Duane S. Boning

Traditional architectural design space exploration (DSE) is highly inefficient, typically requiring tens of thousands of simulator evaluations across various optimization methods. This inefficiency arises because conventional methods treat the simulator as a black-box oracle. In contrast, human architects effectively guide exploration by reasoning through physical constraints, performance bottlenecks, data reuse, and workload structures. To bridge this gap, we introduce AgentDSE, a simulator-in-the-loop methodology driven by a general-purpose large language model (LLM) coding agent. AgentDSE automates this architectural-reasoning loop without requiring model fine-tuning, precomputed design databases, or domain-specific optimizer code. Across deep neural network (DNN) accelerator mapping, hardware/software co-design, and CPU cache-hierarchy optimization, AgentDSE achieves competitive or better design quality with up to two orders of magnitude fewer evaluations. AgentDSE also produces inspectable traces that surface architectural hypotheses, performance cliffs, implicit priors, and simulator artifacts, making every search decision traceable rather than buried in optimizer state.

Traditional architectural design space exploration (DSE) is highly inefficient, typically requiring tens of thousands of simulator evaluations across various optimization methods. This inefficiency arises because conventional methods treat the simulator as a black-box oracle. In contrast, human architects effectively guide exploration by reasoning through physical constraints, performance bottlenecks, data reuse, and workload structures. To bridge this gap, we introduce AgentDSE, a simulator-in-the-loop methodology driven by a general-purpose large language model (LLM) coding agent. AgentDSE automates this architectural-reasoning loop without requiring model fine-tuning, precomputed design databases, or domain-specific optimizer code. Across deep neural network (DNN) accelerator mapping, hardware/software co-design, and CPU cache-hierarchy optimization, AgentDSE achieves competitive or better design quality with up to two orders of magnitude fewer evaluations. AgentDSE also produces inspectable traces that surface architectural hypotheses, performance cliffs, implicit priors, and simulator artifacts, making every search decision traceable rather than buried in optimizer state.

Failure analysis, or forensic analysis of failed circuit elements, is an integral part of reliability assessment for the development of new materials and systems, particularly heterogeneously-integrated systems where multiple underlying failure mechanisms (modes) exist. To detect the presence of more than one failure mechanism, we usually require destructive, expensive tests (autopsies or analyses) on a number of failed units. Developing selective failure analysis strategies in the presence of limitations on time and resources is of the utmost importance. Using mixture and competing-risks models of Weibull distributions, this paper develops a probabilistic method to determine the minimum number of post-failure physical tests (analyses) required to detect the presence of more than one failure mode from the results of accelerated life tests, showing that (i) a successful detection of the presence of two distinct failure modes can be performed with 95% confidence for failure dataset sizes as small as 10 (without requiring exhaustive failure analysis on all units), and (ii) the minimum number of units to be analyzed (on average), expressed as a fraction of the sample size, decays exponentially with a power of the sample size, demonstrating the feasibility of the detection method. Our method can also handle component distributions other than Weibull. Results on three real experimental datasets are presented, along with statistical validation.

This article presents a new framework for optimizing, in the face of constraints on time and resources, selective failure analysis strategies for reliability modeling of systems where multiple failure mechanisms are present. We propose a new method to improve reliability parameter estimates for multiple failure mechanisms (under Weibull mixture and competing-risks models) via a dynamic, iterative strategy for selecting and determining the failure cause of a small number of failed units. Results are statistically validated for both uncensored and right-censored data. Our approach yields, on average, a log-likelihood increase of up to two orders of magnitude. This article also develops a procedure for optimal selection of accelerated life test conditions to enable accurate characterization of the parameters of physical failure mechanism models. This work provides a unified framework for strategic decision-making for reliability analysis, allowing the practitioner to maximize resource usage and minimize errors in the presence of experimental constraints.

Recent advances in large language models (LLMs) have demonstrated significant potential in hardware design automation, particularly in using natural language to synthesize Register-Transfer Level (RTL) code. Despite this progress, a gap remains between model capability and the demands of real-world RTL design, including syntax errors, functional hallucinations, and weak alignment to designer intent. Reinforcement Learning with Verifiable Rewards (RLVR) offers a promising approach to bridge this gap, as hardware provides executable and formally checkable signals that can be used to further align model outputs with design intent. However, in long, structured RTL code sequences, not all tokens contribute equally to functional correctness, and naïvely spreading gradients across all tokens dilutes learning signals. A key insight from our entropy analysis in RTL generation is that only a small fraction of tokens (e.g., always, if, assign, posedge) exhibit high uncertainty and largely influence control flow and module structure. To address these challenges, we present EARL, an Entropy-Aware Reinforcement Learning framework for Verilog generation. EARL performs policy optimization using verifiable reward signals and introduces entropy-guided selective updates that gate policy gradients to high-entropy tokens. This approach preserves training stability and concentrates gradient updates on functionally important regions of code. Our experiments on VerilogEval and RTLLM show that EARL improves functional pass rates over prior LLM baselines by up to 14.7%, while reducing unnecessary updates and improving training stability. These results indicate that focusing RL on critical, high-uncertainty tokens enables more reliable and targeted policy improvement for structured RTL code generation.

