arXiv subjects

Marin Soljačić

Publications and source records attributed to Marin Soljačić.

At least 55 records · Page 3Linked to original sources

arXiv2025

Alkali Intercalation of Moire Heterostructures for Low-Loss Plasmonics

Two-dimensional metals generically support gapless plasmons with wavelengths well below the wavelength of free-space radiation at the same frequency. Typically, however, this substantial confinement of electromagnetic energy is associated with commensurately high losses, and mitigating such losses may only be achieved through judicious band structure engineering near the Fermi level. In a clean system, an isolated, moderately flat, band at the Fermi level with sufficiently high carrier density can support a plasmon that is immune to propagation losses up to some order in the electron-phonon interaction. However, proposed materials that satisfy these criteria have been ferromagnetic, structurally unstable, or otherwise difficult to fabricate. Here, we propose a class of band structure engineered materials that evade these typical pitfalls -- Moire heterostructures of hexagonal boron nitride intercalated with alkali atoms. We find that only sodium atoms engender a sufficiently isolated band with plasmons lossless at first order in the electron-phonon interaction. We calculate higher order electron-phonon losses and find that at frequencies of about $1$eV the electron-phonon decay mechanism is negligible -- leading to a contribution to the decay rate of about 10^7 Hz in a small frequency range. We next calculate losses from the electron-electron interaction and find that this is the dominant process -- leading plasmons to decay to lower frequency plasmons at a rate of around 10^14 Hz.

cond-mat.mtrl-sci↗

arXiv2025

End-to-end design of multicolor scintillators for enhanced energy resolution in X-ray imaging

Scintillators have been widely used in X-ray imaging due to their ability to convert high-energy radiation into visible light, making them essential for applications such as medical imaging and high-energy physics. Recent advances in the artificial structuring of scintillators offer new opportunities for improving the energy resolution of scintillator-based X-ray detectors. Here, we present a three-bin energy-resolved X-ray imaging framework based on a three-layer multicolor scintillator used in conjunction with a physics-aware image postprocessing algorithm. The multicolor scintillator is able to preserve X-ray energy information through the combination of emission wavelength multiplexing and energy-dependent isolation of X-ray absorption in specific layers. The dominant emission color and the radius of the spot measured by the detector are used to infer the incident X-ray energy based on prior knowledge of the energy-dependent absorption profiles of the scintillator stack. Through ab initio Monte Carlo simulations, we show that our approach can achieve an energy reconstruction accuracy of 49.7%, which is only 2% below the maximum accuracy achievable with realistic scintillators. We apply our framework to medical phantom imaging simulations where we demonstrate that it can effectively differentiate iodine and gadolinium-based contrast agents from bone, muscle, and soft tissue.

physics.ins-det↗

arXiv2025

Quantized crystalline-electromagnetic responses in insulators

We introduce new classes of gapped topological phases characterized by quantized crystalline-electromagnetic responses, termed "multipolar Chern insulators". These systems are characterized by nonsymmorphic momentum-space symmetries and mirror symmetries, leading to quantization of momentum-weighted Berry curvature multipole moments. We construct lattice models for such phases and confirm their quantized responses through numerical calculations. These systems exhibit bound charge and momentum densities at lattice and magnetic defects, and currents induced by electric or time-varying strain fields. Our work extends the classification of topological matter by uncovering novel symmetry-protected topological phases with quantized responses.

cond-mat.mes-hall↗

arXiv2025

Nanophotonic thermal management in X-ray tubes

In X-ray tubes, more than 99% of the kilowatts of power supplied to generate X-rays via bremsstrahlung are lost in the form of heat generation in the anode. Therefore, thermal management is a critical barrier to the development of more powerful X-ray tubes with higher brightness and spatial coherence, which are needed to translate imaging modalities such as phase-contrast imaging to the clinic. In rotating anode X-ray tubes, the most common design, thermal radiation is a bottleneck that prevents efficient cooling of the anode$\unicode{x2014}$the hottest part of the device by far. We predict that nanophotonically patterning the anode of an X-ray tube enhances heat dissipation via thermal radiation, enabling it to operate at higher powers without increasing in temperature. The focal spot size, which is related to the spatial coherence of generated X-rays, can also be made smaller at a constant temperature. A major advantage of our "nanophotonic thermal management" approach is that in principle, it allows for complete control over the spectrum and direction of thermal radiation, which can lead to optimal thermal routing and improved performance.

