arXiv subjects

Marin Soljačić

Publications and source records attributed to Marin Soljačić.

At least 37 records · Page 2Linked to original sources

arXiv2026

Active learning for photonic crystals

Active learning for photonic crystals explores the integration of analytic approximate Bayesian last layer neural networks (LL-BNNs) with uncertainty-driven sample selection to accelerate photonic band gap prediction. We employ an analytic LL-BNN formulation, corresponding to the infinite Monte Carlo sample limit, to obtain uncertainty estimates that are strongly correlated with the true predictive error on unlabeled candidate structures. These uncertainty scores drive an active learning strategy that prioritizes the most informative simulations during training. Applied to the task of predicting band gap sizes in two-dimensional, two-tone photonic crystals, our approach achieves up to a 2.7x reduction on average in required training data compared to a random sampling baseline while maintaining predictive accuracy. The efficiency gains arise from concentrating computational resources on high uncertainty regions of the design space rather than sampling uniformly. Given the substantial cost of full band structure simulations, especially in three dimensions, this data efficiency enables rapid and scalable surrogate modeling. Our results suggest that analytic LL-BNN based active learning can substantially accelerate topological optimization and inverse design workflows for photonic crystals, and more broadly, offers a general framework for data efficient regression across scientific machine learning domains.

physics.optics↗

arXiv2026

Supercollimating photonic crystal scintillators

Scintillators convert X-ray energy into visible or near-visible photons, enabling applications in high-energy particle detection and X-ray imaging. Increasing scintillator thickness improves X-ray absorption but degrades spatial resolution due to diffraction-induced lateral spreading of emitted light, resulting in a fundamental trade-off between detection efficiency and image resolution. Here, we propose a class of three-dimensional photonic crystal scintillators that overcomes this limitation through supercollimation, in which light propagates with suppressed diffraction. We develop a multiscale modeling framework that integrates nanophotonic band-structure simulations with Monte Carlo particle transport to quantitatively evaluate the performance of such scintillators. Our results show that supercollimating photonic crystal scintillators can enhance spatial resolution by up to an order of magnitude relative to conventional bulk scintillators of equal thickness. This improvement leads to a substantial increase in detector quantum efficiency (DQE), particularly at high spatial frequencies, enabling fine features to be preserved at reduced X-ray dose. We further demonstrate that comparable image quality can be achieved with approximately an order-of-magnitude lower X-ray dose. By directly engineering light transport within the bulk of the scintillator, this work establishes a nanophotonic route to simultaneously improving resolution and dose efficiency in X-ray imaging.

physics.optics↗

arXiv2026

Hybrid Nanophotonic Scintillators for Enhanced X-ray Absorption, Emission, and Time Resolution

Scintillators convert ionizing radiation into visible photons, enabling applications from cosmic ray detection to medical imaging. Two independent strategies for improving scintillator performance via nanoscale patterning have recently been demonstrated: engineering material properties to enhance absorption of ionizing radiation and integrating nanophotonic structures to enhance the spontaneous emission rate ("nanophotonic scintillators"). Here, we propose a nanophotonic scintillator that simultaneously enhances both the initial energy conversion and the spontaneous emission rate, by periodically stacking a fast-emitting scintillator and a visible-light-transparent material with strong X-ray attenuation ("stopping layer") to form a one-dimensional (1D) photonic crystal (PhC) scintillator. Photoelectric absorption in the stopping layer increases the number of photoelectrons that deposit energy in neighboring scintillator layers and contribute to scintillation. At the same time, the spontaneous emission rate is enhanced by the nanophotonic structuring itself. We design a 1D PhC comprising an organic scintillator and indium tin oxide (ITO) as the stopping layer and numerically simulate the enhancement in scintillation yield and decay rate. The total detected light output is enhanced by up to a factor of 700 compared to a bulk organic scintillator of equal thickness. We further investigate a 1D PhC structure integrating inorganic and organic scintillators for time-of-flight positron emission tomography (TOF-PET): replacing the non-scintillating stopping layer with an inorganic scintillator further increases the light yield, and the coincidence time resolution (CTR) is enhanced up to 3.5 times compared to a bulk inorganic scintillator of equal thickness. Our work presents a unified approach to improve key scintillation parameters within a single nanophotonic structure.