Probabilistic Time Series Forecasting (PTSF) plays a critical role in domains requiring accurate and uncertainty-aware predictions for decision-making. However, existing methods offer suboptimal distribution modeling and suffer from a mismatch between training and evaluation metrics. Surprisingly, we found that augmenting a strong point estimator with a zero-mean Gaussian, whose standard deviation matches its training error, can yield state-of-the-art performance in PTSF. In this work, we propose RDIT, a plug-and-play framework that combines point estimation and residual-based conditional diffusion with a bidirectional Mamba network. We theoretically prove that the Continuous Ranked Probability Score (CRPS) can be minimized by adjusting to an optimal standard deviation and then derive algorithms to achieve distribution matching. Evaluations on eight multivariate datasets across varied forecasting horizons demonstrate that RDIT achieves lower CRPS, rapid inference, and improved coverage compared to strong baselines.

Heterogeneous photonic-electronic systems, such as co-packaged optics and photonic-electronic artificial intelligence (AI) accelerators, are rapidly gaining traction but also pose significant design challenges due to distinct design methodologies. Digital and analog electronics are typically described using hardware description languages and SPICE, respectively, whereas photonic devices and systems are represented using permittivity tensors on the Yee grid and the Scattering matrix formulation. This disparity necessitates an end-to-end photonic-electronic cosimulation tool to streamline co-design. Most preliminary cosimulation approaches rely on translating photonic compact models into Verilog-A or SPICE models to simulate everything there, which not only introduces the additional complexity of model conversion but also has potential numerical stability problems. Additionally, another critical functionality missing from the current implementation is enabling gradient calculation in these co-simulators, which will be crucial for end-to-end gradient-based electronic-photonic system optimization. To address these challenges, we introduce SPIPE, a differentiable SPICE-level co-simulation framework for integrated photonic-electronic systems. SPIPE is the first co-simulator to overcome model conversion issues and to provide differentiability. Numerical experiments on several circuits confirm the accuracy of SPIPE when compared to analytical solutions and real-world experimental data. Furthermore, in cases where existing simulators are applicable, SPIPE achieves a runtime reduction of 2!+85!A compared to an industry-standard simulator. SPIPE features an integrated simulation interface with a low usage barrier, opening avenues for more accessible and effective photonic-electronic co-design. SPIPE is open sourced: https://github.com/zhengqigao/spipe.

DONNs leverage light propagation for efficient analog AI and signal processing. Advances in nanophotonic fabrication and metasurface-based wavefront engineering have opened new pathways to realize high-capacity DONNs across various spectral regimes. Training such DONN systems to determine the metasurface structures remains challenging. Heuristic methods are fast but oversimplify metasurfaces modulation, often resulting in physically unrealizable designs and significant performance degradation. Simulation-in-the-loop optimizes implementable metasurfaces via adjoint methods, but is computationally prohibitive and unscalable. To address these limitations, we propose SP2RINT, a spatially decoupled, progressive training framework that formulates DONN training as a PDE-constrained learning problem. Metasurface responses are first relaxed into freely trainable transfer matrices with a banded structure. We then progressively enforce physical constraints by alternating between transfer matrix training and adjoint-based inverse design, avoiding per-iteration PDE solves while ensuring final physical realizability. To further reduce runtime, we introduce a physics-inspired, spatially decoupled inverse design strategy based on the natural locality of field interactions. This approach partitions the metasurface into independently solvable patches, enabling scalable and parallel inverse design with system-level calibration. Evaluated across diverse DONN training tasks, SP2RINT achieves digital-comparable accuracy while being 1825 times faster than simulation-in-the-loop approaches. By bridging the gap between abstract DONN models and implementable photonic hardware, SP2RINT enables scalable, high-performance training of physically realizable meta-optical neural systems. Our code is available at https://github.com/ScopeX-ASU/SP2RINT

