Search for: All records

Practical reproducibility is the ability to reproduce results is a manner that is cost-effective enough to become a vehicle of mainstream scientific exploration. Since computational research artifacts usually require some form of computing to interpret, open and programmable infrastructure, such as a range of NSF-supported testbeds spanning infrastructure from datacen- ter through networks to wireless systems, is a necessary – but not sufficient – requirement for reproducibility. The question arises what other services and tools should build on the availability of such programmable infrastructure to foster the development and sharing of findable, accessible, integrated, and reusable (FAIR) experiments that underpin practical reproducibility. In this paper, we propose three such services addressing the problems of packaging for reuse, findability, and accessibility, respectively. We describe how we developed these services in Chameleon, an NSF-funded testbed for computer science research which has supported the research of a community of 8,000+ users, and discuss their strengths and limitations. 

The fixed preemption point (FPP) model has been studied as an alternative to fully preemptive and non-preemptive models, as restricting preemptions to specific, predictable locations within a task’s execution can simplify overhead analysis without disallowing preemptions entirely. Prior work has produced response-time analyses for global Earliest Deadline First (G-EDF) scheduling under the FPP model. However, scheduling decisions based solely on task deadlines may be too coarsegrained and may not lead to the lowest response times. In this paper, we propose global FPP EDF-like (G-FPP-EL) scheduling, which assigns a priority point in time for each non-preemptive region of a task. We adapt compliant-vector analysis (CVA) to our model and present general response-time bounds for G-FPPEL schedulers. We then demonstrate that it is possible to design G-FPP-EL schedulers acheiving response-time bounds optimal under CVA and argue that such schedulers should replace global FPP EDF. 

We study a model-free federated linear quadratic regulator (LQR) problem where M agents with unknown, distinct yet similar dynamics collaboratively learn an optimal policy to minimize an average quadratic cost while keeping their data private. To exploit the similarity of the agents' dynamics, we propose to use federated learning (FL) to allow the agents to periodically communicate with a central server to train policies by leveraging a larger dataset from all the agents. With this setup, we seek to understand the following questions: (i) Is the learned common policy stabilizing for all agents? (ii) How close is the learned common policy to each agent's own optimal policy? (iii) Can each agent learn its own optimal policy faster by leveraging data from all agents? To answer these questions, we propose a federated and model-free algorithm named FedLQR. Our analysis overcomes numerous technical challenges, such as heterogeneity in the agents' dynamics, multiple local updates, and stability concerns. We show that FedLQR produces a common policy that, at each iteration, is stabilizing for all agents. We provide bounds on the distance between the common policy and each agent's local optimal policy. Furthermore, we prove that when learning each agent's optimal policy, FedLQR achieves a sample complexity reduction proportional to the number of agents M in a low-heterogeneity regime, compared to the single-agent setting. 

A powerful concept behind much of the recent progress in machine learning is the extraction of common features across data from heterogeneous sources or tasks. Intuitively, using all of one's data to learn a common representation function benefits both computational effort and statistical generalization by leaving a smaller number of parameters to fine-tune on a given task. Toward theoretically grounding these merits, we propose a general setting of recovering linear operators M from noisy vector measurements y=Mx+w, where the covariates x may be both non-i.i.d. and non-isotropic. We demonstrate that existing isotropy-agnostic meta-learning approaches incur biases on the representation update, which causes the scaling of the noise terms to lose favorable dependence on the number of source tasks. This in turn can cause the sample complexity of representation learning to be bottlenecked by the single-task data size. We introduce an adaptation, 𝙳𝚎-𝚋𝚒𝚊𝚜 & 𝙵𝚎𝚊𝚝𝚞𝚛𝚎-𝚆𝚑𝚒𝚝𝚎𝚗 (𝙳𝙵𝚆), of the popular alternating minimization-descent (AMD) scheme proposed in Collins et al., (2021), and establish linear convergence to the optimal representation with noise level scaling down with the total source data size. This leads to generalization bounds on the same order as an oracle empirical risk minimizer. We verify the vital importance of 𝙳𝙵𝚆 on various numerical simulations. In particular, we show that vanilla alternating-minimization descent fails catastrophically even for iid, but mildly non-isotropic data. Our analysis unifies and generalizes prior work, and provides a flexible framework for a wider range of applications, such as in controls and dynamical systems. 