physics.optics↗

arXiv2025

Noise immunity in quantum optical systems through non-Hermitian topology

Multimode nonlinear optical systems are highly valued for their ability to withstand large amounts of optical power, transmit data with high bandwidth, perform physical computations, generate quantum correlations, and much more. For many of these applications, both classical and quantum noise place limitations on important performance metrics. Moreover, it is well known that nonlinear systems tend to develop noise due to their sensitivity to initial conditions, thus motivating the development of general principles which can be used to control and mitigate noise in such systems. In this work, we show that non-reciprocity can enable the unidirectional flow of noise from the non-equilibrium steady states of nonlinear driven-dissipative systems. We demonstrate that this unidirectional flow results from the non-Hermitian skin effect (NHSE) for quantum noise. This effect provides a robust mechanism for isolating parts of the system from degradation due to excess injected noise. Our findings may lead to novel design principles for multimode amplifiers or oscillators that support high power handling while mitigating the buildup of noise. Moreover, our approach provides a new means to connect non-Hermitian topology with nonlinear many-body systems, which is expected to lead to topological descriptions of interacting states of light.

quant-ph↗

arXiv2025

Multimodal Foundation Models for Material Property Prediction and Discovery

Artificial intelligence is transforming computational materials science, improving the prediction of material properties, and accelerating the discovery of novel materials. Recently, publicly available material data repositories have grown rapidly. This growth encompasses not only more materials but also a greater variety and quantity of their associated properties. Existing machine learning efforts in materials science focus primarily on single-modality tasks, i.e. relationships between materials and a single physical property, thus not taking advantage of the rich and multimodal set of material properties. Here, we introduce Multimodal Learning for Materials (MultiMat), which enables self-supervised multi-modality training of foundation models for materials. We demonstrate our framework's potential using data from the Materials Project database on multiple axes: (i) MultiMat achieves state-of-the-art performance for challenging material property prediction tasks; (ii) MultiMat enables novel and accurate material discovery via latent space similarity, enabling screening for stable materials with desired properties; and (iii) MultiMat encodes interpretable emergent features that may provide novel scientific insights.

cs.LG↗

arXiv2025

KAN: Kolmogorov-Arnold Networks

Inspired by the Kolmogorov-Arnold representation theorem, we propose Kolmogorov-Arnold Networks (KANs) as promising alternatives to Multi-Layer Perceptrons (MLPs). While MLPs have fixed activation functions on nodes ("neurons"), KANs have learnable activation functions on edges ("weights"). KANs have no linear weights at all -- every weight parameter is replaced by a univariate function parametrized as a spline. We show that this seemingly simple change makes KANs outperform MLPs in terms of accuracy and interpretability. For accuracy, much smaller KANs can achieve comparable or better accuracy than much larger MLPs in data fitting and PDE solving. Theoretically and empirically, KANs possess faster neural scaling laws than MLPs. For interpretability, KANs can be intuitively visualized and can easily interact with human users. Through two examples in mathematics and physics, KANs are shown to be useful collaborators helping scientists (re)discover mathematical and physical laws. In summary, KANs are promising alternatives for MLPs, opening opportunities for further improving today's deep learning models which rely heavily on MLPs.

cs.LG↗

arXiv2025

Predicting band gap from chemical composition: A simple learned model for a material property with atypical statistics