Reinforcement learning (RL) has recently emerged as a compelling approach for enhancing the reasoning capabilities of large language models (LLMs), where an LLM generator serves as a policy guided by a verifier (reward model). However, current RL post-training methods for LLMs typically use verifiers that are fixed (rule-based or frozen pretrained) or trained discriminatively via supervised fine-tuning (SFT). Such designs are susceptible to reward hacking and generalize poorly beyond their training distributions. To overcome these limitations, we propose Tango, a novel framework that uses RL to concurrently train both an LLM generator and a verifier in an interleaved manner. A central innovation of Tango is its generative, process-level LLM verifier, which is trained via RL and co-evolves with the generator. Importantly, the verifier is trained solely based on outcome-level verification correctness rewards without requiring explicit process-level annotations. This generative RL-trained verifier exhibits improved robustness and superior generalization compared to deterministic or SFT-trained verifiers, fostering effective mutual reinforcement with the generator. Extensive experiments demonstrate that both components of Tango achieve state-of-the-art results among 7B/8B-scale models: the generator attains best-in-class performance across five competition-level math benchmarks and four challenging out-of-domain reasoning tasks, while the verifier leads on the ProcessBench dataset. Remarkably, both components exhibit particularly substantial improvements on the most difficult mathematical reasoning problems. Code is at: https://github.com/kaiwenzha/rl-tango.

Inverse design has emerged as a transformative approach for photonic device optimization, enabling exploration of high-dimensional, non-intuitive design spaces to create ultra-compact, high-performance devices, advancing photonic inte-grated circuits (PICs) in computing and interconnects. However, practical challenges, such as suboptimal device performance compared to manual designs, limited manufacturability, high sensitivity to variations, computational inefficiency, and lack of interpretability, have hindered its adoption in commercial hardware. Recent advancements in AI-assisted photonic simulation and design offer transformative potential, accelerating simulations and design generation by orders of magnitude over traditional numerical methods. Despite these breakthroughs, the lack of an open-source, standardized infrastructure and evaluation bench-mark limits accessibility and cross-disciplinary collaboration. To address this, we introduce MAPS, a multi-fidelity AI-augmented photonic simulation and inverse design infrastructure, designed to bridge this gap. MAP S features three synergistic components: 1 MAPS-Data: A dataset acquisition framework for generating multi-fidelity, richly labeled device designs using intelligent sampling strategies, providing high-quality data for AI-for-optics research. 2 MAPS-Train: A flexible AI-for-photonics training framework, offering hierarchical data loading pipeline, customizable model construction, support for data- and physics-driven losses, and comprehensive evaluation metrics. 3 MAPS-InvDes: An advanced adjoint method-based inverse design toolkit that abstracts complex physics but exposes flexible optimization steps, integrates pre-trained AI models, and incorporates fabrication-aware variation models, for real-world applicability. This infrastructure MAPS provides a unified, open-source platform for developing, benchmarking, and advancing AI-assisted photonic design workflows, accelerating innovation in photonic hardware optimization and scientific machine learning.

Nanophotonic device design aims to optimize pho-tonic structures to meet specific requirements across various applications. Inverse design has unlocked non-intuitive, high-dimensional design spaces, enabling the discovery of compact, high-performance device topologies beyond traditional heuristic or analytic methods. The adjoint method, which calculates analytical gradients for all design variables using just two electromagnetic simulations, enables efficient navigation of this complex space. However, many inverse-designed structures, while numerically plausible, are difficult to fabricate and highly sensitive to physical variations, limiting their practical use. The discrete material distributions with numerous local-optimal structures also pose significant optimization challenges, often causing gradient-based methods to converge on suboptimal designs. In this work, we formulate inverse design as a fabrication-restricted, discrete, prob-abilistic optimization problem and introduce BOSON<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup>, an end-to-end, adaptive, variation-aware subspace optimization framework to address the challenges of manufacturability, robustness, and optimizability. We explicitly consider the fabrication process and differentiably optimize the design in the fabricable subspace. To overcome optimization difficulty, we propose dense target-enhanced gradient flows to mitigate misleading local optima and introduce a conditional subspace optimization strategy to create high-dimensional tunnels to escape local optima. Furthermore, we significantly reduce the prohibitive runtime associated with optimizing across exponential variation samples through an adaptive sampling-based robust optimization method, ensuring both efficiency and variation robustness. On three representative photonic device benchmarks, our proposed inverse design methodology BOSON<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup> delivers fabricable structures and achieves the best convergence and performance under realistic variations, outperforming prior arts with 74.3% post-fabrication performance.

