arXiv subjects

Edward Chen

Publications and source records attributed to Edward Chen.

At least 19 recordsLinked to original sources

arXiv2026

On the electronic and vibrational dimensionality of nanometer-scale silicon structures

We discuss the problem of assessing the electronic and vibrational dimensionality of a semiconductor nanostructure: How thin and/or wide must a nanostructure be in order to induce electron and phonon confinement? Clarifying the physical justification for common criteria found in the literature, we view the electron coherence length (defined as the electron and phonon inelastic mean free path) as their `field of view' and argue (or, better yet, `speculate') that this sets the important length scale. Considering the example of Si nanosheets at room temperature, and drawing from results found in the literature, we estimate that the critical length below which electrons are subject to quantum confinement is of the order of (or smaller than) 8 nm, when their coherence length is determined by energy losses to phonons and remote phonons in gated structures. On the contrary, no single length-scale can be given for phonons: Taking their coherence length as determined by scattering with electrons and anharmonic three-phonon processes, short wavelength acoustic and optical phonons may be confined only by structures as small as 10 nm. Long-wavelength acoustic phonons, instead, may exhibit a coherence length of the order of 1 micrometer, so that they may be confined over much larger distances.

cond-mat.mes-hall↗

arXiv2026

Electron transport in a 1.6~nm-thick double-gated (100) silicon nanosheet: A theoretical study accounting for phonon confinement and remote-phonon scattering

We study theoretically electron transport in an top-and bottom-gated (100) 1.6 nm-thin silicon nanosheet with SiO2/HfO2 gate stacks, focusing on the intrinsic physical processes that affect transport: the confinement of phonons and the presence of interface hybrid plasmon-phonon excitations (IPPs or `remote phonons'). The band structure is calculated using local empirical pseudopotentials; an approximated elastic continuum model is used to consider the confinement of acoustic phonons; the dielectric continuum limit is used to deal with the IPPs. We find that the electron mobility is affected significantly by the boundary conditions chosen to deal with phonon confinement. The more realistic assumption of phonons clamped at the SiO2/HfO2 interfaces and optical phonons at the Si/SiO2 interfaces results in a room temperature mobility much smaller than what is obtained using the common assumption of bulk phonons in the elastic, high-temperature approximation. We also find that, as a result of the complicated structure of the primed subbands, the high-field saturated velocity is significantly lower than its bulk value, as it had been measured in the past in the case of Si inversion layers but never explained theoretically. Finally, we find that IPP scattering does depress the low-field mobility but to a small extent, thanks to the presence of the interfacial SiO2 layers and to the proximity of the metal gates. Moreover, by keeping electrons `cooler', IPP scattering results in a higher saturated velocity. Therefore, the presence of high-kappa materials in the gate-insulator stacks should not affect negatively the performance of field effect transistors based on Si nanosheets.

cond-mat.mes-hall↗

arXiv2026

On the static dielectric constant of thin dielectrics in extremely scaled silicon nanosheet transistors

We argue that the static dielectric constant of small (thin and/or narrow) semiconductor and insulator nanostructures depends strongly on the their environment. We do so by considering the electronic response simply reviewing, briefly but critically, the existing literature. Regarding the ionic response, in addition to reviewing the literature, we use a simple model to account for the confinement of optical phonons in thin films and show that the reduction of their density of states has a negligible effect on the dielectric constant, in contrast to some claims found in the literature. In general, we argue that in realistic structures, such as double-gated Si nanosheets, the use of the bulk dielectric constants for both the channel and the gate insulators, is justified.

cond-mat.mes-hall↗

arXiv2026

FlowRL: A Taxonomy and Modular Framework for Reinforcement Learning with Diffusion Policies

Thanks to their remarkable flexibility, diffusion models and flow models have emerged as promising candidates for policy representation. However, efficient reinforcement learning (RL) upon these policies remains a challenge due to the lack of explicit log-probabilities for vanilla policy gradient estimators. While numerous attempts have been proposed to address this, the field lacks a unified perspective to reconcile these seemingly disparate methods, thus hampering ongoing development. In this paper, we bridge this gap by introducing a comprehensive taxonomy for RL algorithms with diffusion/flow policies. To support reproducibility and agile prototyping, we introduce a modular, JAX-based open-source codebase that leverages JIT-compilation for high-throughput training. Finally, we provide systematic and standardized benchmarks across Gym-Locomotion, DeepMind Control Suite, and IsaacLab, offering a rigorous side-by-side comparison of diffusion-based methods and guidance for practitioners to choose proper algorithms based on the application. Our work establishes a clear foundation for understanding and algorithm design, a high-efficiency toolkit for future research in the field, and an algorithmic guideline for practitioners in generative models and robotics. Our code is available at https://github.com/typoverflow/flow-rl.

