Search

Numerous retrieval models, including sparse, dense and llm-based methods, have demonstrated remarkable performance in predicting the relevance between queries and corpora. However, the preliminary effectiveness analysis experiments indicate that these models fail to achieve satisfactory performance on the majority of queries and corpora, revealing their effectiveness restricted to specific scenarios. Thus, to tackle this problem, we propose a novel Distributed Collaborative Retrieval Framework (DCRF), outperforming each single model across all queries and corpora. Specifically, the framework integrates various retrieval models into a unified system and dynamically selects the optimal results for each user's query. It can easily aggregate any retrieval model and expand to any application scenarios, illustrating its flexibility and scalability.Moreover, to reduce maintenance and training costs, we design four effective prompting strategies with large language models (LLMs) to evaluate the quality of ranks without reliance of labeled data. Extensive experiments demonstrate that proposed framework, combined with 8 efficient retrieval models, can achieve performance comparable to effective listwise methods like RankGPT and ListT5, while offering superior efficiency. Besides, DCRF surpasses all selected retrieval models on the most datasets, indicating the effectiveness of our prompting strategies on rank-oriented automatic evaluation.

Multi-modal hashing methods have gained popularity due to their fast speed and low storage requirements. Among them, the supervised methods demonstrate better performance by utilizing labels as supervisory signals compared with unsupervised methods. Currently, for almost all supervised multi-modal hashing methods, there is a hidden assumption that training sets have no noisy labels. However, labels are often annotated incorrectly due to manual labeling in real-world scenarios, which will greatly harm the retrieval performance. To address this issue, we first discover a significant distribution consistency pattern through experiments, i.e., the 1-0 distribution of the presence or absence of each category in the label is consistent with the high-low distribution of similarity scores of the hash codes relative to category centers. Then, inspired by this pattern, we propose a novel Distribution-Consistency-Guided Multi-modal Hashing (DCGMH), which aims to filter and reconstruct noisy labels to enhance retrieval performance. Specifically, the proposed method first randomly initializes several category centers, which are used to compute the high-low distribution of similarity scores; Noisy and clean labels are then separately filtered out via the discovered distribution consistency pattern to mitigate the impact of noisy labels; Subsequently, a correction strategy, which is indirectly designed via the distribution consistency pattern, is applied to the filtered noisy labels, correcting high-confidence ones while treating low-confidence ones as unlabeled for unsupervised learning, thereby further enhancing the model's performance. Extensive experiments on three widely used datasets demonstrate the superiority of the proposed method compared to state-of-the-art baselines in multi-modal retrieval tasks. The code is available at https://github.com/LiuJinyu1229/DCGMH.

Thermal silica is a common dielectric used in all silicon-photonic circuits. And bound hydroxyl ions (Si-OH) can provide a significant component of optical loss in this material on account of the wet nature of the thermal oxidation process. A convenient way to quantify this loss relative to other mechanisms is through OH-absorption at 1380 nm. Here, using ultra-high-Q thermal-silica wedge microresonators, the OH absorption loss peak is measured and distinguished from the scattering loss base line over a wavelength range from 680 nm to 1550 nm. Record-high on-chip resonator Q factors are observed for near-visible and visible wavelengths, and the absorption limited Q factor is as high as 8 billion in the telecom band. OH ion content level around 2.4 ppm (weight) is inferred from both Q measurements and by Secondary Ion Mass Spectroscopy (SIMS) depth profiling., Comment: 4 pages, 3 figures

Understanding strongly correlated quantum materials, such as high $T_\textrm{c}$ superconductors, iron-based superconductors, and twisted bilayer graphene systems, remains to be one of the outstanding challenges in condensed matter physics. Quantum simulation with ultra-cold atoms in particular optical lattices, which provide orbital degrees of freedom, is a powerful tool to contribute new insights to this endeavor. Here, we report the experimental realization of an unconventional Bose-Einstein condensate of $^{87}$Rb atoms populating degenerate $p$-orbitals in a triangular optical lattice, exhibiting remarkably long coherence times. Using time-of-flight spectroscopy, we observe that this state spontaneously breaks the rotational symmetry and its momentum spectrum agrees with the theoretically predicted coexistence of exotic stripe and loop current orders. Like certain strongly correlated electronic systems with intertwined orders, as high-$T_\textrm{c}$ cuprate superconductors, twisted bilayer graphene, and the recently discovered chiral density-wave state in kagome superconductors $\textrm{AV}_3 \textrm{Sb}_5$ (A=K, Rb, Cs), the newly demonstrated quantum state, in spite of its markedly different energy scale and the bosonic quantum statistics, exhibits multiple symmetry breakings at ultralow temperatures. These findings hold the potential to enhance our comprehension of the fundamental physics governing these intricate quantum materials., Comment: 7+10 pages, 3+12 figures