Time series data are everywhere -- from finance to healthcare -- and each domain brings its own unique complexities and structures. While advanced models like Transformers and graph neural networks (GNNs) have gained popularity in time series forecasting, largely due to their success in tasks like language modeling, their added complexity is not always necessary. In our work, we show that simple feedforward neural networks (SFNNs) can achieve performance on par with, or even exceeding, these state-of-the-art models, while being simpler, smaller, faster, and more robust. Our analysis indicates that, in many cases, univariate SFNNs are sufficient, implying that modeling interactions between multiple series may offer only marginal benefits. Even when inter-series relationships are strong, a basic multivariate SFNN still delivers competitive results. We also examine some key design choices and offer guidelines on making informed decisions. Additionally, we critique existing benchmarking practices and propose an improved evaluation protocol. Although SFNNs may not be optimal for every situation (hence the ``almost'' in our title) they serve as a strong baseline that future time series forecasting methods should always be compared against.

Inverse design has emerged as a transformative approach for photonic device optimization, enabling the exploration of high-dimensional, non-intuitive design spaces to create ultra-compact devices and advance photonic integrated circuits (PICs) in computing and interconnects. However, practical challenges, such as suboptimal device performance, limited manufacturability, high sensitivity to variations, computational inefficiency, and lack of interpretability, have hindered its adoption in commercial hardware. Recent advancements in AI-assisted photonic simulation and design offer transformative potential, accelerating simulations and design generation by orders of magnitude over traditional numerical methods. Despite these breakthroughs, the lack of an open-source, standardized infrastructure and evaluation benchmark limits accessibility and cross-disciplinary collaboration. To address this, we introduce MAPS, a multi-fidelity AI-augmented photonic simulation and inverse design infrastructure designed to bridge this gap. MAPS features three synergistic components: (1) MAPS-Data: A dataset acquisition framework for generating multi-fidelity, richly labeled devices, providing high-quality data for AI-for-optics research. (2) MAPS-Train: A flexible AI-for-photonics training framework offering a hierarchical data loading pipeline, customizable model construction, support for data- and physics-driven losses, and comprehensive evaluations. (3) MAPS-InvDes: An advanced adjoint inverse design toolkit that abstracts complex physics but exposes flexible optimization steps, integrates pre-trained AI models, and incorporates fabrication variation models. This infrastructure MAPS provides a unified, open-source platform for developing, benchmarking, and advancing AI-assisted photonic design workflows, accelerating innovation in photonic hardware optimization and scientific machine learning.

Optical simulation plays an important role in photonic hardware design flow. The finite-difference time-domain (FDTD) method is widely adopted to solve time-domain Maxwell equations. However, FDTD is known for its prohibitive runtime cost as it iteratively solves Maxwell equations and takes minutes to hours to simulate a single device. Recently, AI has been applied to realize orders-of-magnitude speedup in partial differential equation solving. However, AI-based FDTD solvers for photonic devices have not been clearly formulated. Directly applying off-the-shelf models to predict the optical field dynamics shows unsatisfying fidelity and efficiency since the model primitives are agnostic to the unique physical properties of Maxwell equations and lack algorithmic customization. In this work, we thoroughly investigate the synergy between neural operator designs and the physical property of Maxwell equations and introduce a physics-inspired AI-based FDTD prediction framework PIC2O-Sim. PIC2O-Sim features a causality-aware dynamic convolutional neural operator as its backbone model that honors the space–time causality constraints via careful receptive field configuration and explicitly captures the permittivity-dependent light propagation behavior via an efficient dynamic convolution operator. Meanwhile, we explore the trade-offs among prediction scalability, fidelity, and efficiency via a multi-stage partitioned time-bundling technique in autoregressive prediction. Multiple key techniques have been introduced to mitigate iterative error accumulation while maintaining efficiency advantages during autoregressive field prediction. Extensive evaluations on three challenging photonic device simulation tasks have shown the superiority of our PIC2O-Sim method, showing 51.2% lower roll-out prediction error, 23.5 times fewer parameters than state-of-the-art neural operators, providing 133–310× or 31–89× higher simulation speed than an open-source single-process or eight-process parallel FDTD numerical solver.

Guidance techniques are simple yet effective for improving conditional generation in diffusion models. Albeit their empirical success, the practical implementation of guidance diverges significantly from its theoretical motivation. In this paper, we reconcile this discrepancy by replacing the scaled marginal distribution target, which we prove theoretically invalid, with a valid scaled joint distribution objective. Additionally, we show that the established guidance implementations are approximations to the intractable optimal solution under no future foresight constraint. Building on these theoretical insights, we propose rectified gradient guidance (REG), a versatile enhancement designed to boost the performance of existing guidance methods. Experiments on 1D and 2D demonstrate that REG provides a better approximation to the optimal solution than prior guidance techniques, validating the proposed theoretical framework. Extensive experiments on class-conditional ImageNet and text-to-image generation tasks show that incorporating REG consistently improves FID and Inception/CLIP scores across various settings compared to its absence.