physics.optics↗

arXiv2026

Exceptional topology on nonorientable manifolds

We classify gapped phases and characteristic nodal points of non-Hermitian band structures on two-dimensional nonorientable parameter spaces. Such spaces arise in a wide range of physical systems in the presence of nonsymmorphic parameter space symmetries. For gapped phases, we find that nonorientable spaces provide a natural setting for exploring fundamental structural problems in braid group theory, such as torsion and conjugacy. Gapless systems, which host exceptional points (EPs), explicitly violate fermion doubling, even in two-band models. We demonstrate that EPs traversing the nonorientable parameter space exhibit non-Abelian charge inversion. These braided phases and their transitions leave distinct signatures in the form of bulk Fermi arc degeneracies, offering a concrete route toward experimental realization and verification.

cond-mat.mes-hall↗

arXiv2026

A Framework for Closed-Loop Robotic Assembly, Alignment and Self-Recovery of Precision Optical Systems

Robotic automation has transformed scientific workflows in domains such as chemistry and materials science, yet free-space optics, which is a high precision domain, remains largely manual. Optical systems impose strict spatial and angular tolerances, and their performance is governed by tightly coupled physical parameters, making generalizable automation particularly challenging. In this work, we present a robotics framework for the autonomous construction, alignment, and maintenance of precision optical systems. Our approach integrates hierarchical computer vision systems, optimization routines, and custom-built tools to achieve this functionality. As a representative demonstration, we perform the fully autonomous construction of a tabletop laser cavity from randomly distributed components. The system performs several tasks such as laser beam centering, spatial alignment of multiple beams, resonator alignment, laser mode selection, and self-recovery from induced misalignment and disturbances. By achieving closed-loop autonomy for highly sensitive optical systems, this work establishes a foundation for autonomous optical experiments for applications across technical domains.

cs.RO↗

arXiv2026

Gradient-based search of quantum phases: discovering unconventional fractional Chern insulators

The discovery and understanding of new quantum phases has time and again transformed both fundamental physics and technology, yet progress often relies on slow, intuition-based theoretical considerations or experimental serendipity. Here, we introduce a general gradient-based framework for targeted phase discovery. We define a differentiable function, dubbed "target-phase loss function", which encodes fingerprints of a quantum state, thereby recasting phase search as a tractable optimization problem in Hamiltonian space. The method is broadly applicable to a wide range of symmetry-broken and topological orders and can be interfaced with most many-body numerical solvers. As a demonstration, we apply it to spinless fermions on the kagome lattice using exact diagonalization and discover two distinctive fractional Chern insulators (FCIs): (i) at filling $ν= 1/3$, a "non-ideal" Abelian FCI whose band geometry lies far beyond the Landau-level mimicry paradigm and all recent generalizations; and (ii) at $ν= 1/2$, a non-Abelian FCI stabilized purely by finite-range two-body interactions. These results provide the first explicit realization of such types of FCIs and establish a versatile paradigm for systematic quantum-phase discovery.

cond-mat.str-el↗

arXiv2026

Unsupervised Decomposition and Recombination with Discriminator-Driven Diffusion Models

Decomposing complex data into factorized representations can reveal reusable components and enable synthesizing new samples via component recombination. We investigate this in the context of diffusion-based models that learn factorized latent spaces without factor-level supervision. In images, factors can capture background, illumination, and object attributes; in robotic videos, they can capture reusable motion components. To improve both latent factor discovery and quality of compositional generation, we introduce an adversarial training signal via a discriminator trained to distinguish between single-source samples and those generated by recombining factors across sources. By optimizing the generator to fool this discriminator, we encourage physical and semantic consistency in the resulting recombinations. Our method outperforms implementations of prior baselines on CelebA-HQ, Virtual KITTI, CLEVR, and Falcor3D, achieving lower FID scores and better disentanglement as measured by MIG and MCC. Furthermore, we demonstrate a novel application to robotic video trajectories: by recombining learned action components, we generate diverse sequences that significantly increase state-space coverage for exploration on the LIBERO benchmark.

