Find us at the Viking Game Room - In basement of Smith Memorial Student Union, from 4pm to 6pm.
Location: FMH 462 (Large Conference Room), from 4pm to 5pm.
Location: FMH 462 (Large Conference Room), from 4pm to 5pm.
Location: FMH 462 (Large Conference Room), from 4pm to 5pm.
Location: FMH 462 (Large Conference Room), from 4pm to 5pm.
Location: FMH 462 (Large Conference Room), from 4pm to 5pm.
Location: Oregon State University
Location: FMH 462, from 4pm - 5pm.
Find our booth at Party in the Park, from 11am to 2pm!
Talk about your research with grad students to share and/or get feedback (open to all departments who might need to talk through their math).
Clarify graduate level math concepts through conversation.
Practice giving presentations with feedback from researching peers.
Every other Thursday at 4PM in FMH 462 (Large Conference Room)
Title: Math at Intel
Title: Machine Learning for geothermal resource assessment
Abstract:
Improving the accuracy of geothermal resource assessment is crucial in the transition toward sustainable energy. To improve accuracy and reduce expert decision bias, the implementation of data-driven methods for the assessment of these resources has gained popularity during the past years. However the utilization of Machine Learning (ML) algorithms in geological datasets is challenging due to the intrinsic characteristics of geological data (i.e., categorical data for structural settings, a small number of examples, and sparsity of hydrothermal systems that translate into severe class imbalance) This work presents the challenges, workflow, and outcome of applying these tools to improve the assessment of hydrothermal systems.
Title: Consensus-based Communication-aware Formation Control for a Mobile Multi-agent System
Abstract:
Over the past few decades, unmanned aerial vehicle (UAV) technology has played a significant role in military and civilian applications. To meet the challenges of the future, in addition to improving the functionality and utility of individual aircraft, there is a need to consider how to develop more effective UAV management and organizations. Consequently, among the many developments in UAVs, formation control has become an important concept in recent years. Formation control requires multiple UAVs to adapt to the mission including generating a formation, staying in formation, and changing formation. In this paper, we further constrain the formation controller model, not only estimating the desired separation with acceptable accuracy but also ensuring a consensus among estimates. Thus, optimizes the overall communication performance of a dynamical multi-agent system.
Title: Introduction to convex algebraic geometry
Abstract:
Convex algebraic geometry is the study of convex optimization using powerful techniques from algebra and algebraic geometry. Central to this study are convex semialgebraic sets, such as spectrahedra. We will see how these objects can be used in polynomial optimization problems through semidefinite programming and sums-of-squares theory.
Title: High-Order Numerical Methods and High-Performance GPU Computing
Abstract:
The world's fastest supercomputer, ORNL's Frontier, was released in June 2022, reaching peak performance of 1.1 exaflops (that is, 1.1 quintillion arithmetic operations per second), ushering in the age of exascale computing. Frontier, as well as the US Department of Energy's other upcoming exascale supercomputers, derive the vast majority of their performance from graphics processing units (GPUs). GPU-based computing marks a radical departure from traditional CPU-based architectures, and taking advantage of the performance offered by these supercomputers requires research and development of new numerical methods, algorithms, and software.
This has prompted significant interest in higher-order numerical methods, which possess many attractive features that make them well-suited for large-scale GPU computing. However, the use of high-order methods also presents new challenges that must be addressed and opens new research directions for investigation. In this talk, I will give a brief overview of the algorithms underpinning high-performance, high-order numerical methods, and discuss some topics of ongoing research interest, including matrix-free linear solvers and physical constraint preservation. I will also mention some relevant topics for potential student projects.
In addition to the technical content, I will discuss (and answer any questions about) some of my experiences in graduate school, internships, postdocs, and working at a national lab.
Title: Improving the value of observations in data assimilation: significance, challenges, and recent advances.
Abstract:
Data assimilation systems (DAS) combine information from a numerical model, observational data, and error statistics to analyze and predict the state of a dynamical system. Four-dimensional variational data assimilation (4D-Var) provides an estimate (analysis) to the true state by solving a large-scale model-constrained optimization problem. Valuable insight on the relative importance and contribution of various DAS components to reduce the forecast errors may be obtained by performing sensitivity studies to assess the analysis and forecast impact induced by variations in the input parameters. This talk presents theoretical and practical aspects of the sensitivity analysis in a 4D-Var DAS including evaluation of the forecast sensitivity to observations (FSO), prior state estimate, and parameterized error covariance models. An FSO-based optimization approach is formulated to identify deficiencies in the weight assigned to various observing system components and adaptively improve the use of observations. The practical applicability of this methodology is illustrated in numerical experiments with the Naval Research Laboratory Atmospheric Variational Data Assimilation System-Accelerated Representer (NAVDAS-AR) and the U.S. Navy’s Global Environmental Model (NAVGEM).
Introduction to different computational topics. By Austen Nelson.
Vim
Dotfiles
Bash
Modules and Environments
Slurm
MPI
Other tips and tricks
Speaker: Kirill Voronin
Title: How sparse linear algebra and fast Fourier transform are used in the Intel (R) Math Kernel Library.
Come celebrate Pi-Day and relax from final exam prep on 3/14/19 with free pizza from Hotlips and a talk by Fariborz Maseeh Department of Mathematics and Statistics Professor Derek Garton! There will be lots of time to talk math, eat, socialize, and learn more about PSU's Student Chapter of SIAM. Bring a friend!
Speaker: Nathan Lawrence
Title: Low-dimensional characterization of human faces from gappy data
Speaker: John Caughman
Included Pie-cutting session!
Speaker: Derek Garton
Title: The Pollard RHO algorithm and randomness in discrete dynamical systems
Title: Acoustic headwave propagation in ocean ambient noise
Title: Derivative Information in Model-Constrained optimization.
Speaker: John Caughman
Included Pie-cutting session and Math trivia!