arXiv subjects

Ju Li

Publications and source records attributed to Ju Li.

At least 55 records · Page 3Linked to original sources

arXiv2025

Frankenstein Optimizer: Harnessing the Potential by Revisiting Optimization Tricks

Gradient-based optimization drives the unprecedented performance of modern deep neural network models across diverse applications. Adaptive algorithms have accelerated neural network training due to their rapid convergence rates; however, they struggle to find ``flat minima" reliably, resulting in suboptimal generalization compared to stochastic gradient descent (SGD). By revisiting various adaptive algorithms' mechanisms, we propose the Frankenstein optimizer, which combines their advantages. The proposed Frankenstein dynamically adjusts first- and second-momentum coefficients according to the optimizer's current state to directly maintain consistent learning dynamics and immediately reflect sudden gradient changes. Extensive experiments across several research domains such as computer vision, natural language processing, few-shot learning, and scientific simulations show that Frankenstein surpasses existing adaptive algorithms and SGD empirically regarding convergence speed and generalization performance. Furthermore, this research deepens our understanding of adaptive algorithms through centered kernel alignment analysis and loss landscape visualization during the learning process. Code is available at https://github.com/acctouhou/Frankenstein_optimizer

cs.LG↗

arXiv2025

Ultrafast switchable polar and magnetic orders by nonlinear light-matter interaction

An outstanding challenge in materials science and physics is the harnessing of light for switching charge order in e.g., ferroelectrics. Here we propose a mechanism through which electrons in ferroelectric bilayers excited with light cause ionic structural transitions. Using perturbation theory within a many-body formalism, we show that the ionic coupling is mediated by a resonant change in electronic occupation functions, ultimately governed by the quantum geometric tensor (QGT) of the ground state. Furthermore, we show that such transitions are generally accompanied by multiferroic order switching. We demonstrate two examples of light-induced structural and polarization switching under this mechanism using first-principle calculations on bilayer CrI$_{3}$ and MoTe$_{2}$. We show that the two materials can switch between atomic stackings with a light intensity threshold of only 10-100 GW/cm$^2$, a value 1-3 orders of magnitude lower than that required by direct light-ion coupling thanks to the superior efficiency of resonant light-electron coupling. Since such switching is fast, highly controllable, contactless, and reversible, it is promising for use in optically controlled nonvolatile memory, nanophotonics and polar electronics.

cond-mat.mtrl-sci↗

arXiv2025

Collaborative AI Enhances Image Understanding in Materials Science

The Copilot for Real-world Experimental Scientist (CRESt) system empowers researchers to control autonomous laboratories through conversational AI, providing a seamless interface for managing complex experimental workflows. We have enhanced CRESt by integrating a multi-agent collaboration mechanism that utilizes the complementary strengths of the ChatGPT and Gemini models for precise image analysis in materials science. This innovative approach significantly improves the accuracy of experimental outcomes by fostering structured debates between the AI models, which enhances decision-making processes in materials phase analysis. Additionally, to evaluate the generalizability of this approach, we tested it on a quantitative task of counting particles. Here, the collaboration between the AI models also led to improved results, demonstrating the versatility and robustness of this method. By harnessing this dual-AI framework, this approach stands as a pioneering method for enhancing experimental accuracy and efficiency in materials research, with applications extending beyond CRESt to broader scientific experimentation and analysis.

cs.AI↗

arXiv2025

Contrastive Learning of English Language and Crystal Graphs for Multimodal Representation of Materials Knowledge

Artificial intelligence (AI) is increasingly used for the inverse design of materials, such as crystals and molecules. Existing AI research on molecules has integrated chemical structures of molecules with textual knowledge to adapt to complex instructions. However, this approach has been unattainable for crystals due to data scarcity from the biased distribution of investigated crystals and the lack of semantic supervision in peer-reviewed literature. In this work, we introduce a contrastive language-crystals model (CLaC) pre-trained on a newly synthesized dataset of 126k crystal structure-text pairs. To demonstrate the advantage of using synthetic data to overcome data scarcity, we constructed a comparable dataset extracted from academic papers. We evaluate CLaC's generalization ability through various zero-shot cross-modal tasks and downstream applications. In experiments, CLaC achieves state-of-the-art zero-shot generalization performance in understanding crystal structures, surpassing latest large language models.

cs.CL↗

arXiv2025

Strong Long-Wave Infrared Optical Response in a Topological Semiconductor with a Mexican Hat Band Structure