cs.CV↗

arXiv2026

Structures resistant to Manipulation by all Wavefronts in two dimensions

Using light to manipulate small particles is a powerful tool with numerous practical applications across biophysics and nanotechnology. This experimental technique has achieved significant performance gains by employing shaped wavefronts, most commonly generated with spatial light modulators. Wavefront shaping has also enabled the manipulation of seemingly arbitrary objects beyond the reach of conventional beams. Contrary to this established assumption, we show here the existence of a wide variety of objects resistant to manipulation, even with the optimal wavefront shaping protocol. The counterintuitive shapes of these objects are found using inverse design in two dimensions, providing a foundation for their natural extension to three dimensions. Specifically, we show that the maximal pulling force is reduced by up to four orders of magnitude, and the maximal trapping stiffness is reduced by up to nearly two orders of magnitude. Our findings could prove useful for the development of micromachines that require a predictable mechanical response to arbitrary waves.

physics.optics↗

arXiv2026

Wavefront Engineering for Scintillation-Based Imaging

Recent research in nanophotonics for scintillation-based imaging has demonstrated promising improvements in scintillator performance. In parallel, advances in nanophotonics have enabled wavefront control through metasurfaces, a capability that has transformed fields such as microscopy by allowing tailored control of optical propagation. This naturally raises the following question, which we address in this perspective: can wavefront-control strategies be leveraged to improve scintillation-based imaging? To answer this question, we explore nanophotonic- and metasurface-enabled wavefront control in scintillators to mitigate image blurring arising from their intrinsically diffuse light emission. While depth-of-field extension in scintillation faces fundamental limitations absent in microscopy, this approach reveals promising avenues, including stacked scintillators, selective spatial-frequency enhancement, and X-ray energy-dependent imaging. These results clarify the key distinctions in adapting wavefront engineering to scintillation and its potential to enable tailored detection strategies.

physics.optics↗

arXiv2026

Essentially No Energy Barrier Between Independent Fermionic Neural Quantum State Minima

Neural quantum states (NQS) have proven highly effective in representing quantum many-body wavefunctions, but their loss landscape remains poorly understood and debated. Here, we demonstrate that the NQS loss landscape is more benign and similar to conventional deep learning than previously thought, exhibiting mode connectivity: independently trained NQS are connected by paths in parameter space with essentially no energy barrier. To construct these paths, we develop GeoNEB, a path optimizer integrating efficient stochastic reconfiguration with the nudged elastic band method for constructing minimum energy paths. For the strongly interacting six-electron quantum dot modeled by a $1.6$M-parameter Psiformer, we find two independent minima with expected energy barrier $\sim10^{-5}$ times smaller than the system's overall energy scale and $\sim10^{-3}$ times smaller than the linear path's barrier. The path respects physical symmetry in addition to achieving low energy, with the angular momentum remaining well quantized throughout. Our work is the first to construct optimized paths between independently trained NQS, and it suggests that the NQS loss landscape may not be as pathological as once feared.

cond-mat.dis-nn↗

arXiv2025

Symbolic Learning of Topological Bands in Photonic Crystals

Topological photonic crystals (PhCs) that support disorder-resistant modes, protected degeneracies, and robust transport have recently been explored for applications in waveguiding, optical isolation, light trapping, and lasing. However, designing PhCs with prescribed topological properties remains challenging because of the highly nonlinear mapping from the continuous real-space design of PhCs to the discrete output space of band topology. Here, we introduce a machine learning approach to address this problem, employing Kolmogorov--Arnold networks (KANs) to predict and inversely design the band symmetries of two-dimensional PhCs with two-fold rotational (C2) symmetry. We show that a single-hidden-layer KAN, trained on a dataset of C2-symmetric unit cells, achieves high accuracy in classifying the topological classes of the lowest lying bands. We use the symbolic regression capabilities of KANs to extract algebraic formulas that express the topological classes directly in terms of the Fourier components of the dielectric function. These formulas not only retain the full predictive power of the network but also provide novel insights and enable deterministic inverse design. Using this approach, we generate photonic crystals with target topological bands, achieving high accuracy even for high-contrast, experimentally realizable structures beyond the training domain.