cs.LG↗

arXiv2026

Discovering Implicit Large Language Model Alignment Objectives

Large language model (LLM) alignment relies on complex reward signals that often obscure the specific behaviors being incentivized, creating critical risks of misalignment and reward hacking. Existing interpretation methods typically rely on pre-defined rubrics, risking the omission of "unknown unknowns", or fail to identify objectives that comprehensively cover and are causal to the model behavior. To address these limitations, we introduce Obj-Disco, a framework that automatically decomposes an alignment reward signal into a sparse, weighted combination of human-interpretable natural language objectives. Our approach utilizes an iterative greedy algorithm to analyze behavioral changes across training checkpoints, identifying and validating candidate objectives that best explain the residual reward signal. Extensive evaluations across diverse tasks, model sizes, and alignment algorithms demonstrate the framework's robustness. Experiments with popular open-source reward models show that the framework consistently captures > 90% of reward behavior, a finding further corroborated by human evaluation. Additionally, a case study on alignment with an open-source reward model reveals that Obj-Disco can successfully identify latent misaligned incentives that emerge alongside intended behaviors. Our work provides a crucial tool for uncovering the implicit objectives in LLM alignment, paving the way for more transparent and safer AI development.

cs.LG↗

arXiv2026

ALMo: Interactive Aim-Limit-Defined, Multi-Objective System for Personalized High-Dose-Rate Brachytherapy Treatment Planning and Visualization for Cervical Cancer

In complex clinical decision-making, clinicians must often track a variety of competing metrics defined by aim (ideal) and limit (strict) thresholds. Sifting through these high-dimensional tradeoffs to infer the optimal patient-specific strategy is cognitively demanding and historically prone to variability. In this paper, we address this challenge within the context of High-Dose-Rate (HDR) brachytherapy for cervical cancer, where planning requires strictly managing radiation hot spots while balancing tumor coverage against organ sparing. We present ALMo (Aim-Limit-defined Multi-Objective system), an interactive decision support system designed to infer and operationalize clinician intent. ALMo employs a novel optimization framework that minimizes manual input through automated parameter setup and enables flexible control over toxicity risks. Crucially, the system allows clinicians to navigate the Pareto surface of dosimetric tradeoffs by directly manipulating intuitive aim and limit values. In a retrospective evaluation of 25 clinical cases, ALMo generated treatment plans that consistently met or exceeded manual planning quality, with 65% of cases demonstrating dosimetric improvements. Furthermore, the system significantly enhanced efficiency, reducing average planning time to approximately 17 minutes, compared to the conventional 30-60 minutes. While validated in brachytherapy, ALMo demonstrates a generalized framework for streamlining interaction in multi-criteria clinical decision-making.

cs.LG↗

arXiv2025

Dominant scattering mechanisms in the low/high electric field transport in cryogenic 2D confinement in Silicon (110) with high-$\kappa$ oxides

The performance of silicon nano-devices at cryogenic temperatures is critical for quantum qubit control circuits and space applications. Using multi-valley Monte Carlo simulations, we investigate electron transport in Si~(110) systems. At low electric fields, phonon absorption becomes negligible, and mobility is governed by competition between remote Coulomb scattering~(RCS) at low inversion charge density and surface roughness scattering~(SRS) at high density, leading to a mobility peak. High-$\kappa$ dielectrics such as $\mathrm{HfO_2}$ introduce remote phonon scattering~(RPS), which suppresses mobility. Under high electric fields, phonon emission dominates at 4~K, limiting velocity enhancement and resulting in limited current improvement

cond-mat.mes-hall↗

arXiv2025

Interactive Multi-Objective Probabilistic Preference Learning with Soft and Hard Bounds