In solid-state materials science, substantial efforts have been devoted to the calculation and modeling of the electronic band gap. While a wide range of ab initio methods and machine learning algorithms have been created that can predict this quantity, the development of new computational approaches for studying the band gap remains an active area of research. Here we introduce a simple machine learning model for predicting the band gap using only the chemical composition of the crystalline material. To motivate the form of the model, we first analyze the empirical distribution of the band gap, which sheds new light on its atypical statistics. Specifically, our analysis enables us to frame band gap prediction as a task of modeling a mixed random variable, and we design our model accordingly. Our model formulation incorporates thematic ideas from chemical heuristic models for other material properties in a manner that is suited towards the band gap modeling task. The model has exactly one parameter corresponding to each element, which is fit using data. To predict the band gap for a given material, the model computes a weighted average of the parameters associated with its constituent elements and then takes the maximum of this quantity and zero. The model provides heuristic chemical interpretability by intuitively capturing the associations between the band gap and individual chemical elements.

cond-mat.mtrl-sci↗

arXiv2024

Quantum sensitivity of parametric oscillators

Many quantum systems exhibit high sensitivity to their initial conditions, where microscopic quantum fluctuations can significantly influence macroscopic observables. Understanding how quantum states may influence the behavior of nonlinear dynamic systems may open new avenues in controlling light-matter interactions. To explore this issue, we analyze the sensitivity of a fundamental quantum optical process - parametric oscillation - to quantum initializations. Focusing on optical parametric oscillators (OPOs), we demonstrate that the quantum statistics of arbitrary initial states are imprinted in the early-stage dynamics and can persist in the steady-state probabilities. We derive the "quantum sensitivity" of parametric oscillators, linking the initial quantum state to the system's steady-state outcomes, highlighting how losses and parametric gain govern the system's quantum sensitivity. Moreover, we show that these findings extend beyond OPOs to a broader class of nonlinear systems, including Josephson junction based superconducting circuits. Our work opens the way to a new class of experiments that can test the sensitivity of macroscopic systems to quantum initial conditions and offers a pathway for controlling systems with quantum degrees of freedom.

quant-ph↗

arXiv2024

Topological quantum walk in synthetic non-Abelian gauge fields

We theoretically introduce synthetic non-Abelian gauge fields for topological quantum walks. The photonic mesh lattice configuration is generalized with polarization multiplexing to achieve a four-dimensional Hilbert space, based on which we provide photonic building blocks for realizing various quantum walks in non-Abelian gauge fields. It is found that SU(2) gauge fields can lead to Peierls substitution in both momenta and quasienergy. In one and two dimensions, we describe detailed photonic setups to realize topological quantum walk protocols whose Floquet winding numbers and Rudner-Lindner-Berg-Levin invariants can be effectively controlled by the gauge fields. Finally, we show how non-Abelian gauge fields facilitate convenient simulation of entanglement in conjunction with polarization-dependent and spatial-mode-dependent coin operations. Our results shed light on the study of synthetic non-Abelian gauge fields in photonic Floquet systems.

physics.optics↗

arXiv2024

QuanTA: Efficient High-Rank Fine-Tuning of LLMs with Quantum-Informed Tensor Adaptation

We propose Quantum-informed Tensor Adaptation (QuanTA), a novel, easy-to-implement, fine-tuning method with no inference overhead for large-scale pre-trained language models. By leveraging quantum-inspired methods derived from quantum circuit structures, QuanTA enables efficient high-rank fine-tuning, surpassing the limitations of Low-Rank Adaptation (LoRA)--low-rank approximation may fail for complicated downstream tasks. Our approach is theoretically supported by the universality theorem and the rank representation theorem to achieve efficient high-rank adaptations. Experiments demonstrate that QuanTA significantly enhances commonsense reasoning, arithmetic reasoning, and scalability compared to traditional methods. Furthermore, QuanTA shows superior performance with fewer trainable parameters compared to other approaches and can be designed to integrate with existing fine-tuning algorithms for further improvement, providing a scalable and efficient solution for fine-tuning large language models and advancing state-of-the-art in natural language processing.

