arXiv subjects

Zixi Li

Publications and source records attributed to Zixi Li.

13 recordsLinked to original sources

arXiv2026

Entropy as a Structural Prior: How a Log-Barrier on DiT Belief Space Drives Musical Diversity and Development

Confidence-based loss weighting is usually avoided in generative models because it accelerates errors when the model is confidently wrong, but this intuition breaks down in supervised diffusion training. We introduce the Eisbach log-barrier, a parameter-free weight derived from the entropy of the DiT output's spatial energy distribution: high entropy damps the gradient, while low entropy preserves it. Applied to LoRA fine-tuning of Stable Audio 3 Medium on MusicCaps, it unexpectedly yields stronger thematic development, clearer acoustic differentiation, and higher textural diversity than unweighted training, the opposite of mode collapse. This works because in supervised diffusion the gradient direction is locked to ground truth, so confidence only scales the step size, and because temporal entropy downweights flat samples while preserving high-contrast ones. The result is an online, self-referential data curriculum that emerges purely from the forward pass, with analyzed noise-level dynamics and testable predictions.

cs.SD↗

arXiv2026

A Closed-Form Upper Bound for Admissible Learning-Rate Steps in Belief-Space Dynamics

Learning-rate steps are usually treated as hyperparameters. This paper isolates a local beliefspace calculation: when an update is modeled as a projected forward step on the probability simplex, admissibility means contractivity in the natural KL/Bregman geometry. Under this model, the upper bound of an admissible step is not a tuning slogan but a formula.

cs.LG↗

arXiv2026

Heterogeneously Integrated Diamond-on-Lithium Niobate Quantum Photonic Platform

Diamond photonics has enabled efficient interfaces for quantum memories and is predicted to be a critical component of quantum networks. However, scalable network architectures require spatial, temporal, and spectral control of photons, which relies on nonlinear and electro-optic functionalities that diamond alone cannot provide. Here, we demonstrate heterogeneous integration of a thin-film lithium niobate (TFLN) platform, which has strong chi-2 nonlinearity and electro-optic effects, with thin diamond films. We demonstrate high-Q diamond photonic crystal cavities (Q factors exceeding 5x10^4 at 735 nm) that are lithographically aligned with TFLN photonic backbone and critically coupled to it. This allows us to realize low-loss diamond-TFLN "escalators" (loss ~1 dB/coupler) that support efficient light transfer between them. At cryogenic temperatures (5K), we can collect photons emitted from silicon vacancies (SiVs) embedded within the diamond structure via the TFLN photonic circuit. This approach establishes a scalable route toward integrated photonic circuits for practical quantum networking and other technologies.

physics.optics↗

arXiv2025

Reasoning: From Reflection to Solution

What is reasoning? This question has driven centuries of philosophical inquiry, from Aristotle's syllogisms to modern computational complexity theory. In the age of large language models achieving superhuman performance on benchmarks like GSM8K (95\% accuracy) and HumanEval (90\% pass@1), we must ask: have these systems learned to \emph{reason}, or have they learned to \emph{pattern-match over reasoning traces}? This paper argues for a specific answer: \textbf{reasoning is iterative operator application in state spaces, converging to fixed points}. This definition is not merely philosophical -- it has concrete architectural implications that explain both the failures of current systems and the path to genuine reasoning capabilities. Our investigation begins with a puzzle (OpenXOR), progresses through theory (OpenOperator), and culminates in a working solution (OpenLM) that achieves 76\% accuracy where state-of-the-art LLMs achieve 0\%. This is not about criticizing existing systems, but about \emph{understanding what reasoning requires} and \emph{building architectures that provide it}.

cs.LG↗

arXiv2025

Asterisk Operator

We propose the \textbf{Asterisk Operator} ($\ast$-operator), a novel unified framework for abstract reasoning based on Adjacency-Structured Parallel Propagation (ASPP). The operator formalizes structured reasoning tasks as local, parallel state evolution processes guided by implicit relational graphs. We prove that the $\ast$-operator maintains local computational constraints while achieving global reasoning capabilities, providing an efficient and convergent computational paradigm for abstract reasoning problems. Through rigorous mathematical analysis and comprehensive experiments on ARC2 challenges and Conway's Game of Life, we demonstrate the operator's universality, convergence properties, and superior performance. Our innovative Embedding-Asterisk distillation method achieves 100\% accuracy on ARC2 validation with only 6M parameters, representing a significant breakthrough in neural-symbolic reasoning. \textbf{Keywords:} Abstract Reasoning, Adjacency Structure, Parallel Propagation, Asterisk Operator, Convergence, Universal Approximation

cs.AI↗

arXiv2025

TreeGPT: Pure TreeFFN Encoder-Decoder Architecture for Structured Reasoning Without Attention Mechanisms

