URu$_2$Si$_2$

We report on magneto-transport properties of the Kondo semiconducting compound CeRu$_2$Al$_{10}$, focusing on its exotic phase below $T_0$ = 27 K. In this phase, an excess thermal conductivity $κ$ emerges and is gradually suppressed by magnetic field, strikingly resembling those observed in the hidden-order phase of URu$_2$Si$_2$. Our analysis indicates that low-energy magnetic excitation is the most likely origin, as was also proposed for URu$_2$Si$_2$ recently, despite the largely reduced magnetic moments. Likewise, other transport properties such as resistivity, thermopower and Nernst effect exhibit distinct features characterizing the very different charge dynamics above and below $T_0$, sharing similarities to URu$_2$Si$_2$, too. Given the exotic nature of the ordered phases in both compounds, whether a unified interpretation to all these observations exists appears to be extremely interesting.

preprint2016arXiv

Temperature-field phase diagram of geometrically frustrated CePdAl

From the electrical resistivity and ac susceptibility measurements down to $T$ = 40 mK, a rich temperature-field phase diagram has been constructed for the geometrically frustrated heavy-fermion compound CePdAl. In the limit of zero temperature, the field-induced suppression of the antiferromagnetism in this compound results in a sequence of phase transitions/corssovers, rather than a single critical point, in a critical region of field 3$-$5 T. In the lowest temperature region, a power-law description of the temperature-dependent resistivity yields $A$$\cdot$$T^n$ with $n$$>$2 for all applied fields. However, a quadratic-in-temperature resistivity term can be extracted when assuming an additional electron scattering channel from magnon-like excitations, with $A$ diverging upon approaching the critical regime. Our observations suggest CePdAl to be a new playground for exotic physics arising from the competition between Kondo screening and geometrical frustration.

preprint2015arXiv

Large Seebeck Effect by Charge-Mobility Engineering

The Seebeck effect describes the generation of an electric potential in a conducting solid exposed to a temperature gradient. Besides fundamental relevance in solid state physics, it serves as a key quantity to determine the performance of functional thermoelectric materials. In most cases, it is dominated by an energy-dependent electronic density of states at the Fermi level, in line with the prevalent efforts toward superior thermoelectrics through the engineering of electronic structure. Here, we demonstrate an alternative source for the Seebeck effect based on charge-carrier relaxation: A charge mobility that changes rapidly with temperature can result in a sizeable addition to the Seebeck coefficient. This new Seebeck source is demonstrated explicitly for Ni-doped CoSb3, where a dramatic mobility change occurs due to the crossover between two different charge-relaxation regimes. Our findings unveil the origin of pronounced features in the Seebeck coefficient of many other elusive materials characterized by a significant mobility mismatch. As the physical origin for the latter can vary greatly, our proposal provides a unifying framework for the understanding of a large panoply of thermoelectric phenomena. When utilized appropriately, this effect can also provide a novel route to the design of improved thermoelectric materials for applications in solid-state cooling or power generation.

preprint2014arXiv

Resonant Charge Relaxation as a Likely Source of the Enhanced Thermopower in FeSi

The enhanced thermopower of the correlated semiconductor FeSi is found to be robust against the sign of the relevant charge carriers. At $T$\,$\approx$\,70 K, the position of both the high-temperature shoulder of the thermopower peak and the nonmagnetic-enhanced paramagnetic crossover, the Nernst coefficient $ν$ assumes a large maximum and the Hall mobility $μ_H$ diminishes to below 1 cm$^2$/Vs. These cause the dimension-less ratio $ν$/$μ_H$ $-$ a measure of the energy dispersion of the charge scattering time $τ(ε)$ $-$ to exceed that of classical metals and semiconductors by two orders of magnitude. Concomitantly, the resistivity exhibits a hump and the magnetoresistance changes its sign. Our observations hint at a resonant scattering of the charge carriers at the magnetic crossover, imposing strong constraints on the microscopic interpretation of the robust thermopower enhancement in FeSi.

preprint2013arXiv

Evidence for a Kondo destroying quantum critical point in YbRh2Si2

The heavy-fermion metal YbRh$_{2}$Si$_{2}$ is a weak antiferromagnet below $T_{N} = 0.07$ K. Application of a low magnetic field $B_{c} = 0.06$ T ($\perp c$) is sufficient to continuously suppress the antiferromagnetic (AF) order. Below $T \approx 10$ K, the Sommerfeld coefficient of the electronic specific heat $γ(T)$ exhibits a logarithmic divergence. At $T < 0.3$ K, $γ(T) \sim T^{-ε}$ ($ε: 0.3 - 0.4$), while the electrical resistivity $ρ(T) = ρ_{0} + aT$ ($ρ_{0}$: residual resistivity). Upon extrapolating finite-$T$ data of transport and thermodynamic quantities to $T = 0$, one observes (i) a vanishing of the "Fermi surface crossover" scale $T^{*}(B)$, (ii) an abrupt jump of the initial Hall coefficient $R_{H}(B)$ and (iii) a violation of the Wiedemann Franz law at $B = B_{c}$, the field-induced quantum critical point (QCP). These observations are interpreted as evidence of a critical destruction of the heavy quasiparticles, i.e., propagating Kondo singlets, at the QCP of this material.