This paper introduces KirchhoffNet, a novel class of neural network models inspired by the principles of analog electronic circuitry, specifically Kirchhoff’s laws. KirchhoffNet operates as an analog circuit, where the network input is represented by initial node voltages, and the output corresponds to the node voltages at a specific time. The dynamics of the node voltages are governed by learnable parameters on the edges, and the evolution of these voltages follows a system of ordinary differential equations (ODEs). Despite the absence of traditional neural network components such as convolutional layers, KirchhoffNet achieves outstanding performance across a wide range of machine learning tasks. We further demonstrate that KirchhoffNet is capable of computing diffusion models, making it a promising candidate for accelerating modern generative AI applications. Most notably, KirchhoffNet can be implemented as a high-speed & low-power analog integrated circuit, which introduces a compelling advantage: irrespective of the number of parameters in the network, its on-chip forward calculation can always be completed within a short time. This property makes KirchhoffNet a highly attractive and scalable paradigm for implementing large-scale neural networks, opening new avenues in the realm of analog neural networks for artificial intelligence.

Nanophotonic device design aims to optimize photonic structures to meet specific requirements across various applications. Inverse design has unlocked non-intuitive, high-dimensional design spaces, enabling the discovery of high-performance devices beyond heuristic or analytic methods. The adjoint method, which calculates gradients for all variables using just two simulations, enables efficient navigation of this complex space. However, many inverse-designed structures, while numerically plausible, are difficult to fabricate and sensitive to variations, limiting their practical use. The discrete nature with numerous local-optimal structures also pose significant optimization challenges, often causing gradient-based methods to converge on suboptimal designs. In this work, we formulate inverse design as a fabrication-restricted, discrete, probabilistic optimization problem and introduce BOSON-1, an end-to-end, variation-aware subspace optimization framework to address the challenges of manufacturability, robustness, and optimizability. To overcome optimization difficulty, we propose dense target-enhanced gradient flows to mitigate misleading local optima and introduce a conditional subspace optimization strategy to create high-dimensional tunnels to escape local optima. Furthermore, we significantly reduce the runtime associated with optimizing across exponential variation samples through an adaptive sampling-based robust optimization, ensuring both efficiency and variation robustness. On three representative photonic device benchmarks, our proposed inverse design methodology BOSON^-1 delivers fabricable structures and achieves the best convergence and performance under realistic variations, outperforming prior arts with 74.3% post-fabrication performance. We open-source our codes at https://github.com/ScopeX-ASU/BOSON.

In this paper, we leverage a foundational principle of analog electronic circuitry, Kirchhoff's current and voltage laws, to introduce a distinctive class of neural network models termed KirchhoffNet. Essentially, KirchhoffNet is an analog circuit that can function as a neural network, utilizing its initial node voltages as the neural network input and the node voltages at a specific time point as the output. The evolution of node voltages within the specified time is dictated by learnable parameters on the edges connecting nodes. We demonstrate that KirchhoffNet is governed by a set of ordinary differential equations (ODEs), and notably, even in the absence of traditional layers (such as convolution layers), it attains state-of-the-art performances across diverse and complex machine learning tasks. Most importantly, KirchhoffNet can be potentially implemented as a low-power analog integrated circuit, leading to an appealing property --- irrespective of the number of parameters within a KirchhoffNet, its on-chip forward calculation can always be completed within a short time. This characteristic makes KirchhoffNet a promising and fundamental paradigm for implementing large-scale neural networks, opening a new avenue in analog neural networks for AI. Our source code and model checkpoints are publicly available: https://github.com/zhengqigao/kirchhoffnet.

Abstract This paper presents a method for statistical analysis of ball grid array (BGA) pin degradation in packaged devices. A novel testing system is demonstrated featuring parallel monitoring of numerous pins, fast thermal cycling, and precise degradation data generation even in moderate stresses. The resistance change data due to temperature cycling stress is compared to stress on the pin due to its location on the chip. The results are vital for localizing vulnerabilities in pins that are not location-specific in the chip. The methods introduced in this study can be expanded to analyze advanced packaging assemblies.

