We present a new algorithm for solving large-scale security-constrained optimal power flow in polar form (AC-SCOPF). The method builds on Nonlinearly Constrained augmented Lagrangian (NCL), an augmented Lagrangian method in which the subproblems are solved using an interior-point method. NCL has two key advantages for large-scale SCOPF. First, NCL handles difficult problems such as infeasible ones or models with complementarity constraints. Second, the augmented Lagrangian term naturally regularizes the Newton linear systems within the interior-point method, enabling solution of the Newton systems with a pivoting-free factorization that can be efficiently parallelized on GPUs. We assess the performance of our implementation, called MadNCL, on large-scale corrective AC-SCOPFs, with complementarity constraints modeling the corrective actions. Numerical results show that MadNCL can solve AC-SCOPF with 500 buses and 256 contingencies fully on the GPU in less than 3 minutes, whereas Knitro takes more than 3 hours to find an equivalent solution.
This paper introduces a framework based on Large Deviation Theory (LDT) to accurately and efficiently compute the rare probabilities of voltage collapse. We formulate the problem as finding the most probable failure point (the instanton) on the stability boundary and derive both first-order and second-order approximations for the collapse probability. The second-order method incorporates the local curvature of the stability boundary, yielding higher accuracy. This LDT framework generalizes methods based on Mahalanobis distance and is extensible to non-Gaussian uncertainties. We validate our approach on test systems, demonstrating that the LDT estimates converge to Monte Carlo results in the rare-event regime where direct sampling becomes computationally prohibitive.
We report on the solution optimized for characterizing SNSPDs by rejecting electromagnetic interference (EMI) from various sources. The proposed readout method enhances measurement stability and enables reliable device characterization at low bias currents, where the signal-to-noise ratio is typically limited. By effectively suppressing EMI-induced noise, the method improves the ability to distinguish genuine detection events from spurious signals and reduces the effort required for data analysis. The approach has been applied to preliminary measurements of SNSPDs exposed to α particles emitted from a 241Am source, demonstrating stable operation and clean signal acquisition. While a detailed study of α detection is underway, the method establishes a foundation for further characterization of SNSPDs with various incident particles. The demonstrated EMI rejection technique is expected to facilitate future research in particle detection and support ongoing SNSPD development for applications in nuclear and accelerator-based experiments.
Forensics investigations continue at Fukushima Daiichi related to understanding and interpreting severe accident progression at the three affected units following the Great East Japan Earthquake on March 11, 2011. As part of this work, the U.S. Department of Energy sponsored a study in 2016 to examine ex-vessel accident progression in Unit 1 (1F1) based on what was known at the time using the MELTSPREAD and CORQUENCH codes developed by Argonne National Laboratory. Since then, TEPCO has adeptly utilized robot and drone technologies to gain a clearer understanding of conditions within the reactor containments, and for 1F1, these investigations have provided important insights related to MCCI. In parallel, progress has been made on improving melt spreading and MCCI/debris coolability models as part of the Reduction of Severe Accident Uncertainties (ROSAU) program using insights from the experiments as well as findings at Fukushima shared through numerous international collaborations and programs (FACE, ARC-F, …). The objective of this work is to apply these improved models to perform an integrated analysis of the 1F1 ex-vessel accident sequence to gauge their ability to replicate actual end-state debris conditions reported by TEPCO. The analysis spans the time from vessel failure up through cavity flooding and the follow-on debris cooling phase until cavity ablation is arrested. Since this paper is published in a special edition commemorating Jean-Marie Seiler, the MCCI/coolability analysis portion of this work makes note of his key contributions in this field that have been implemented in CORQUENCH as well as other severe accident codes.
Rapid and accurate building damage assessment after tornadoes is critical for emergency response and recovery, yet automated methods struggle with the visual complexity of tornado-induced wreckage, primarily due to severe domain shift and extreme class imbalance. We introduce ASTRA, a large-scale framework evaluating 79 open-source models (67 CNNs and 12 Transformers-based architectures) across 2,420 experiments on an expertly curated Quad-State Tornado Damage (QSTD) benchmark of 5,517 street-view images annotated under the IN-CORE engineering taxonomy. We find that achieving strong performance hinges on the interaction among architecture, optimization, and pre-training source, not on architecture alone. Attention-based models were acutely optimization-sensitive: although bottom-tier under the default configuration, they recovered by + 25 to + 38 Macro-F1-points once optimization was corrected, either by switching from Adam to SGD or, equivalently, lowering the learning rate. A controlled ablation attributes this swing primarily to the learning rate, with a smaller optimizer effect specific to Vision Transformers; CNNs showed no comparable reversal. Furthermore, we uncover a counterintuitive behavior in which scene-centric Places365 pre-training did not surpass object-centric ImageNet after optimization, despite its closer semantic proximity to building and streetscape scenes and its competitive zero-shot performance. In a strict zero-shot test on the held-out Tuscaloosa-Moore Tornado Damage dataset of 2,393 images, naive baselines failed to transfer, while QSTD-optimized models achieved substantially better performance. The champion model, ConvNeXt-Base, achieved a + 34.6 Macro-F1-point gain over its baseline, despite temporal, sensor, and geographic shifts. Consequently, ASTRA offers practical optimization guidelines and a rigorously validated, open-source foundation for automated tornado-damage assessment. Project Website: https://crumeike.github.io/astra-page.