Light sources and photodetectors operating in the far- to mid-infrared (FIR/MIR) band ($8$-$12~\rm μm$, $0.1$-$0.15~\rm eV$) remain relatively poorly developed compared to their counterparts operating in the visible and near-infrared ranges, despite extensive application potential for thermal imaging, standoff sensing, and other technologies. This is attributable in part to the lack of narrow-gap materials ($<0.1~\rm eV$) with high optical gain and absorption. In this work, a narrow-gap semiconductor, $\rm Pb_{0.7}Sn_{0.3}Se$, is demonstrated to exhibit an optical response $>10\times$ larger than that of $\rm Hg_{x}Cd_{1-x}Te$ (MCT), the dominant material for FIR/MIR photodetectors. A previous theoretical investigation indicated that chalcogen $p$ and metal $d$ band inversion in this material creates a Mexican hat band structure (MHBS), which results in a dramatic increase in the joint density of states at the optical transition edge compared to typical semiconductors. This prediction is experimentally validated here using single-crystal specimens of $\rm Pb_{0.7}Sn_{0.3}Se$ measured using temperature-dependent spectroscopic ellipsometry over a wavelength range of $1.7$-$20~\rm μm$ ($0.73$-$0.062~\rm eV$). These measurements demonstrate a large enhancement in extinction coefficient and refractive index characteristic of a MHBS in the vicinity of the absorption edge, in agreement with theoretical predictions. The realization of topological semiconductors with a MHBS is expected to lead to high-efficiency detectors operating in the FIR/MID range.

physics.optics↗

arXiv2025

Linear and Nonlinear Edelstein Effects in Chiral Topological Semimetals

Recently, there has been growing interest in achieving on-demand control of magnetism through electrical and optical means. In this work, we provide first-principles predictions for the linear and nonlinear Edelstein effects (LEE and NLEE) in the chiral topological semimetal CoSi. The LEE and NLEE represent first- and second-order magnetic responses to external electric fields, enabling precise manipulation of magnetization via electrical and optical methods. We demonstrate that although both LEE and NLEE require time-reversal symmetry breaking, they can still be realized in non-magnetic materials, as time-reversal symmetry can be spontaneously broken by heat and dissipation, according to the second law of thermodynamics. Meanwhile, due to different inversion symmetry selection rules, the LEE and NLEE manifest opposite and identical signs in the two enantiomers of CoSi, respectively. We further quantify the magnitude of LEE and NLEE, showing that electrically or optically induced magnetization can reach 10 Bohr magneton per unit cell when the external electric field strength is comparable with the internal atomic electric field, which is on the order of 1 V/A. Our work offers a systematical approach for predicting the electrical and optical control of magnetism in real materials, paving the way for potential applications in areas such as spintronics and magnetic memories.

cond-mat.mtrl-sci↗

arXiv2024

Non-Hermitian Spin-Spin Interaction Mediated by Chiral Phonons

Non-Hermiticity and chirality are two fundamental properties known to give rise to various intriguing phenomena. However, the interplay between these properties has been rarely explored. In this work, we bridge this gap by introducing an off-diagonal non-Hermitian spin-spin interaction mediated by chiral phonons. This interaction arises from the spin-selectivity due to the locking between phonon momentum and angular momentum in chiral materials. The resulting non-Hermitian interaction mediated by the vacuum field of chiral phonons can reach the kHz range for electron spins and can be further enhanced by externally driven mechanical waves, potentially leading to observable effects in the quantum regime. Moreover, the long-range nature of phonon-mediated interactions enables the realization of the long-desired non-Hermitian interaction among multiple spins. The effect proposed in this work may have wide-ranging applications in cascaded quantum systems, non-Hermitian many-body physics, and non-Hermitian cooling.

quant-ph↗

arXiv2024

10 GHz Robust polarization modulation towards high-speed satellite-based quantum communication

In practical satellite-based quantum key distribution (QKD) systems, the preparation and transmission of polarization-encoding photons suffer from complex environmental effects and high channel-loss. Consequently, the hinge to enhancing the secure key rate (SKR) lies in achieving robust, low-error and high-speed polarization modulation. Although the schemes that realize self-compensation exhibit remarkable robustness. Their modulation speed is constrained to approximately 2 GHz to avoid the interaction between the electrical signal and the reverse optical pulses. Here we utilize the non-reciprocity of the lithium niobate modulators and eliminate the modulation on the reverse optical pulses. As this characteristic is widely available in the radio-frequency band, the modulation speed is no longer limited by the self-compensating optics and can be further increased. The measured average intrinsic QBER of the different polarization states at 10 GHz system repetition frequency is as low as 0.53% over 10 min without any compensation. And the experiment simulation shows that the proposed scheme extends the transmission distance to more than 350 km. Our work can be be efficient performed to the high-speed and high-loss satellite-based quantum communication scenario.

quant-ph↗

arXiv2024

Comment on "High-Power Collective Charging of a Solid-State Quantum Battery"