physics.optics↗

arXiv2025

Refining Heuristic Predictors of Fractional Chern Insulators using Machine Learning

We develop an interpretable, data-driven framework to quantify how single-particle band geometry governs the stability of fractional Chern insulators (FCIs). Using large-scale exact diagonalization, we evaluate an FCI metric that yields a continuous spectral measure of FCI stability across parameter space. We then train Kolmogorov-Arnold networks (KANs) -- a recently developed interpretable neural architecture -- to regress this metric from two band-geometric descriptors: the trace violation $T$ and the Berry curvature fluctuations $σ_B$. Applied to spinless fermions at filling $ν=1/3$ in models on the checkerboard and kagome lattices, our approach yields compact analytical formulas that predict FCI stability with over $>80 \%$ accuracy in both regression and classification tasks, and remain reliable even in data-scarce regimes. The learned relations reveal model-dependent trends, clarifying the limits of Landau-level-mimicking heuristics. Our framework provides a general method for extracting simple, phenomenological "laws" that connect many-body phase stability to chosen physical descriptors, enabling rapid hypothesis formation and targeted design of quantum phases.

cond-mat.str-el↗

arXiv2025

AI-Driven Robotics for Optics

Optics is foundational to research in many areas of science and engineering, including nanophotonics, quantum information, materials science, biomedical imaging, and metrology. However, the design, assembly, and alignment of optical experiments remain predominantly manual, limiting throughput and reproducibility. Automating such experiments is challenging due to the strict, non-negotiable precision requirements and the diversity of optical configurations found in typical laboratories. Here, we introduce a platform that integrates generative artificial intelligence, computer vision, and robotics to automate free-space optical experiments. The platform translates user-defined goals into valid optical configurations, assembles them using a robotic arm, and performs micrometer-scale fine alignment using a robot-deployable tool. It then executes a range of automated measurements, including beam characterization, polarization mapping, and spectroscopy, with consistency surpassing that of human operators. This work demonstrates the first flexible, AI-driven automation platform for optics, offering a path towards remote operation, cloud labs, and high-throughput discovery in the optical sciences.

physics.optics↗

arXiv2025

L$^2$M: Mutual Information Scaling Law for Long-Context Language Modeling

We present a universal theoretical framework for understanding long-context language modeling based on a bipartite mutual information scaling law that we rigorously verify in natural language. We demonstrate that bipartite mutual information captures multi-token interactions distinct from and scaling independently of conventional two-point mutual information, and show that this provides a more complete characterization of the dependencies needed for accurately modeling long sequences. Leveraging this scaling law, we formulate the Long-context Language Modeling (L$^2$M) condition, which lower bounds the necessary scaling of a model's history state -- the latent variables responsible for storing past information -- for effective long-context modeling. We validate the framework and its predictions on transformer and state-space models. Our work provides a principled foundation to understand long-context modeling and to design more efficient architectures with stronger long-context capabilities, with potential applications beyond natural language.

cs.CL↗

arXiv2025

Learning simple heuristic rules for classifying materials based on chemical composition

In the past decade, there has been a significant interest in the use of machine learning approaches in materials science research. Conventional deep learning approaches that rely on complex, nonlinear models have become increasingly important in computational materials science due to their high predictive accuracy. In contrast to these approaches, we have shown in a recent work that a remarkably simple learned heuristic rule -- based on the concept of topogivity -- can classify whether a material is topological using only its chemical composition. In this paper, we go beyond the topology classification scenario by also studying the use of machine learning to develop simple heuristic rules for classifying whether a material is a metal based on chemical composition. Moreover, we present a framework for incorporating chemistry-informed inductive bias based on the structure of the periodic table. For both the topology classification and the metallicity classification tasks, we empirically characterize the performance of simple heuristic rules fit with and without chemistry-informed inductive bias across a wide range of training set sizes. We find evidence that incorporating chemistry-informed inductive bias can reduce the amount of training data required to reach a given level of test accuracy.