preprint2013arXiv

Highly Dispersive Electron Relaxation and Colossal Thermoelectricity in the Correlated Semiconductor FeSb$_2$

We show that the colossal thermoelectric power, $S(T)$, observed in the correlated semiconductor FeSb$_2$ below 30\,K is accompanied by a huge Nernst coefficient $ν(T)$ and magnetoresistance MR$(T)$. Markedly, the latter two quantities are enhanced in a strikingly similar manner. While in the same temperature range, $S(T)$ of the reference compound FeAs$_2$, which has a seven-times larger energy gap, amounts to nearly half of that of FeSb$_2$, its $ν(T)$ and MR$(T)$ are intrinsically different to FeSb$_2$: they are smaller by two orders of magnitude and have no common features. With the charge transport of FeAs$_2$ successfully captured by the density functional theory, we emphasize a significantly dispersive electron-relaxation time $τ(ε_k)$ due to electron-electron correlations to be at the heart of the peculiar thermoelectricity and magnetoresistance of FeSb$_2$.

preprint2012arXiv

Nernst Effect as a Probe of Local Kondo Scattering in Heavy Fermions

A large, strongly temperature-dependent Nernst coefficient, $ν$, is observed between $T$ = 2 K and 300 K for CeCu$_2$Si$_2$ and Ce$_{0.8}$La$_{0.2}$Cu$_2$Si$_2$. The enhanced $ν(T)$ is determined by the asymmetry of the on-site Kondo (conduction electron$-4f$ electron) scattering rate. Taking into account the measured Hall mobility, $μ_H$, the highly unusual thermopower, $S$, of these systems can be semiquantitatively described by $S(T)$ $=$ $-$$ν(T)/μ_H(T)$, which explicitly demonstrates that the thermopower originates from the local Kondo scattering process over a wide temperature range from far above to well below the coherence temperature ($\approx$ 20 K for CeCu$_2$Si$_2$). Our results suggest that the Nernst effect can act as a proper probe of local charge-carrier scattering. This promises an impact on exploring the unconventional enhancement of the thermopower in correlated materials suited for potential applications.

preprint2011arXiv

Enhanced electron correlations in FeSb$_2$

FeSb$_2$ has been recently identified as a new model system for studying many-body renormalizations in a $d$-electron based narrow gap semiconducting system, strongly resembling FeSi. The electron-electron correlations in FeSb$_2$ manifest themselves in a wide variety of physical properties including electrical and thermal transport, optical conductivity, magnetic susceptibility, specific heat and so on. We review some of the properties that form a set of experimental evidences revealing the crucial role of correlation effects in FeSb$_2$. The metallic state derived from slight Te doping in FeSb$_2$, which has large quasiparticle mass, will also be introduced.

preprint2009arXiv

Simultaneously optimizing the interdependent thermoelectric parameters in Ce(Ni$_{1-x}$Cu$_x$)$_2$Al$_3$

Substitution of Cu for Ni in the Kondo lattice system CeNi$_2$Al$_3$ results in a simultaneous optimization of the three interdependent thermoelectric parameters: thermoelectric power, electrical and thermal conductivities, where the electronic change in conduction band induced by the extra electron of Cu is shown to be crucial. The obtained thermoelectric figure of merit $zT$ amounts to 0.125 at around 100 K, comparable to the best values known for Kondo compounds. The realization of ideal thermoelectric optimization in Ce(Ni$_{1-x}$Cu$_x$)$_2$Al$_3$ indicates that proper electronic tuning of Kondo compounds is a promising approach to efficient thermoelectric materials for cryogenic application.

preprint2006arXiv

Anomalous elastic softening of SmRu_{4}P_{12} under high pressure

The filled skutterudite compound SmRu_4P_{12} undergoes a complex evolution from a paramagnetic metal (phase I) to a probable multipolar ordering insulator (phase II) at T_{MI} = 16.5 K, then to a magnetically ordered phase (phase III) at T_{N} = 14 K. Elastic properties under hydrostatic pressures were investigated to study the nature of the ordering phases. We found that distinct elastic softening above T_{MI} is induced by pressure, giving evidence of quadrupole degeneracy of the ground state in the crystalline electric field. It also suggests that quadrupole moment may be one of the order parameters below T_{MI} under pressure. Strangely, the largest degree of softening is found in the transverse elastic constant C_{T} at around 0.5-0.6 GPa, presumably having relevancy to the competing and very different Gruneisen parameters Ωof T_{MI} and T_{N}. Interplay between the two phase transitions is also verified by the rapid increase of T_{MI} under pressure with a considerably large Ωof 9. Our results can be understood on the basis of the proposed octupole scenario for SmRu_4P_{12}.