Optimization methods are frequently exploited in the design of silicon photonic devices. In this paper, we demonstrate that pushing the objective function to its minimum during optimization often results in devices that gradually become more sensitive to perturbations of design variables. The dominant strategy of selecting the design with the smallest objective function can lead to fabrication failure or yield loss due to manufacturing process variations. To address this issue, we propose an intuitive selection criterion that can identify designs not only possessing small objective functions but that are also robust to variations. Our simulation results on the Y-splitter, direction coupler, and bent waveguide designs demonstrate that the proposed method can achieve 2x higher coverage of robust designs with almost negligible run time, compared to the two baseline methods.

Estimating the reliability of electronic devices involves identification of failure mechanisms and prediction of lifetimes. For parameter estimation and failure mode identification in Weibull competing-risks models, a differential-evolution-based global optimization approach has recently been developed, with the superiority of that approach demonstrated over the best-known local methods for the problem. In an effort to design a method faster than differential evolution for this problem, the present paper develops a new type of expectation maximization (EM) algorithm that is capable of handling bound constraints while optimizing the parameters of the Weibull component distributions. The differential-evolution-based approach guarantees a feasible, but not necessarily high-quality, solution in every run, while the proposed method offers no such guarantee. Despite this lack of guarantee, the proposed method is seen to produce results of a quality highly competitive with differential evolution. Numerical results on ten test cases, based on three real test datasets and two synthetic datasets, show that in terms of solution quality, the proposed method is competitive with differential evolution, while offering an average savings of about 64% in the computation time. Comparative performance analyses with the standard EM algorithm and the best-known local method L-BFGS-B are also provided. The numerical results are statistically validated. A new approach to model improvement via selective failure analysis is demonstrated as an application of the proposed algorithm. The proposed algorithm has the potential to be used for general-purpose likelihood maximization involving latent variables in diverse domains.

The finite-difference time-domain (FDTD) method, which is important in photonic hardware design flow, is widely adopted to solve time-domain Maxwell equations. However, FDTD is known for its prohibitive runtime cost, taking minutes to hours to simulate a single device. Recently, AI has been applied to realize orders-of-magnitude speedup in partial differential equation (PDE) solving. However, AI-based FDTD solvers for photonic devices have not been clearly formulated. Directly applying off-the-shelf models to predict the optical field dynamics shows unsatisfying fidelity and efficiency since the model primitives are agnostic to the unique physical properties of Maxwell equations and lack algorithmic customization. In this work, we thoroughly investigate the synergy between neural operator designs and the physical property of Maxwell equations and introduce a physics-inspired AI-based FDTD prediction framework PIC2O-Sim which features a causality-aware dynamic convolutional neural operator as its backbone model that honors the space-time causality constraints via careful receptive field configuration and explicitly captures the permittivity-dependent light propagation behavior via an efficient dynamic convolution operator. Meanwhile, we explore the trade-offs among prediction scalability, fidelity, and efficiency via a multi-stage partitioned time-bundling technique in autoregressive prediction. Multiple key techniques have been introduced to mitigate iterative error accumulation while maintaining efficiency advantages during autoregressive field prediction. Extensive evaluations on three challenging photonic device simulation tasks have shown the superiority of our PIC2O-Sim method, showing 51.2% lower roll-out prediction error, 23.5 times fewer parameters than state-of-the-art neural operators, providing 300-600x higher simulation speed than an open-source FDTD numerical solver.

Accurate estimation of rare failure occurrence probability is crucial for ensuring the proper and reliable functioning of integrated circuits (ICs). Conventional Monte Carlo methods are inefficient, demanding an exorbitant number of samples to achieve reliable estimates. Inspired by the exact sampling capabilities of normalizing flows, we revisit this problem and propose normalizing flow assisted importance sampling, termed NOFIS. NOFIS first learns a sequence of proposal distributions associated with predefined nested subset events by minimizing KL divergence losses. Next, it estimates the rare event probability by utilizing importance sampling in conjunction with the last proposal. The efficacy of our NOFIS method is substantiated through comprehensive qualitative visualizations, affirming the optimality of the learned proposal distribution, as well as 10 quantitative experiments, which highlight NOFIS's superior accuracy over baseline approaches.

We propose an automatic approach for implementing light processing functions on programmable photonic integrated circuits. Our approach offers two unprecedented and significant advantages. Firstly, it is generalizable to any topology, including triangular, square, and hexagonal meshes, and any linear light processing function with a frequency domain representation. Secondly, it achieves power efficiency by minimizing power consumption, even when taking into account non-ideal thermal crosstalk. Our key is to employ a differentiable scattering matrix simulation and formulate functional synthesis as an L1-regularized optimization problem. Our proposed approach demonstrates superior power efficiency and synthesized result quality compared to several common baselines, with a slight and manageable increase in algorithm runtime.