High-stakes decision-making involves navigating multiple competing objectives with expensive evaluations. For instance, in brachytherapy, clinicians must balance maximizing tumor coverage (e.g., an aspirational target or soft bound of >95% coverage) against strict organ dose limits (e.g., a non-negotiable hard bound of <601cGy to the bladder). Selecting Pareto-optimal solutions that match implicit preferences is challenging, as exhaustive Pareto frontier exploration is computationally and cognitively prohibitive, necessitating interactive frameworks to guide users. While decision-makers (DMs) often possess domain knowledge to narrow the search via such soft-hard bounds, current methods often lack systematic approaches to iteratively refine these multi-faceted preference structures. Furthermore, DMs often require confidence that they have not overlooked superior alternatives, a paramount necessity in high-stakes scenarios. We present Active-MoSH, an interactive local-global framework designed for this process. Its local component integrates probabilistic preference learning with an active sampling strategy to adaptively refine Pareto subsets while minimizing cognitive burden. To bolster decision confidence, Active-MoSH's global component, C-MoSH, leverages multi-objective sensitivity analysis to identify potentially overlooked, high-value points beyond immediate feedback. We demonstrate Active-MoSH's performance benefits through diverse synthetic and real-world applications. A high-stakes case study with real cervical cancer brachytherapy treatment plans and an image selection user study further validate our hypotheses regarding the framework's ability to improve convergence, enhance DM confidence, and provide expressive preference articulation.

cs.AI↗

arXiv2024

Modeling Multi-Objective Tradeoffs with Monotonic Utility Functions

Countless science and engineering applications in multi-objective optimization (MOO) necessitate that decision-makers (DMs) select a Pareto-optimal (PO) solution which aligns with their preferences. Evaluating individual solutions is often expensive, and the high-dimensional trade-off space makes exhaustive exploration of the full Pareto frontier (PF) infeasible. We introduce a novel, principled two-step process for obtaining a compact set of PO points that aligns with user preferences, which are specified a priori as general monotonic utility functions (MFs). Our process (1) densely samples the user's region of interest on the PF, then (2) sparsifies the results into a small, diverse set for the DM. We instantiate this framework with soft-hard functions (SHFs), an intuitive class of MFs that operationalizes the common expert heuristic of imposing soft and hard bounds. We provide extensive empirical validation of our framework instantiated with SHFs on diverse domains, including brachytherapy, engineering design, and large language models. For brachytherapy, our approach returns a compact set of points with over 3% greater SHF-defined utility than the next best approach. Among the other domains, our approach consistently leads in utility, as a final compact set of just 5 points captures over 99% of the utility offered by the entire dense set.

cs.LG↗

arXiv2023

Dynamic Model Agnostic Reliability Evaluation of Machine-Learning Methods Integrated in Instrumentation & Control Systems

In recent years, the field of data-driven neural network-based machine learning (ML) algorithms has grown significantly and spurred research in its applicability to instrumentation and control systems. While they are promising in operational contexts, the trustworthiness of such algorithms is not adequately assessed. Failures of ML-integrated systems are poorly understood; the lack of comprehensive risk modeling can degrade the trustworthiness of these systems. In recent reports by the National Institute for Standards and Technology, trustworthiness in ML is a critical barrier to adoption and will play a vital role in intelligent systems' safe and accountable operation. Thus, in this work, we demonstrate a real-time model-agnostic method to evaluate the relative reliability of ML predictions by incorporating out-of-distribution detection on the training dataset. It is well documented that ML algorithms excel at interpolation (or near-interpolation) tasks but significantly degrade at extrapolation. This occurs when new samples are "far" from training samples. The method, referred to as the Laplacian distributed decay for reliability (LADDR), determines the difference between the operational and training datasets, which is used to calculate a prediction's relative reliability. LADDR is demonstrated on a feedforward neural network-based model used to predict safety significant factors during different loss-of-flow transients. LADDR is intended as a "data supervisor" and determines the appropriateness of well-trained ML models in the context of operational conditions. Ultimately, LADDR illustrates how training data can be used as evidence to support the trustworthiness of ML predictions when utilized for conventional interpolation tasks.

cs.LG↗

arXiv2023

Studies of two-dimensional material resistive random-access memory by kinetic Monte Carlo simulations

Resistive memory based on 2D WS2, MoS2, and h-BN materials has been studied, including experiments and simulations. The influences with different active layer thicknesses have been discussed, including experiments and simulations. The thickness with the best On/Off ratio is also found for the 2D RRAM. This work reveals fundamental differences between a 2D RRAM and a conventional oxide RRAM. Furthermore, from the physical parameters extracted with the KMC model, the 2D materials have a lower diffusion activation energy from the vertical direction, where a smaller bias voltage and a shorter switching time can be achieved. It was also found the diffusion activation energy from the CVD-grown sample is much lower than the mechanical exfoliated sample. The result shows MoS2 has the fastest switching speed among three 2D materials.