cs.LG↗

arXiv2024

Nonreciprocal scintillation using one-dimensional magneto-optical photonic crystals

Scintillation describes the conversion of high-energy particles into light in transparent media and finds diverse applications such as high-energy particle detection and industrial and medical imaging. This process operates on multiple timescales, with the final radiative step consisting of spontaneous emission, which can be modeled within the framework of quasi-equilibrium fluctuational electrodynamics. Scintillation can therefore be controlled and enhanced via nanophotonic effects, which has been proposed and experimentally demonstrated. Such designs have thus far obeyed Lorentz reciprocity, meaning there is a direct equivalence between scintillation emission and absorption by the scintillator. However, scintillators that do not obey Lorentz reciprocity have not been explored, even though they represent a novel platform for probing emission which is both nonequilibrium and nonreciprocal in nature. In this work, we propose to harness nonreciprocity to achieve directional control of scintillation emission, granting an additional degree of control over scintillation. Such directionality of light output is important in improving collection efficiencies along the directions where detectors are located. We present the design of a nonreciprocal scintillator using a one-dimensional magnetophotonic crystal in the Voigt configuration. Our work demonstrates the potential of controlling nonequilibrium emission such as scintillation by breaking reciprocity and expands the space of nanophotonic design for achieving such control.

physics.optics↗

arXiv2024

OccamLLM: Fast and Exact Language Model Arithmetic in a Single Step

Despite significant advancements in text generation and reasoning, Large Language Models (LLMs) still face challenges in accurately performing complex arithmetic operations. Language model systems often enable LLMs to generate code for arithmetic operations to achieve accurate calculations. However, this approach compromises speed and security, and fine-tuning risks the language model losing prior capabilities. We propose a framework that enables exact arithmetic in a single autoregressive step, providing faster, more secure, and more interpretable LLM systems with arithmetic capabilities. We use the hidden states of a LLM to control a symbolic architecture that performs arithmetic. Our implementation using Llama 3 with OccamNet as a symbolic model (OccamLlama) achieves 100\% accuracy on single arithmetic operations ($+,-,\times,÷,\sin{},\cos{},\log{},\exp{},\sqrt{}$), outperforming GPT 4o with and without a code interpreter. Furthermore, OccamLlama outperforms GPT 4o with and without a code interpreter on average across a range of mathematical problem solving benchmarks, demonstrating that OccamLLMs can excel in arithmetic tasks, even surpassing much larger models. We will make our code public shortly.

cs.CL↗

arXiv2024

Weyl points on non-orientable manifolds

Weyl fermions are hypothetical chiral particles that can also manifest as excitations near three-dimensional band crossing points in lattice systems. These quasiparticles are subject to the Nielsen-Ninomiya "no-go" theorem when placed on a lattice, requiring the total chirality across the Brillouin zone to vanish. This constraint results from the topology of the (orientable) manifold on which they exist. Here, we ask to what extent the concepts of topology and chirality of Weyl points remain well-defined when the underlying manifold is non-orientable. We show that the usual notion of chirality becomes ambiguous in this setting, allowing for systems with a non-zero total chirality. This circumvention of the Nielsen-Ninomiya theorem stems from a generic discontinuity of the vector field whose zeros are Weyl points. Furthermore, we discover that Weyl points on non-orientable manifolds carry an additional $\mathbb{Z}_2$ topological invariant which satisfies a different no-go theorem. We implement such Weyl points by imposing a non-symmorphic symmetry in the momentum space of lattice models. Finally, we experimentally realize all aspects of their phenomenology in a photonic platform with synthetic momenta. Our work highlights the subtle but crucial interplay between the topology of quasiparticles and of their underlying manifold.

