Search

Higher-from symmetries act on sub-dimensional spatial manifolds of a quantum system and they can emerge as an exact symmetry at low energies even when they are explicitly broken at the microscopic level, making them difficult to characterize. In this work, we propose a quantitative criterion for the existence of 1-form symmetries motivated by quantum error correction (QEC). We show that the loss of the emergent 1-form symmetry is an information-theoretic transition revealed from the ensemble of post-measurement states. We analytically determine the regimes in which a 1-form symmetry emerges in product states on one- and two-dimensional lattices. The latter can be solved by mapping the ensemble of post-measurement states to the partition sum of a random bond Ising model along the Nishimori line. In analytically intractable regimes, we demonstrate how to detect 1-form symmetries with a global Minimal-Weight Perfect Matching (MWPM) decoder and numerically examine the information-theoretic transition of the 1-form symmetry, including systems with $\mathbb{Z}_2$ topological order. As an application of our protocol, we show that once the 1-form symmetry is detected to exist, a topological quantum phase transitions characterized by the spontaneous breaking of the 1-form symmetry can be accurately detected by a disorder parameter. By exploiting ideas from quantum error correction, our work develops an information-theoretic criterion for emergent 1-from symmetries, which furthers our understanding of exotic symmetries and offers practical routes toward their characterization., Comment: 24 pages, 18 figures

Deep learning models can suffer from severe performance degradation when relying on spurious correlations between input features and labels, making the models perform well on training data but have poor prediction accuracy for minority groups. This problem arises especially when training data are limited or imbalanced. While most prior work focuses on learning invariant features (with consistent correlations to y), it overlooks the potential harm of spurious correlations between features. We hereby propose Elastic Representation (ElRep) to learn features by imposing Nuclear- and Frobenius-norm penalties on the representation from the last layer of a neural network. Similar to the elastic net, ElRep enjoys the benefits of learning important features without losing feature diversity. The proposed method is simple yet effective. It can be integrated into many deep learning approaches to mitigate spurious correlations and improve group robustness. Moreover, we theoretically show that ElRep has minimum negative impacts on in-distribution predictions. This is a remarkable advantage over approaches that prioritize minority groups at the cost of overall performance., Comment: Accepted at AISTATS 2025

Graph Neural Networks (GNNs) have recently become the predominant tools for studying graph data. Despite state-of-the-art performance on graph classification tasks, GNNs are overwhelmingly trained in a single domain under supervision, thus necessitating a prohibitively high demand for labels and resulting in poorly transferable representations. To address this challenge, we propose the Label-Propagation Tensor Graph Neural Network (LP-TGNN) framework to bridge the gap between graph data and traditional domain adaptation methods. It extracts graph topological information holistically with a tensor architecture and then reduces domain discrepancy through label propagation. It is readily compatible with general GNNs and domain adaptation techniques with minimal adjustment through pseudo-labeling. Experiments on various real-world benchmarks show that our LP-TGNN outperforms baselines by a notable margin. We also validate and analyze each component of the proposed framework in the ablation study.

Phases with topological order exhibit further complexity in the presence of global symmetries: States with the same topological order are distinguished by how their anyonic excitations transform under these symmetries, leading to a classification in terms of symmetry-enriched topological phases. In this work, we develop a generic scheme to study an analogous situation for three-dimensional fracton phases by means of isometric tensor network states (isoTNS) with finite bond dimension, which allow us to tune between wavefunctions of different symmetry fractionalization. We focus on the X-Cube model, a paradigmatic fracton model hosting two types of excitations: lineons, which are mobile in a single direction only, and fractons that are completely immobile as individual particles. By deforming the local tensors that describe the ground state of the fixed point model, we find a family of exact wavefunctions for which the symmetry fractionalization under an anti-unitary symmetry on both types of excitations is directly visible. These wavefunctions have non-vanishing correlation lengths and are non-stabilizer states. At the critical points between the phases, power-law correlations are supported in certain spatial directions. Furthermore, based on the isoTNS description of the wavefunction, we determine a linear-depth quantum circuit to sequentially realize these states on a quantum processor, including a holographic scheme for which a pair of two-dimensional qubit arrays suffices to encode the three-dimensional state using measurements. Our approach provides a construction to enrich phases with exotic topological or fracton order based on the language of tensor networks and offers a tractable route to implement and characterize fracton order with quantum processors., Comment: 22 pages, 10 figures