cond-mat.mtrl-sci↗

arXiv2023

Image-Force Barrier Lowering of Schottky Barriers in Two-Dimensional Materials as a Function of Metal Contact Angle

Two-dimensional (2D) semiconductors are a promising solution for the miniaturization of electronic devices and for the exploration of novel physics. However, practical applications and demonstrations of physical phenomena are hindered by high Schottky barriers at the contacts to 2D semiconductors. While the process of image-force barrier lowering (IFBL) can considerably decrease the Schottky barrier, IFBL is not fully understood for the majority of prevalent contact geometries. We introduce a novel technique to determine the IFBL potential energy with application spanning far beyond that of any existing method. We do so by solving Poisson's equation with the boundary conditions of two metal surfaces separated by an angle Omega. We then prove that our result can also be obtained with the method of images provided a non-Euclidean, cone-manifold space is used. The resulting IFBL is used to calculate the expected contact resistance of the most prevalent geometric contacts. Finally, we investigate contact resistance and show how the stronger IFBL counteracts the effect of larger depletion width with increasing contact angle. We find that top contacts experience lower contact resistance than edge contacts. Remarkably, our results identify tunable parameters for reducing Schottky barriers and likewise contact resistance to edge-contacted 2D materials, enhancing potential applications.

cond-mat.mes-hall↗

arXiv2022

Systems-theoretic Hazard Analysis of Digital Human-System Interface Relevant to Reactor Trip

Human-system interface is one of the key advanced design features applied to modern digital instrumentation and control systems of nuclear power plants. The conventional design is based on a compact workstation-based system within the control room. The compact workstation provides both a strategic operating environment while also a convenient display for plant status information necessary to the operator. The control environment is further enhanced through display panels, visual and auditory alarms, and procedure systems. However, just like the legacy control, the HSI should incorporate diversity to demonstrate sufficient defense-in-depth protection against common cause failures of the safety system. Furthermore, the vulnerability of the HSI is affected by a plethora of factors, such as human error, cyberattacks, software common cause failures, etc., that complicate the design and analysis. Therefore, this work aims to identify and evaluate existing system vulnerabilities to support the licensing, deployment and operation of HSI designs, especially the functions that are relevant to a reactor trip. We performed a systematic hazard analysis to investigate potential vulnerabilities within the HSI design using the novel redundancy-guided systems-theoretic hazard analysis. This method was developed and demonstrated by Idaho National Laboratory under a project initiated by the Risk-Informed Systems Analysis Pathway of the U.S. Department of Energy's Light Water Reactor Sustainability Program. The goal of the project is to develop a strong technical basis for risk assessment strategies to support effective, reliable, and licensable digital instrumentation and control technologies.

cs.HC↗

arXiv2022

An Application of a Modified Beta Factor Method for the Analysis of Software Common Cause Failures

This paper presents an approach for modeling software common cause failures (CCFs) within digital instrumentation and control (I&C) systems. CCFs consist of a concurrent failure between two or more components due to a shared failure cause and coupling mechanism. This work emphasizes the importance of identifying software-centric attributes related to the coupling mechanisms necessary for simultaneous failures of redundant software components. The groups of components that share coupling mechanisms are called common cause component groups (CCCGs). Most CCF models rely on operational data as the basis for establishing CCCG parameters and predicting CCFs. This work is motivated by two primary concerns: (1) a lack of operational and CCF data for estimating software CCF model parameters; and (2) the need to model single components as part of multiple CCCGs simultaneously. A hybrid approach was developed to account for these concerns by leveraging existing techniques: a modified beta factor model allows single components to be placed within multiple CCCGs, while a second technique provides software-specific model parameters for each CCCG. This hybrid approach provides a means to overcome the limitations of conventional methods while offering support for design decisions under the limited data scenario.

cs.SE↗

arXiv2022

Failure Mechanism Traceability and Application in Human System Interface of Nuclear Power Plants using RESHA