In the Letter [Physical Review Letters 120, 117702 (2018)], Ferraro \emph{et al.} claimed a quantum advantage in the Dicke quantum battery (QB), whereby $N$ two-level systems (TLS) are coupled to a common photonic mode of a cavity. They argued that compared with the so-called Rabi QB, the Dicke QB exhibits a $\sqrt{N}$ quantum enhancement in the charging power because of the entanglement created by the common photonic mode. In this Comment, however, we demonstrate that the apparent $\sqrt{N}$ enhancement actually comes from the stronger cavity electric (or magnetic) field under the setup discussed in [Physical Review Letters 120, 117702 (2018)]. This is a trivial classical effect, and there is no true "quantum advantage" in the charging power of the Dicke QB. While somewhat similar questions regarding the origin of the claimed "quantum advantage" have been raised before, here we would like to make it clear that the $\sqrt{N}$ enhancement in [Physical Review Letters 120, 117702 (2018)] is purely a classical effect, not attributable to quantum entanglement or collective phenomena. Therefore, we believe that using the term "quantum battery" in this context may be inappropriate and misleading.

quant-ph↗

arXiv2024

Generative Model for Constructing Reaction Path from Initial to Final States

Mapping the chemical reaction pathways and their corresponding activation barriers is a significant challenge in molecular simulation. Given the inherent complexities of 3D atomic geometries, even generating an initial guess of these paths can be difficult for humans. This paper presents an innovative approach that utilizes neural networks to generate initial guesses for reaction pathways based on the initial state and learning from a database of low-energy transition paths. The proposed method is initiated by inputting the coordinates of the initial state, followed by progressive alterations to its structure. This iterative process culminates in the generation of the guess reaction path and the coordinates of the final state. The method does not require one-the-fly computation of the actual potential energy surface, and is therefore fast-acting. The application of this geometry-based method extends to complex reaction pathways illustrated by organic reactions. Training was executed on the Transition1x dataset of organic reaction pathways. The results revealed the generation of reactions that bore substantial similarities with the test set of chemical reaction paths. The method's flexibility allows for reactions to be generated either to conform to predetermined conditions or in a randomized manner.

physics.comp-ph↗

arXiv2024

Efficient Quantum Transduction Using Anti-Ferromagnetic Topological Insulators

Transduction of quantum information between distinct quantum systems is an essential step in various applications, including quantum communications and quantum computing. However, mediating photons of vastly different frequencies and designing high-performance transducers are highly nontrivial, due to multifaceted and sometimes conflicting requirements. In this work, we first discuss some general principles for quantum transducer design, and then propose solid-state anti-ferromagnetic topological insulators to serve as particularly effective transducers. First, the anti-ferromagnetic order can minimize detrimental influences on nearby quantum systems caused by magnetic interactions. Second, topological insulators exhibit band-inversion, which can greatly enhance their optical responses. This property, coupled with robust spin-orbit coupling and high spin density, results in strong nonlinear interaction in magnetic topological insulators, thereby substantially improving transduction efficiency. Using MnBi2Te4 as an example, we discuss the potential experimental realization of quantum transduction based on magnetic topological materials. Particularly, we showcase that quantum transduction efficiency exceeding 90% can be achieved with modest experimental requirements, while the transduction bandwidth can reach the GHz range. The strong nonlinear photonic interactions in magnetic topological insulators can find diverse applications besides quantum transduction, such as quantum squeezing.

cond-mat.mtrl-sci↗

arXiv2024

Multi-task learning for molecular electronic structure approaching coupled-cluster accuracy

Machine learning (ML) plays an important role in quantum chemistry, providing fast-to-evaluate predictive models for various properties of molecules. However, most existing ML models for molecular electronic properties use density functional theory (DFT) databases as ground truth in training, and their prediction accuracy cannot surpass that of DFT. In this work, we developed a unified ML method for electronic structures of organic molecules using the gold-standard CCSD(T) calculations as training data. Tested on hydrocarbon molecules, our model outperforms DFT with the widely-used hybrid and double hybrid functionals in computational costs and prediction accuracy of various quantum chemical properties. As case studies, we apply the model to aromatic compounds and semiconducting polymers on both ground state and excited state properties, demonstrating its accuracy and generalization capability to complex systems that are hard to calculate using CCSD(T)-level methods.

physics.chem-ph↗

arXiv2024

Exponentially Enhanced non-Hermitian Cooling

Certain non-Hermitian systems exhibit the skin effect, whereby the wavefunctions become exponentially localized at one edge of the system. Such exponential amplification of wavefunction has received significant attention due to its potential applications in e.g., classical and quantum sensing. However, the opposite edge of the system, featured by the exponentially suppressed wavefunctions, remains largely unexplored. Leveraging this phenomenon, we introduce a non-Hermitian cooling mechanism, which is fundamentally distinct from traditional refrigeration or laser cooling techniques. Notably, non-Hermiticity will not amplify thermal excitations, but rather redistribute them. Hence, thermal excitations can be cooled down at one edge of the system, and the cooling effect can be exponentially enhanced by the number of auxiliary modes, albeit with a lower bound that depends on the dissipative interaction with the environment. Non-Hermitian cooling does not rely on intricate properties such as exceptional points or non-trivial topology, and it can apply to a wide range of Bosonic modes such as photons, phonons, magnons, etc.