We present a new method for preprocessing semiconductor fabrication sensor signals that improves fault detection model performance. Machine learning has allowed for advances in fault detection and classification, but there are still difficulties in applying these techniques to the monitoring of fabrication processes when some variation in processing time is acceptable. The new method uses domain knowledge - specifically, process recipe steps - to create constraints that better align signals along the time dimension, which addresses this problem of nonlinear siznalaliznment,

Programmable photonic integrated circuits (PPICs) are an emerging technology recently proposed as an alternative to custom-designed application-specific integrated photonics. Light routing is one of the most important functions that need to be realized on a PPIC. Previous literature has investigated the light routing problem from an algorithmic or experimental perspective, e.g., adopting graph theory to route an optical signal. In this paper, we also focus on the light routing problem, but from a complementary and theoretical perspective, to answer questions about what is possible to be routed. Specifically, we demonstrate that not all path lengths (defined as the number of tunable basic units that an optical signal traverses) can be realized on a square-mesh PPIC, and a rigorous realizability condition is proposed and proved. We further consider multi-path routing, where we provide an analytical expression on path length sum, upper bounds on path length mean/variance, and the maximum number of realizable paths. All of our conclusions are proven mathematically. Illustrative potential optical applications using our observations are also presented at the end.

In this review, recent trends in microelectronics packaging reliability are summarized. We review the technology from early packaging concepts, including wire bond and BGA, to advanced techniques used in HI schemes such as 3D stacking, interposers, fan-out packaging, and more recently developed silicon interconnect fabric integration. This review includes approaches for both design modification studies and packaged device validation. Methods are explored for compatibility in new complex packaging assemblies. Suggestions are proposed for optimizations of the testing practices to account for the challenges anticipated in upcoming HI packaging schemes.

We use Bayesian optimization with a new tunable acquisition function to design a photonic Y-splitter robust to fabrication variations. Compared to conventional acquisition functions, our method yields more robust solutions across varying metrics and datasets.

Neural ordinary differential equations (Neural ODEs) is a family of continuous-depth neural networks where the evolution of hidden states is governed by learnable temporal derivatives. We identify a significant limitation in applying traditional Batch Normalization (BN) to Neural ODEs, due to a fundamental mismatch — BN was initially designed for discrete neural networks with no temporal dimension, whereas Neural ODEs operate continuously over time. To bridge this gap, we introduce temporal adaptive Batch Normalization (TA-BN), a novel technique that acts as the continuous-time analog to traditional BN. Our empirical findings reveal that TA-BN enables the stacking of more layers within Neural ODEs, enhancing their performance. Moreover, when confined to a model architecture consisting of a single Neural ODE followed by a linear layer, TA-BN achieves 91.1% test accuracy on CIFAR-10 with 2.2 million parameters, making it the first unmixed Neural ODE architecture to approach MobileNetV2-level parameter efficiency. Extensive numerical experiments on image classification and physical system modeling substantiate the superiority of TA-BN compared to baseline methods.

Assessing the reliability of electronic devices, circuits and packages requires accurate lifetime predictions and identification of failure modes. This paper demonstrates a new approach to the extraction of underlying failure mechanism distribution parameters from data corresponding to a combined distribution of two distinct mechanisms. Specifically, a differential evolution approach is developed for parameter identification in competing-risks and mixture models. Use of multiple metrics for performance evaluation shows that our approach outperforms the best-known methods in the literature. Numerical results are shown for simulated data and also for package-level and device-level real failure data. On the modeling of industrial package failure data, our approach provides up to 92% reduction in mean squared error, up to 7% increase in log-likelihood and up to 61% decrease in the maximum Kolmogorov-Smirnov distance. On ring oscillator data obtained from our laboratory experiments, the corresponding improvements are 94%, 5% and 77%, respectively. For both simulated and real datasets, the improvement in performance is validated through statistical tests of significance. An application of the approach is demonstrated for empirical extraction of the temperature-dependence of parameters from lifetime data at different test temperatures.

A rare event is defined by a low probability of occurrence. Accurate estimation of such small probabilities is of utmost importance across diverse domains. Conventional Monte Carlo methods are inefficient, demanding an exorbitant number of samples to achieve reliable estimates. Inspired by the exact sampling capabilities of normalizing flows, we revisit this challenge and propose normalizing flow assisted importance sampling, termed NOFIS. NOFIS first learns a sequence of proposal distributions associated with predefined nested subset events by minimizing KL divergence losses. Next, it estimates the rare event probability by utilizing importance sampling in conjunction with the last proposal. The efficacy of our NOFIS method is substantiated through comprehensive qualitative visualizations, affirming the optimality of the learned proposal distribution, as well as a series of quantitative experiments encompassing $10$ distinct test cases, which highlight NOFIS's superiority over baseline approaches.