cond-mat.mtrl-sci↗

arXiv2025

Programmable control of the spatiotemporal quantum noise of light

Optoelectronic systems based on multiple modes of light can often exceed the performance of their single-mode counterparts. However, multimode nonlinear interactions often introduce considerable amounts of noise, limiting the ultimate performance of these systems. It is therefore crucial to develop ways to simultaneously control complex nonlinear interactions while also gaining control over their noise. Here, we show that noise buildup in nonlinear multimode systems can be strongly suppressed by controlling the input wavefront. We demonstrate this approach in a multimode fiber by using an active wavefront-shaping protocol to focus a region of high intensity - yet low intensity noise - at the output. Our programmable control of both the input and output reduces the beam noise by 12 dB beyond what linear attenuation achieves, reaching levels near the quantum shot-noise limit. We show that this is possible because the optimally shaped wavefront maximally decouples the output intensity fluctuations from the input laser fluctuations. These findings are supported by a new theoretical and simulation framework that efficiently captures spatiotemporal quantum noise dynamics in highly multimode nonlinear systems. Our results highlight the potential of programmable wavefront shaping to enable nonlinear multimode technologies that overcome noise buildup to operate at quantum-noise limits.

physics.optics↗

arXiv2025

Controlling X-ray emission with optical nanostructures

Nonlinear processes lie at the heart of many technologies such as frequency converters and entangled photon sources. Historically, observation and manipulation of these processes, for instance through nanostructures, has been limited to optical and lower frequencies. Recently, however, second-order nonlinear processes which couple X-ray and optical photons have been observed and used to probe the electronic structure and optical response of materials. Observing and controlling these processes remains challenging due to their low efficiency and the difficulty of fabricating devices with spatial features on the scale of X-ray wavelengths. Here, we show how optical nanostructures can be used to manipulate X-ray/optical nonlinear processes, using a quantum theory which describes these second-order nonlinear interactions. As an example, we show how photonic crystals shape both the spectral and spatial characteristics of X-rays emitted through X-ray to optical parametric down-conversion, leading to a fill-factor-normalized rate enhancement of 2.2 over an unstructured medium, in addition to control over the directionality of X-ray emission. The ability to control X-ray nonlinear processes may lead to more monochromatic, heralded X-ray sources, enhanced ghost imaging of lattice and electronic dynamics, and imaging and spectroscopy beyond the standard quantum limit. Our framework illuminates a path towards controlling quantum optical effects at X-ray frequencies.

physics.optics↗

arXiv2025

Tunable Nanophotonic Devices and Cavities based on a Two-Dimensional Magnet

Central to the field of nanophotonics is the ability to engineer the flow of light through nanoscale structures. These structures often have permanent working spectral ranges and optical properties that are fixed during fabrication. Quantum materials, with their correlated and intertwined degrees of freedom, offer a promising avenue for dynamically controlling photonic devices without altering their physical structure. Here, we fabricate photonic crystal slabs from CrSBr, a van der Waals antiferromagnetic semiconductor, and demonstrate unprecedented in situ control over their optical properties. Leveraging the combination of the exceptionally large refractive index of CrSBr near its excitonic resonances and its tunability via external fields, we achieve precise manipulation of photonic modes at near-visible and infrared wavelengths, showcasing a new paradigm for nanophotonic device design. The resulting guided resonances of the photonic crystal are tightly packed in the spectrum with very small mode volumes, are highly tunable via external magnetic fields, and exhibit high Q-factors exceeding 500. These resonances self-hybridize with the excitonic degrees of freedom, resulting in intrinsic strong light-matter coupling. Our findings underscore the potential of quantum materials for developing in situ tunable photonic elements and cavities.

physics.optics↗