cond-mat.mes-hall↗

arXiv2024

End-to-End Optimization of Metasurfaces for Imaging with Compressed Sensing

We present a framework for the end-to-end optimization of metasurface imaging systems that reconstruct targets using compressed sensing, a technique for solving underdetermined imaging problems when the target object exhibits sparsity (i.e. the object can be described by a small number of non-zero values, but the positions of these values are unknown). We nest an iterative, unapproximated compressed sensing reconstruction algorithm into our end-to-end optimization pipeline, resulting in an interpretable, data-efficient method for maximally leveraging metaoptics to exploit object sparsity. We apply our framework to super-resolution imaging and high-resolution depth imaging with a phase-change material. In both situations, our end-to-end framework computationally discovers optimal metasurface structures for compressed sensing recovery, automatically balancing a number of complicated design considerations to select an imaging measurement matrix from a complex, physically constrained manifold with millions ofdimensions. The optimized metasurface imaging systems are robust to noise, significantly improving over random scattering surfaces and approaching the ideal compressed sensing performance of a Gaussian matrix, showing how a physical metasurface system can demonstrably approach the mathematical limits of compressed sensing.

eess.IV↗

arXiv2024

On a Novel Application of Wasserstein-Procrustes for Unsupervised Cross-Lingual Learning

The emergence of unsupervised word embeddings, pre-trained on very large monolingual text corpora, is at the core of the ongoing neural revolution in Natural Language Processing (NLP). Initially introduced for English, such pre-trained word embeddings quickly emerged for a number of other languages. Subsequently, there have been a number of attempts to align the embedding spaces across languages, which could enable a number of cross-language NLP applications. Performing the alignment using unsupervised cross-lingual learning (UCL) is especially attractive as it requires little data and often rivals supervised and semi-supervised approaches. Here, we analyze popular methods for UCL and we find that often their objectives are, intrinsically, versions of the Wasserstein-Procrustes problem. Hence, we devise an approach to solve Wasserstein-Procrustes in a direct way, which can be used to refine and to improve popular UCL methods such as iterative closest point (ICP), multilingual unsupervised and supervised embeddings (MUSE) and supervised Procrustes methods. Our evaluation experiments on standard datasets show sizable improvements over these approaches. We believe that our rethinking of the Wasserstein-Procrustes problem could enable further research, thus helping to develop better algorithms for aligning word embeddings across languages. Our code and instructions to reproduce the experiments are available at https://github.com/guillemram97/wp-hungarian.

cs.CL↗

arXiv2024

Stochastic logic in biased coupled photonic probabilistic bits

Optical computing often employs tailor-made hardware to implement specific algorithms, trading generality for improved performance in key aspects like speed and power efficiency. An important computing approach that is still missing its corresponding optical hardware is probabilistic computing, used e.g. for solving difficult combinatorial optimization problems. In this study, we propose an experimentally viable photonic approach to solve arbitrary probabilistic computing problems. Our method relies on the insight that coherent Ising machines composed of coupled and biased optical parametric oscillators can emulate stochastic logic. We demonstrate the feasibility of our approach by using numerical simulations equivalent to the full density matrix formulation of coupled optical parametric oscillators.

physics.optics↗

arXiv2024

TENG: Time-Evolving Natural Gradient for Solving PDEs With Deep Neural Nets Toward Machine Precision

Partial differential equations (PDEs) are instrumental for modeling dynamical systems in science and engineering. The advent of neural networks has initiated a significant shift in tackling these complexities though challenges in accuracy persist, especially for initial value problems. In this paper, we introduce the $\textit{Time-Evolving Natural Gradient (TENG)}$, generalizing time-dependent variational principles and optimization-based time integration, leveraging natural gradient optimization to obtain high accuracy in neural-network-based PDE solutions. Our comprehensive development includes algorithms like TENG-Euler and its high-order variants, such as TENG-Heun, tailored for enhanced precision and efficiency. TENG's effectiveness is further validated through its performance, surpassing current leading methods and achieving $\textit{machine precision}$ in step-by-step optimizations across a spectrum of PDEs, including the heat equation, Allen-Cahn equation, and Burgers' equation.

cs.LG↗