Hexagonal optical lattices offer a tunable platform to study exotic orbital physics in solid state materials. Here, we present a versatile high-precision scheme to implement a hexagonal optical lattice potential, which is engineered by overlapping two independent triangular optical sublattices generated by laser beams with slightly different wavelengths around 1064 nm. This enables us to precisely control the detailed structure of the hexagonal lattice by adjusting the relative position and the relative lattice depth of the two triangular optical sublattices. Taking advantage of the sensitive dependence of the second Bloch band on small lattice deformations, we propose a strategy to optimize the optical lattice geometry with an extremely high precision. This method can also be extended to other lattice configurations involving more than two sublattices. Our work provides the experimental requirements in the search for novel orbital physics of ultracold atoms, for example, in the flat $p$-band of the hexagonal optical lattice., Comment: 9 pages, 4 figures

We propose a general and experimentally feasible approach to realize simultaneous ground-state cooling of arbitrary number of near-degenerate, or even fully degenerate mechanical modes, overcoming the limit imposed by the formation of mechanical dark modes. Multiple optical modes are employed to provide different dissipation channels that prevent complete destructive interference of the cooling pathway, and thus eliminating the dark modes. The cooling rate and limit are explicitly specified, in which the distinguishability of the optical modes to the mechanical modes is found to be critical for an efficient cooling process. In a realistic multi-mode optomechanical system, ground-state cooling of all mechanical modes is demonstrated by sequentially introducing optical drives, proving the feasibility and scalability of the proposed scheme. The work may provide new insights in preparing and manipulating multiple quantum states in macroscopic systems.

We consider the shift of charge-to-mass ratio for extremal black holes in the context of effective field theory, motivated by the Weak Gravity Conjecture. We constrain extremality corrections in different regimes subject to unitarity and causality constraints. In the asymptotic IR, we demonstrate that for any supersymmetric theory in flat space, and for all minimally coupled theories, logarithmic running at one loop pushes the Wilson coefficient of certain four-derivative operators to be larger at lower energies, guaranteeing the existence of sufficiently large black holes with $Q>M$. We identify two exceptional cases of nonsupersymmetric theories involving large numbers of light states and Planck-scale nonminimal couplings, in which the sign of the running is reversed, leading to black holes with negative corrections to $Q/M$ in the deep IR, but argue that these do not rule out extremal black holes as the requisite charged states for the WGC. We separately show that causality and unitarity imply that the leading threshold corrections to the effective action from integrating out massive states, in any weakly coupled theory, can be written as a sum of squares and is manifestly positive for black hole backgrounds. Quite beautifully, the shift in the extremal $Q/M$ ratio is directly proportional to the shift in the on-shell action, guaranteeing that these threshold corrections push $Q>M$ in compliance with the WGC. Our results apply for black holes with or without dilatonic coupling and charged under any number of ${\rm U}(1)$s., Comment: 41 pages, 5 figures

In this paper, we construct the supersymmetric spinning polynomials. These are orthogonal polynomials that serve as an expansion basis for the residue or discontinuity of four-point scattering amplitudes, respecting four-dimensional super Poincare invariance. The polynomials are constructed by gluing on-shell supersymmetric three-point amplitudes of one massive two massless multiplets, and are identified with algebraic Jacobi-polynomials. Equipped with these we construct the supersymmetric EFThedron, which geometrically defines the allowed region of Wilson coefficients respecting UV unitarity and super Poincare invariance., Comment: 18 pages, 3 figures

In this paper, we explore the open string amplitude's dual role as a space-time S-matrix and a 1D CFT correlation function. We pursue this correspondence in two directions. First, beginning with a general disk integrand dressed with a Koba-Nielsen factor, we demonstrate that exchange symmetry for the factorization residue of the amplitude forces the integrand to be expandable on SL(2,R) conformal blocks. Furthermore, positivity constraints associated with unitarity imply the SL(2,R) blocks must come in linear combinations for which the Virasoro block emerges at the "kink" in the space of solutions. In other words, Virasoro symmetry arises at the boundary of consistent factorization. Next, we consider the low energy EFT description, where unitarity manifests as the EFThedron in which the couplings must live. The existence of a worldsheet description implies, through the Koba-Nielsen factor, monodromy relations which impose algebraic identities amongst the EFT couplings. We demonstrate at finite derivative order that the intersection of the "monodromy plane" and the EFThedron carves out a tiny island for the couplings, which continues to shrink as the derivative order is increased. At the eighth derivative order, on a three-dimensional monodromy plane, the intersection fixes the width of this island to around 1.5$\%$ (of $\zeta(3)$) and 0.2$\%$ (of $\zeta(5)$) with respect to the super string answer. This leads us to conjecture that the four-point open superstring amplitude can be completely determined by the geometry of the intersection of the monodromy plane and the EFThedron., Comment: 13 figures

In search of natural building blocks for supergravity amplitudes, a tentative criteria is term-by-term bonus z^-2 large momentum scaling. For a given choice of deformation legs, we present such an expansion in the form of a BCFW representation in N=7 supergravity based on a special shift. We will show that this improved scaling behavior, with respect to the fully N=8 representation, is due to its automatic incorporation of the so called bonus relations., Comment: 16 pages, 2 figures