Evolutionary computation excels in complex optimization but demands deep domain knowledge, restricting its accessibility. Large Language Models (LLMs) offer a game-changing solution with their extensive knowledge and could democratize the optimization paradigm. Although LLMs possess significant capabilities, they may not be universally effective, particularly since evolutionary optimization encompasses multiple stages. It is therefore imperative to evaluate the suitability of LLMs as evolutionary optimizer (EVO). Thus, we establish a series of rigid standards to thoroughly examine the fidelity of LLM-based EVO output in different stages of evolutionary optimization and then introduce a robust error-correction mechanism to mitigate the output uncertainty. Furthermore, we explore a cost-efficient method that directly operates on entire populations with excellent effectiveness in contrast to individual-level optimization. Through extensive experiments, we rigorously validate the performance of LLMs as operators targeted for combinatorial problems. Our findings provide critical insights and valuable observations, advancing the understanding and application of LLM-based optimization.

Since searches for the low-lying excited baryons $\Xi(1/2^-)$ and $\Sigma(1/2^-)$ are crucial to deepening our understanding of the light baryon spectrum, we have investigated the Cabibbo-favored process $\Xi_c^+ \to \Lambda {\bar{K}}^0 \pi^+$ by taking into account the $S$-wave pseudoscalar meson-octet baryon interactions within the chiral unitary approach, which could dynamically generate the resonances $\Xi(1/2^-)$ and $\Sigma(1/2^-)$. The contributions from the excited kaons are double Cabibbo-suppressed, and the contribution from the $\Sigma(1385)$ is also suppressed due to Korner-Pati-Woo theory, thus those states are expected to play negligible contributions in this process. We have predicted the $\bar{K}^0 \Lambda$ and $\pi^+\Lambda$ invariant mass distributions, which have the clear signals of the $\Xi(1/2^-)$ and $\Sigma(1/2^-)$. Thus, the $\Xi_c^+ \to \Lambda {\bar{K}}^0 \pi^+$ is an ideal process to search for the low-lying baryons $\Xi(1/2^-)$ and $\Sigma(1/2^-)$, and we make a call for a precise measurements of this process in experiments, such as Belle II, LHCb, and the proposed Super Tau-Charm Facility (STCF)., Comment: 8 pages, 9 figures

We have investigated the process $\Lambda_c^+ \to \pi^+\eta n$ by taking into account the contributions from the nucleon resonance $N(1535)$ and the scalar meson $a_0(980)$, which could be dynamically generated by the interaction of the $S$-wave pseudosalar meson-octet baryon and the $S$-wave pseudosalar meson-pseudosalar meson, respectively. Our results show that, in $\eta n$ invariant mass distribution, there is a significant near-threshold enhancement structure, which could be associated with $N(1535)$. On the other hand, one can find a clear cusp structure of $a_0(980)$ in $\pi^+\eta$ invariant mass distribution. We further estimate the ratio $R$ = $\mathcal{B}(\Lambda_c^+ \to a_0(980)^+ n)/\mathcal{B}(\Lambda_c^+ \to \pi^+\eta n)\approx 0.313$. Our results can be tested by BESIII, Belle~II, and the proposed Super Tau-Charm Facility experiments in the future., Comment: 8 pages, 9 figures

The newly observed open-flavor tetraquark $T_{c\bar{s}0}(2900)$ has attracted many attentions, and searching for its spin partners is crucial to exploring the internal structure of those states. In this work, we will show that, the $D_s^+\pi^+$ invariant mass distribution of the process $B^+\to D^{*-}D_s^+\pi^+$ measured by LHCb has a resonant-like structure around 2830~MeV, which could be associated with the predicted $T_{c\bar{s}2}$, the spin $J=2$ partner of $T_{c\bar{s}0}(2900)$. Furthermore, we have evaluated the momenta of the angular mass distribution, which are very different for each of the spin assumptions, and have larger strength at the resonant energy than the peaks seen in the angular integrated mass distribution. We make a call for the experimental determination of these magnitudes, which could be used to pin down the existence of the $T_{c\bar{s}2}$., Comment: 6 pages, 5 figures, comments are welcome