We present TreeGPT, an attention-free neural architecture that explores the potential of pure TreeFFN encoder-decoder design for structured reasoning tasks. Unlike traditional transformer approaches that rely on attention mechanisms, TreeGPT employs bidirectional TreeFFN components that process sequences through adjacent connections in parallel, aiming to achieve computational efficiency while maintaining reasoning capabilities. Our approach centers on a TreeFFN Encoder-Decoder mechanism: $$\text{Encoder TreeFFN (L} \rightarrow \text{R)} + \text{Decoder TreeFFN (R} \leftarrow \text{L)} \rightarrow \text{Parallel Processing}$$ where the encoder processes left-to-right dependencies while the decoder handles right-to-left patterns, both using simple neighbor-to-neighbor connections. This design eliminates attention computation while maintaining sequence modeling capabilities. We evaluate our approach on the ARC Prize 2025 dataset, where TreeGPT achieves 99\% validation accuracy using 3.16M parameters. The model converges within 1500 training steps and demonstrates 100\% token-level accuracy on selected evaluation samples. Our preliminary results suggest that for certain structured reasoning tasks, specialized TreeFFN architectures may offer advantages over attention-based approaches. While these findings are encouraging, we acknowledge that further investigation across diverse tasks and datasets would be valuable to establish the broader applicability of attention-free designs.

cs.AI↗

arXiv2025

Pointer: Linear-Complexity Long-Range Modeling without Pre-training

We introduce Pointer, a novel architecture that achieves linear $O(NK)$ complexity for long-range sequence modeling while maintaining superior performance without requiring pre-training. Unlike standard attention mechanisms that compute $O(N^2)$ pairwise interactions, our approach uses layer-wise pointer chaining where each layer's pointer selection depends on previous layer's pointer positions, creating explicit long-distance connections through pointer chains. We demonstrate that this architecture achieves $2$--$10\times$ speedup on long sequences compared to standard transformers, maintains $>95\%$ accuracy on copy tasks at distances up to 2048 tokens, and learns interpretable pointer patterns that reveal structured dependency modeling. Our experiments on efficiency benchmarks, long-range dependency tasks, and interpretability analysis show that Pointer offers a compelling alternative to attention mechanisms for scenarios requiring efficient long-range modeling without pre-training dependencies.

cs.CL↗

arXiv2025

Listwise Direct Preference Optimization with Multi-Dimensional Preference Mixing

Recent alignment methods based on Direct Preference Optimization (DPO) reformulate preference learning as supervised optimization over pairwise comparisons, offering improved efficiency and stability over reinforcement learning from human feedback (RLHF). However, existing DPO-style methods implicitly assume a single fixed preference objective, which limits their ability to model the structured and sometimes conflicting nature of real-world human judgments that span multiple preference dimensions. In this work, we propose Listwise Direct Preference Optimization ($\lambda$-DPO), a unified framework that simultaneously improves supervision granularity and preference flexibility. Instead of collapsing multi-dimensional preference signals into a single ranking, $\lambda$-DPO constructs a mixture of listwise preference distributions weighted by a preference vector $\lambda$ on the probability simplex, enabling a single model to internalize a continuous spectrum of preference trade-offs. To further improve robustness, we introduce a performance-driven stochastic $\lambda$ scheduler that adaptively samples preference weights based on empirical downstream performance, explicitly mitigating the risks of misspecification inherent to static weighting schemes. We evaluate our method across multiple model families and scales on six widely used benchmarks. Experimental results show the consistent improvement against baselines.

cs.LG↗

arXiv2024

High-Q Cavity Interface for Color Centers in Thin Film Diamond

Quantum information technology offers the potential to realize unprecedented computational resources via secure channels capable of distributing entanglement between quantum computers. Diamond, as a host to atom-like defects with optically-accessible spin qubits, is a leading platform to realize quantum memory nodes needed to extend the reach of quantum links. Photonic crystal (PhC) cavities enhance light-matter interaction and are essential ingredients of an efficient interface between spins and photons that are used to store and communicate quantum information respectively. Despite great effort, however, the realization of visible PhC cavities with high quality factor (Q) and design flexibility is challenging in diamond. Here, we demonstrate one- and two-dimensional PhC cavities fabricated in recently developed thin-film diamonds, featuring Q-factors of 1.8x10$^5$ and 1.6x10$^5$, respectively, the highest Qs for visible PhC cavities realized in any material. Importantly, our fabrication process is simple and high-yield, based on conventional planar fabrication techniques, in contrast to previous approaches that rely on complex undercut methods. We also demonstrate fiber-coupled 1D PhC cavities with high photon extraction efficiency, and optical coupling between a single SiV center and such a cavity at 4K achieving a Purcell factor of 13. The demonstrated diamond thin-film photonic platform will improve the performance and scalability of quantum nodes and expand the range of quantum technologies.

quant-ph↗

arXiv2023

Microwave-based quantum control and coherence protection of tin-vacancy spin qubits in a strain-tuned diamond membrane heterostructure

