Search

Particle accelerator operation requires simultaneous optimization of multiple objectives. Multi-Objective Optimization (MOO) is particularly challenging due to trade-offs between the objectives. Evolutionary algorithms, such as genetic algorithm (GA), have been leveraged for many optimization problems, however, they do not apply to complex control problems by design. This paper demonstrates the power of differentiability for solving MOO problems using a Deep Differentiable Reinforcement Learning (DDRL) algorithm in particle accelerators. We compare DDRL algorithm with Model Free Reinforcement Learning (MFRL), GA and Bayesian Optimization (BO) for simultaneous optimization of heat load and trip rates in the Continuous Electron Beam Accelerator Facility (CEBAF). The underlying problem enforces strict constraints on both individual states and actions as well as cumulative (global) constraint for energy requirements of the beam. A physics-based surrogate model based on real data is developed. This surrogate model is differentiable and allows back-propagation of gradients. The results are evaluated in the form of a Pareto-front for two objectives. We show that the DDRL outperforms MFRL, BO, and GA on high dimensional problems.

We study the implementation of fault-tolerant logical Clifford gates on stabilizer quantum error correcting codes based on their symmetries. Our approach is to map the stabilizer code to a binary linear code, compute its automorphism group, and impose constraints based on the Clifford operators permitted. We provide a rigorous formulation of the method for finding automorphisms of stabilizer codes and generalize ZX-dualities to non-CSS codes. We provide a Python package implementing our algorithms which uses the computational algebra system MAGMA. Our algorithms map automorphism group generators to physical circuits, calculate Pauli corrections based on the destabilizers of the code, and determine their logical action. We discuss the fault tolerance of the circuits and include examples of gates through automorphisms for the [[4,2,2]] and perfect [[5,1,3]] codes, bivariate bicycle codes, and the best known distance codes., Comment: 24 pages, 6 figures

The size of the key-value (KV) cache plays a critical role in determining both the maximum context length and the number of concurrent requests supported during inference in modern language models. The KV cache size grows proportionally with the number of attention heads and the tokens processed, leading to increased memory consumption and slower inference for long inputs. In this work, we explore the use of MixAttention, a model architecture modification closely related to a blog published by Character.AI. MixAttention combines sliding window attention, where only a small subset of recent tokens is stored in the KV cache, with KV cache sharing across layers. Our experiments demonstrate that MixAttention significantly reduces memory usage and improves inference speed without sacrificing model performance in both short and long-context tasks. We also explore various configurations of this architecture, identifying those that maintain quality across evaluation metrics while optimizing resource efficiency.

In this paper, we establish Lasota-Yorke inequality for the Frobenius-Perron Operator of a piecewise expanding $C^{1+\varepsilon}$ map of an interval. By adapting this inequality to satisfy the assumptions of the Ionescu-Tulcea and Marinescu ergodic theorem \cite{ionescu1950}, we demonstrate the existence of an absolutely continuous invariant measure (ACIM) for the map. Furthermore, we prove the quasi-compactness of the Frobenius-Perron operator induced by the map. Additionally, we explore significant properties of the system, including weak mixing and exponential decay of correlations., Comment: 15 pages, 1 figure

A cognitive fully adaptive radar (CoFAR) adapts its behavior on its own within a short period of time in response to changes in the target environment. For the CoFAR to function properly, it is critical to understand its operating environment through estimation of the clutter channel impulse response (CCIR). In general, CCIR is sparse but prior works either ignore it or estimate the CCIR by imposing sparsity as an explicit constraint in their optimization problem. In this paper, contrary to these studies, we develop covariance-free Bayesian learning (CoFBL) techniques for estimating sparse CCIR in a CoFAR system. In particular, we consider a multiple measurement vector scenario and estimate a simultaneously sparse (row sparse) CCIR matrix. Our CoFBL framework reduces the complexity of conventional sparse Bayesian learning through the use of the diagonal element estimation rule and conjugate gradient descent algorithm. We show that the framework is applicable to various forms of CCIR sparsity models: group, joint, and joint-cum-group. We evaluate our method through numerical experiments on a data set generated using RFView, a high-fidelity modeling and simulation tool. We derive Bayesian Cram\'{e}r-Rao bounds for the various considered scenarios to benchmark the performance of our algorithms. Our results demonstrate that the proposed CoFBL-based approaches perform better than the existing popular approaches such as multiple focal underdetermined system solver and simultaneous orthogonal matching pursuit.