In recent years, there has been considerable effort to modernize existing and new nuclear power plants with digital instrumentation and control systems. However, there has also been considerable concern both by industry and regulatory bodies for the risk and consequence analysis of these systems. Of concern are digital common cause failures specifically due to software defects. These failures by the software can occur in both the control and monitoring of a system. While many methods have been proposed to identify software failure modes, such as Systems Theoretic Process Analysis, Hazard and Consequence Analysis for Digital Systems, etc., these methods are focused primarily on the control action pathway of a system. In contrast, the information feedback pathway lacks Unsafe Control Actions, which are typically related to software basic events; thus, assessment of software basic events in such systems is unclear. In this work, we present the idea of intermediate processors and Unsafe Information Flow (UIF) to help safety analysts trace failure mechanisms in the feedback pathway and how they can be integrated into a fault tree for improved assessment capability. The concepts presented are demonstrated in two comprehensive case studies, a smart sensor integrated platform for unmanned autonomous vehicles and another on a representative advanced human system interface for safety critical plant monitoring. The qualitative software basic events are identified, and a fault tree analysis is conducted based on a modified Redundancy guided Systems theoretic Hazard Analysis methodology. The case studies demonstrate the use of UIFs and intermediate processors in the fault tree to improve traceability of software failures in highly complex digital instrumentation feedback. The improved method clarifies fault tree construction when multiple component dependencies are present in the system.

cs.SE↗

arXiv2022

Application of Orthogonal Defect Classification for Software Reliability Analysis

The modernization of existing and new nuclear power plants with digital instrumentation and control systems (DI&C) is a recent and highly trending topic. However, there lacks strong consensus on best-estimate reliability methodologies by both the United States (U.S.) Nuclear Regulatory Commission (NRC) and the industry. In this work, we develop an approach called Orthogonal-defect Classification for Assessing Software Reliability (ORCAS) to quantify probabilities of various software failure modes in a DI&C system. The method utilizes accepted industry methodologies for quality assurance that are verified by experimental evidence. In essence, the approach combines a semantic failure classification model with a reliability growth model to predict the probability of failure modes of a software system. A case study was conducted on a representative I&C platform (ChibiOS) running a smart sensor acquisition software developed by Virginia Commonwealth University (VCU). The testing and evidence collection guidance in ORCAS was applied, and defects were uncovered in the software. Qualitative evidence, such as modified condition decision coverage, was used to gauge the completeness and trustworthiness of the assessment while quantitative evidence was used to determine the software failure probabilities. The reliability of the software was then estimated and compared to existing operational data of the sensor device. It is demonstrated that by using ORCAS, a semantic reasoning framework can be developed to justify if the software is reliable (or unreliable) while still leveraging the strength of the existing methods.

cs.SE↗

arXiv2022

Advanced Transient Diagnostic with Ensemble Digital Twin Modeling

The use of machine learning (ML) model as digital-twins for reduced-order-modeling (ROM) in lieu of system codes has grown traction over the past few years. However, due to the complex and non-linear nature of nuclear reactor transients as well as the large range of tasks required, it is infeasible for a single ML model to generalize across all tasks. In this paper, we incorporate issue specific digital-twin ML models with ensembles to enhance the prediction outcome. The ensemble also utilizes an indirect probabilistic tracking method of surrogate state variables to produce accurate predictions of unobservable safety goals. The unique method named Ensemble Diagnostic Digital-twin Modeling (EDDM) can select not only the most appropriate predictions from the incorporated diagnostic digital-twin models but can also reduce generalization error associated with training as opposed to single models.

cs.LG↗

arXiv2022

Quantitative Evaluation of Common Cause Failures in High Safety-significant Safety-related Digital Instrumentation and Control Systems in Nuclear Power Plants

Digital instrumentation and control (DIC) systems at nuclear power plants (NPPs) have many advantages over analog systems. They are proven to be more reliable, cheaper, and easier to maintain given obsolescence of analog components. However, they also pose new engineering and technical challenges, such as possibility of common cause failures (CCFs) unique to digital systems. This paper proposes a Platform for Risk Assessment of DIC (PRADIC) that is developed by Idaho National Laboratory (INL). A methodology for evaluation of software CCFs in high safety-significant safety-related DIC systems of NPPs was developed as part of the framework. The framework integrates three stages of a typical risk assessment, qualitative hazard analysis and quantitative reliability and consequence analyses. The quantified risks compared with respective acceptance criteria provide valuable insights for system architecture alternatives allowing design optimization in terms of risk reduction and cost savings. A comprehensive case study performed to demonstrate the framework capabilities is documented in this paper. Results show that the PRADIC is a powerful tool capable to identify potential digital-based CCFs, estimate their probabilities, and evaluate their impacts on system and plant safety.

eess.SY↗