Robust spin-photon interfaces in solids are essential components in quantum networking and sensing technologies. Ideally, these interfaces combine a long-lived spin memory, coherent optical transitions, fast and high-fidelity spin manipulation, and straightforward device integration and scaling. The tin-vacancy center (SnV) in diamond is a promising spin-photon interface with desirable optical and spin properties at 1.7 K. However, the SnV spin lacks efficient microwave control and its spin coherence degrades with higher temperature. In this work, we introduce a new platform that overcomes these challenges - SnV centers in uniformly strained thin diamond membranes. The controlled generation of crystal strain introduces orbital mixing that allows microwave control of the spin state with 99.36(9) % gate fidelity and spin coherence protection beyond a millisecond. Moreover, the presence of crystal strain suppresses temperature dependent dephasing processes, leading to a considerable improvement of the coherence time up to 223(10) $μ$s at 4 K, a widely accessible temperature in common cryogenic systems. Critically, the coherence of optical transitions is unaffected by the elevated temperature, exhibiting nearly lifetime-limited optical linewidths. Combined with the compatibility of diamond membranes with device integration, the demonstrated platform is an ideal spin-photon interface for future quantum technologies.

cond-mat.mes-hall↗

arXiv2023

Direct-bonded diamond membranes for heterogeneous quantum and electronic technologies

Diamond has superlative material properties for a broad range of quantum and electronic technologies. However, heteroepitaxial growth of single crystal diamond remains limited, impeding integration and evolution of diamond-based technologies. Here, we directly bond single-crystal diamond membranes to a wide variety of materials including silicon, fused silica, sapphire, thermal oxide, and lithium niobate. Our bonding process combines customized membrane synthesis, transfer, and dry surface functionalization, allowing for minimal contamination while providing pathways for near unity yield and scalability. We generate bonded crystalline membranes with thickness as low as 10 nm, sub-nm interfacial regions, and nanometer-scale thickness variability over 200 by 200 $\mu m^2$ areas. We measure spin coherence times $T_2$ for nitrogen-vacancy centers in bonded membranes of up to 623(21) $\mu$s, suitable for advanced quantum applications. We demonstrate multiple methods for integrating high quality factor nanophotonic cavities with the diamond heterostructures, highlighting the platform versatility in quantum photonic applications. Furthermore, we show that our ultra-thin diamond membranes are compatible with total internal reflection fluorescence (TIRF) microscopy, which enables interfacing coherent diamond quantum sensors with living cells while rejecting unwanted background luminescence. The processes demonstrated herein provide a full toolkit to synthesize heterogeneous diamond-based hybrid systems for quantum and electronic technologies.

physics.app-ph↗

arXiv2021

Tunable and Transferable Diamond Membranes for Integrated Quantum Technologies

Color centers in diamond are widely explored as qubits in quantum technologies. However, challenges remain in the effective and efficient integration of these diamond-hosted qubits in device heterostructures. Here, nanoscale-thick uniform diamond membranes are synthesized via "smart-cut" and isotopically (12C) purified overgrowth. These membranes have tunable thicknesses (demonstrated 50 nm to 250 nm), are deterministically transferable, have bilaterally atomically flat surfaces (Rq <= 0.3 nm), and bulk-diamond-like crystallinity. Color centers are synthesized via both implantation and in-situ overgrowth incorporation. Within 110 nm thick membranes, individual germanium-vacancy (GeV-) centers exhibit stable photoluminescence at 5.4 K and average optical transition linewidths as low as 125 MHz. The room temperature spin coherence of individual nitrogen-vacancy (NV-) centers shows Ramsey spin dephasing times (T2*) and Hahn echo times (T2) as long as 150 us and 400 us, respectively. This platform enables the straightforward integration of diamond membranes that host coherent color centers into quantum technologies.

physics.app-ph↗

arXiv2019

An optical and X-ray study of the contact binary, BH Cassiopeiae

We present the revised and high quality CCD time series observations in 2017 and 2018 of the W UMa-type contact binary BH~Cas. The combination of our new determinations of minimum times and those collected from the literatures reveal a long-term period increase with a rate of $\textit{dP/dt}\,=\,+3.27 \times 10^{-7} \textrm{d}\,\textrm{yr}^{-1}$, which can be accounted for by angular momentum transfer as the less-massive component loses its mass to the more massive one. One cyclical oscillation ($A'=0.00300$~days and $P'=20.09$~yrs) superimposed on the long-term period increase tendency is attributed to one additional object that orbits BH Cas. The low-resolution spectra of this source are observed at phase $\sim$0.0 and $\sim$0.5 in 2018 and 2019, and the spectral types are identified as K3V$\,\pm\,$1 and G8V$\,\pm\,$2. According to the luminosity contributions and spectra at these phases, the spectral type of the primary star is close to K3V$\,\pm\,$1. We infer a cool spot on the hotter and less-massive component to fit the asymmetry of light curves. In addition, changes in the location, temperature, and area of spots on the secondary star in different observations may indicate that the magnetic field activity of the secondary is more active than that of the primary. Such strong spot activities and optical flare are both detected at the same time from a W UMa-type star for the first time. With the \emph{XMM-Newton} observations taken in 2003 and 2012, the X-ray light curve is obtained and the X-ray spectra can be well described with a black-body or a bremsstrahlung model. The soft X-ray luminosity of BH Cas is estimated as 8.8\,--\,9.7$\,\times\,10^{30}$\,erg\,s$^{-1}$.

astro-ph.SR↗