We present a novel approach to data preparation for developing multilingual Indic large language model. Our meticulous data acquisition spans open-source and proprietary sources, including Common Crawl, Indic books, news articles, and Wikipedia, ensuring a diverse and rich linguistic representation. For each Indic language, we design a custom preprocessing pipeline to effectively eliminate redundant and low-quality text content. Additionally, we perform deduplication on Common Crawl data to address the redundancy present in 70% of the crawled web pages. This study focuses on developing high-quality data, optimizing tokenization for our multilingual dataset for Indic large language models with 3B and 7B parameters, engineered for superior performance in Indic languages. We introduce a novel multilingual tokenizer training strategy, demonstrating our custom-trained Indic tokenizer outperforms the state-of-the-art OpenAI Tiktoken tokenizer, achieving a superior token-to-word ratio for Indic languages.

In this paper, we prove the quasi-compactness of the Frobenius-Perron operator for a piecewise convex map $\tau$ with a countably infinite number of branches on the interval $I=[0,1]$. We establish that for high enough $n$ iterates of $\tau$, $\tau^n$ are piecewise expanding. Using the Lasota-Yorke Inequality derived from references \cite{hofbauer1982} and \cite{keller1985}, adapted to meet the assumptions of the Ionescu-Tulcea and Marinescu ergodic theorem, we demonstrate the existence of absolutely continuous invariant measure (ACIM) $\mu$ for $\tau$, the exactness of the dynamical system $(I, \tau,\mu)$ and the quasi-compactness of Frobenius-Perron operator $P_\tau$ induced by $\tau$. The last fact implies a multitude of strong ergodic properties of $\tau$., Comment: 23 pages, 6 figures

Large scale, inverse problem solving deep learning algorithms have become an essential part of modern research and industrial applications. The complexity of the underlying inverse problem often poses challenges to the algorithm and requires the proper utilization of high-performance computing systems. Most deep learning algorithms require, due to their design, custom parallelization techniques in order to be resource efficient while showing a reasonable convergence. In this paper we introduces a \underline{S}calable \underline{A}synchronous \underline{G}enerative workflow for solving \underline{I}nverse \underline{P}roblems \underline{S}olver (SAGIPS) on high-performance computing systems. We present a workflow that utilizes a parallelization approach where the gradients of the generator network are updated in an asynchronous ring-all-reduce fashion. Experiments with a scientific proxy application demonstrate that SAGIPS shows near linear weak scaling, together with a convergence quality that is comparable to traditional methods. The approach presented here allows leveraging GANs across multiple GPUs, promising advancements in solving complex inverse problems at scale.

Transition metal dichalcogenide (TMD) nanoscrolls (NS) exhibit significant photoluminescence (PL) signals despite their multilayer structure, which cannot be explained by the strained multilayer description of NS. Here, we investigate the interlayer interactions in NS to address this discrepancy. The reduction of interlayer interactions in NS is attributed to two factors: (1) the symmetry-broken mixed stacking order between neighbouring layers due to misalignment, and (2) the high inhomogeneity in the strain landscape resulting from the unique Archimedean spiral-like geometry with positive eccentricity. These were confirmed through transmission electron microscopy, field emission scanning electron microscopy and atomic force microscopy. To probe the effect of reduction of interlayer interactions in multilayered MoS$_2$ nanoscrolls, low-temperature PL spectroscopy was employed investigating the behaviour of K-point excitons. The effects of reduced interlayer interactions on exciton-phonon coupling (EXPC), exciton energy, and exciton oscillator strength are discussed, providing insights into the unique properties of TMD nanoscrolls., Comment: S.C. and T.C. have contributed equally to this work

In the original coded caching model introduced by Maddah-Ali and Niesen in 2014, the server starts broadcasting only after it receives demands from all the users. So, all the users must be active during the delivery phase. In this work, we consider a coded caching model called hotplug coded caching in which some of the users are offline during the delivery phase. This model was first introduced by Ma and Tuninetti (``On Coded Caching Systems with Offline Users," 2022 IEEE International Symposium on Information Theory). The concept of Hotplug Placement Delivery Arrays (HpPDAs) for the hotplug coded caching systems was introduced in (``Improved Hotplug Caching Schemes Using PDAs and $t$-Designs," \emph{arXiv:2311.02856}, 2024), in which the authors have constructed HpPDAs from $t$-designs. This work provides a new hotplug coded caching scheme from the existing HpPDAs. The performance comparison of the proposed scheme with the existing schemes is presented. When applied for HpPDAs from $t$-designs, our scheme outperforms the baseline scheme by Ma and Tuninetti, and the Improved $t$-scheme by Rajput and Rajan in some memory regimes., Comment: 8 pages, 4 figures and 1 table. arXiv admin note: text overlap with arXiv:2311.02856