quant-ph↗

arXiv2023

Blind quantum machine learning with quantum bipartite correlator

Distributed quantum computing is a promising computational paradigm for performing computations that are beyond the reach of individual quantum devices. Privacy in distributed quantum computing is critical for maintaining confidentiality and protecting the data in the presence of untrusted computing nodes. In this work, we introduce novel blind quantum machine learning protocols based on the quantum bipartite correlator algorithm. Our protocols have reduced communication overhead while preserving the privacy of data from untrusted parties. We introduce robust algorithm-specific privacy-preserving mechanisms with low computational overhead that do not require complex cryptographic techniques. We then validate the effectiveness of the proposed protocols through complexity and privacy analysis. Our findings pave the way for advancements in distributed quantum computing, opening up new possibilities for privacy-aware machine learning applications in the era of quantum technologies.

quant-ph↗

arXiv2023

Precise Fermi-level engineering in a topological Weyl semimetal via fast ion implantation

The precise controllability of the Fermi level is a critical aspect of quantum materials. For topological Weyl semimetals, there is a pressing need to fine-tune the Fermi level to the Weyl nodes and unlock exotic electronic and optoelectronic effects associated with the divergent Berry curvature. However, in contrast to 2D materials, where the Fermi level can be controlled through various techniques, the situation for bulk crystals beyond laborious chemical doping poses significant challenges. Here, we report the meV-level ultra-fine-tuning of the Fermi level of bulk topological Weyl semimetal TaP using accelerator-based high-energy hydrogen implantation and theory-driven planning. By calculating the desired carrier density and controlling the accelerator profiles, the Fermi level can be fine-tuned from 5 meV to only $\sim$0.5 meV (DFT calculations) away from the Weyl nodes. The Weyl nodes are preserved, while the carrier mobility is largely retained. Our work demonstrates the viability of this generic approach to tune the Fermi level in semimetal systems and could serve to achieve property fine-tuning for other bulk quantum materials with ultrahigh precision.

cond-mat.mtrl-sci↗

arXiv2023

Solid-State $\rm ^{229}Th$ Nuclear Laser with Two-Photon Pumping

The radiative excitation of the 8.3 eV isomeric state of thorium-229 is an outstanding challenge due to the lack of tunable far-ultraviolet (F-UV) sources. In this work, we propose an efficient two-photon pumping scheme for thorium-229 using the optonuclear quadrupolar effect, which only requires a 300 nm UV-B pumping laser. We further demonstrate that population inversion between the nuclear isomeric and ground states can be achieved at room temperature using a two-step pumping process. The nuclear laser, which has been pursued for decades, may be realized using a Watt-level UV-B pumping laser and ultrawide bandgap thorium compounds (e.g., $\rm ThF_4$, $\rm Na_2ThF_6$, or $\rm K_2ThF_6$) as the gain medium.

nucl-th↗

arXiv2023

1.5 million materials narratives generated by chatbots

The advent of artificial intelligence (AI) has enabled a comprehensive exploration of materials for various applications. However, AI models often prioritize frequently encountered materials in the scientific literature, limiting the selection of suitable candidates based on inherent physical and chemical properties. To address this imbalance, we have generated a dataset of 1,494,017 natural language-material paragraphs based on combined OQMD, Materials Project, JARVIS, COD and AFLOW2 databases, which are dominated by ab initio calculations and tend to be much more evenly distributed on the periodic table. The generated text narratives were then polled and scored by both human experts and ChatGPT-4, based on three rubrics: technical accuracy, language and structure, and relevance and depth of content, showing similar scores but with human-scored depth of content being the most lagging. The merger of multi-modality data sources and large language model (LLM) holds immense potential for AI frameworks to help the exploration and discovery of solid-state materials for specific applications.

cond-mat.mtrl-sci↗

arXiv2023

$μ$eV-deep neutron bound states in nanocrystals

The nuclear strong force induces the widely studied neutron scattering states and MeV-energy nuclear bound states. Whether this same interaction could lead to low-energy bound states for a neutron in the nuclear force field of a cluster of nuclei is an open question. Here, we computationally demonstrate the existence of -$μ$eV-level neutronic bound states originating from nuclear interaction in nanocrystals with a spatial extent of tens of nanometers. These negative-energy neutron wavefunctions depend on the size, dimension, and nuclear spin polarization of the nanoparticles, providing engineering degrees of freedom for the artificial neutronic "molecule".

quant-ph↗