Abstract Infrasound may be used to detect the approach of hazardous volcanic mudflows, known as lahars, tens of minutes before their flow fronts arrive. We have analyzed signals from more than 20 secondary lahars caused by precipitation events at Fuego Volcano during Guatemala’s rainy season in May through October of 2022. We are able to quantify the capabilities of infrasound monitoring through comparison with seismic data, time lapse camera imagery, and high-resolution video of a well-recorded event on August 17. We determine that infrasound sensors, deployed adjacent to the lahar path and in small-aperture (10 s of meters) arrays, are particularly sensitive to remote detection of lahars, including small-sized events, at distances of at least 5 km. At Fuego Volcano these detections could be used to provide timely alerts of up to 30 min before lahars arrive at a downstream monitoring site, such as in the frequently impacted Ceniza drainage. We propose that continuous infrasound monitoring, from locations adjacent to a drainage, may complement seismic monitoring and serve as a valuable tool to help identify approaching hazards. On the other hand, infrasound arrays located a kilometer or more from the lahar path can be effectively used to track a lahar’s progression. 

We study the problem of learning a linear system model from the observations of M clients. The catch: Each client is observing data from a different dynamical system. This work addresses the question of how multiple clients collaboratively learn dynamical models in the presence of heterogeneity. We pose this problem as a federated learning problem and characterize the tension between achievable performance and system heterogeneity. Furthermore, our federated sample complexity result provides a constant factor improvement over the single agent setting. Finally, we describe a meta federated learning algorithm, FedSysID, that leverages existing federated algorithms at the client level. 

SN 2020zbf is a hydrogen-poor superluminous supernova (SLSN) atz = 0.1947 that shows conspicuous C IIfeatures at early times, in contrast to the majority of H-poor SLSNe. Its peak magnitude isM_g = −21.2 mag and its rise time (≲26.4 days from first light) places SN 2020zbf among the fastest rising type I SLSNe. We used spectra taken from ultraviolet (UV) to near-infrared wavelengths to identify spectral features. We paid particular attention to the C IIlines as they present distinctive characteristics when compared to other events. We also analyzed UV and optical photometric data and modeled the light curves considering three different powering mechanisms: radioactive decay of⁵⁶Ni, magnetar spin-down, and circumstellar medium (CSM) interaction. The spectra of SN 2020zbf match the model spectra of a C-rich low-mass magnetar-powered supernova model well. This is consistent with our light curve modeling, which supports a magnetar-powered event with an ejecta massM_ej = 1.5 M_⊙. However, we cannot discard the CSM-interaction model as it may also reproduce the observed features. The interaction with H-poor, carbon-oxygen CSM near peak light could explain the presence of C IIemission lines. A short plateau in the light curve around 35–45 days after peak, in combination with the presence of an emission line at 6580 Å, can also be interpreted as being due to a late interaction with an extended H-rich CSM. Both the magnetar and CSM-interaction models of SN 2020zbf indicate that the progenitor mass at the time of explosion is between 2 and 5M_⊙. Modeling the spectral energy distribution of the host galaxy reveals a host mass of 10^8.7M_⊙, a star formation rate of 0.24_−0.12^+0.41M_⊙yr⁻¹, and a metallicity of ∼0.4Z_⊙.

 

We address the problem of learning linear system models from observing multiple trajectories from different system dynamics. This framework encompasses a collaborative scenario where several systems seeking to estimate their dynamics are partitioned into clusters according to their system similarity. Thus, the systems within the same cluster can benefit from the observations made by the others. Considering this framework, we present an algorithm where each system alternately estimates its cluster identity and performs an estimation of its dynamics. This is then aggregated to update the model of each cluster. We show that under mild assumptions, our algorithm correctly estimates the cluster identities and achieves an approximate sample complexity that scales inversely with the number of systems in the cluster, thus facilitating a more efficient and personalized system identification process.