In Pliable Private Information Retrieval (PPIR) with a single server, messages are partitioned into $\Gamma$ non-overlapping classes. The user wants to retrieve a message from its desired class without revealing the identity of the desired class to the server. In S. A. Obead, H. Y. Lin and E. Rosnes, Single-Server Pliable Private Information Retrieval With Side Information, arXiv:2305.06857, authors consider the problem of PPIR with Side Information (PPIR-SI), where the user now has side information. The user wants to retrieve any new message (not included in the side information) from its desired class without revealing the identity of the desired class and its side information. A scheme for the PPIR-SI is given by Obead et al. for the case when the users side information is unidentified, and this case is referred to as PPIR with Unidentifiable SI (PPIR-USI). In this paper, we study the problem of PPIR for the single server case when the side information is partially identifiable, and we term this case as PPIR with Identifiable Side Information (PPIR-ISI). The user is well aware of the identity of the side information belonging to $\eta$ number of classes, where $1\leq \eta \leq \Gamma$. In this problem, The user wants to retrieve a message from its desired class without revealing the identity of the desired class to the server. We give a scheme for PPIR-ISI, and we prove that having identifiable side information is advantageous by comparing the rate of the proposed scheme to the rate of the PPIR-USI scheme given by Obead et al. for some cases. Further, we extend the problem of PPIR-ISI for multi-user case, where users can collaborately generate the query sets, and we give a scheme for this problem., Comment: 11 pages and 3 figures Additional references included

This paper investigates whether recent advances in Large Language Models (LLMs) can assist in translating human explanations into a format that can robustly support learning Linear Temporal Logic (LTL) from demonstrations. Both LLMs and optimization-based methods can extract LTL specifications from demonstrations; however, they have distinct limitations. LLMs can quickly generate solutions and incorporate human explanations, but their lack of consistency and reliability hampers their applicability in safety-critical domains. On the other hand, optimization-based methods do provide formal guarantees but cannot process natural language explanations and face scalability challenges. We present a principled approach to combining LLMs and optimization-based methods to faithfully translate human explanations and demonstrations into LTL specifications. We have implemented a tool called Janaka based on our approach. Our experiments demonstrate the effectiveness of combining explanations with demonstrations in learning LTL specifications through several case studies., Comment: 21 Pages, 13 Page Appendix

The linear stability of a suspension of phototactic algae is investigated numerically with particular emphasis on the effects of the angle of incidence of the illuminating oblique collimated irradiation with thermal effects. The suspension is illuminated by the oblique collimated irradiation from the top and heated/cooled from the bottom. The linear stability analysis shows that the suspension becomes more unstable as the angle of incidence increases., Comment: arXiv admin note: text overlap with arXiv:2301.09864

In this study, our focus is mainly to check the effect of light scattering on the onset of thermal-phototactic-bioconvection in an algal suspension where the suspension is illuminated by the collimated oblique irradiation from above while simultaneously applying heating or cooling from below. We conduct a numerical investigation into the linear stability of a suspension containing phototactic algae, focusing particularly on how the angle of incidence of oblique collimated irradiation influences the system. Our solutions reveal a transition of the most unstable mode from a stationary to an overstable state, or vice versa, under certain parameter configurations as the angle of incidence varies. Additionally, we frequently observe oscillatory instabilities in cases where the upper surface is rigid, particularly as the angle of incidence increases within the suspension.

The GlueX Central Drift Chamber (CDC) in Hall D at Jefferson Lab, used for detecting and tracking charged particles, is calibrated and controlled during data taking using a Gaussian process. The system dynamically adjusts the high voltage applied to the anode wires inside the chamber in response to changing environmental and experimental conditions such that the gain is stabilized. Control policies have been established to manage the CDC's behavior. These policies are activated when the model's uncertainty exceeds a configurable threshold or during human-initiated tests during normal production running. We demonstrate the system reduces the time detector experts dedicate to calibration of the data offline, leading to a marked decrease in computing resource usage without compromising detector performance., Comment: AI4EIC2023, 9 pages

Hydra is a system which utilizes computer vision to perform near real time data quality management, initially developed for Hall-D in 2019. Since then, it has been deployed across all experimental halls at Jefferson Lab, with the CLAS12 collaboration in Hall-B being the first outside of GlueX to fully utilize Hydra. The system comprises back end processes that manage the models, their inferences, and the data flow. The front-end components, accessible via web pages, allow detector experts and shift crews to view and interact with the system. This talk will give an overview of the Hydra system as well as highlight significant developments in Hydra's feature set, acute challenges with operating Hydra in all halls, and lessons learned along the way.