Time series anomaly detection is challenging due to the complexity and variety of patterns that can occur. One major difficulty arises from modeling time-dependent relationships to find contextual anomalies while maintaining detection accuracy for point anomalies. In this paper, we propose a framework for unsupervised time series anomaly detection that utilizes point-based and sequence-based reconstruction models. The point-based model attempts to quantify point anomalies, and the sequence-based model attempts to quantify both point and contextual anomalies. Under the formulation that the observed time point is a two-stage deviated value from a nominal time point, we introduce a nominality score calculated from the ratio of a combined value of the reconstruction errors. We derive an induced anomaly score by further integrating the nominality score and anomaly score, then theoretically prove the superiority of the induced anomaly score over the original anomaly score under certain conditions. Extensive studies conducted on several public datasets show that the proposed framework outperforms most state-of-the-art baselines for time series anomaly detection.

Programmable photonic integrated circuits (PPICs) are an emerging technology recently proposed as an alternative to custom-designed application-specific integrated photonics. Light routing is one of the most important functions that need to be realized on a PPIC. Previous literature has investigated the light routing problem from an algorithmic or experimental perspective, e.g., adopting graph theory to route an optical signal. In this paper, we also focus on the light routing problem, but from a complementary and theoretical perspective, to answer questions about what is possible to be routed. Specifically, we demonstrate that not all path lengths (defined as the number of tunable basic units that an optical signal traverses) can be realized on a square-mesh PPIC, and a rigorous realizability condition is proposed and proved. We further consider multi-path routing, where we provide an analytical expression on path length sum, upper bounds on path length mean/variance, and the maximum number of realizable paths. All of our conclusions are proven mathematically. Illustrative potential optical applications using our observations are also presented.

Understanding process variations and their impact in silicon photonics remains challenging. To achieve high-yield manufacturing, a key step is to extract the magnitude and spatial distribution of process variations in the actual fabrication, which is usually based on measurements of replicated test structures. In this paper, we develop a Bayesian-based method to infer the distribution of systematic geometric variations in silicon photonics, without requiring replication of identical test structures. We apply this method to characterization data from multiple silicon nitride ring resonators with different design parameters. We extract distributions with standard deviation of 28 nm, 0.8 nm, and 3.8 nm for the width, thickness, and partial etch depth, respectively, as well as the spatial maps of these variations. Our results show that this characterization and extraction approach can serve as an efficient method to study process variation in silicon photonics, facilitating the design of high-yield silicon photonic circuits in the future.

We develop and apply a Bayesian linear regression technique for few-shot performance modeling of silicon photonic devices under process variation. The key contribution is to leverage information from low-fidelity simulations which are cheap to obtain, and to encode that information into the prior distribution under a Bayesian framework. We then evaluate the maximum-a-posteriori (MAP) estimator according to Bayes' theorem. Our numerical experiments on a rectangular waveguide and a Y-branch show that compared to linear regression, ridge regression, and a multi-layer feedforward neural network, this approach can achieve 7x smaller error given the same number of samples, or reduce running time by 2.4x to reach the same accuracy. Most importantly, we demonstrate that with only 10 expensive high-fidelity simulations and a small number of cheap low-fidelity simulations, our approach can obtain a model with good performance.

We consider the impact of intra-wafer systematic spatial variation, pattern density mismatch, and line edge roughness on splitter-tree-based integrated optical phased arrays. These variations can substantially affect the emitted beam profile in the array dimension. We study the effect on different architecture parameters, and the analysis is shown to be consistent with experimental results.

The utilization of Bayesian methods has been widely acknowledged as a viable solution for tackling various challenges in electronic integrated circuit (IC) design under stochastic process variation, including circuit performance modeling, yield/failure rate estimation, and circuit optimization. As the post-Moore era brings about new technologies (such as silicon photonics and quantum circuits), many of the associated issues there are similar to those encountered in electronic IC design and can be addressed using Bayesian methods. Motivated by this observation, we present a comprehensive review of Bayesian methods in electronic design automation (EDA). By doing so, we hope to equip researchers and designers with the ability to apply Bayesian methods in solving stochastic problems in electronic circuits and beyond.

A correlation between device reliability and process variations in packaged devices is identified. Weibull distribution statistics are used in a novel method for finding the limit of precision possible for time-to-failure averaging. This study identifies increase of variance due to process variation with size scaling in three generations of technologies. There is a linear decrease in precision threshold which correlates to the size of the device. The results present a concern of BTI in technologies from 16nm and below.