May 24: Shashaani is awarded a grant for a three-year project on “Fast and Scalable Stochastic Derivative-free Optimization” from Office of Naval Research. Shashaani will lead this effort seeking to excel stochastic optimization in theory and practice for black-box noisy problems of large scales.
Group News
There are 54 posts filed in Group News (this is page 1 of 3).
Apr 24: Jain unconditionally passes his Final defense! He developed fast sampling techniques to handle big data when calibrating computer models and used his methods to help with wind power generation. Congrats Pranav!!
Apr 24: The group had a good night at the 2024 ISE CA Anderson Awards.
Ha received the ISE Distinguished Dissertation Award, which recognizes recent PhD graduates who have written original, innovative dissertations that reflect outstanding technical contributions and are likely to have significant scientific and societal impact! He will be the department’s nominee for the IISE Pritsker Dissertation Award and/or INFORMS Dantzig Dissertation Award. (Jeon-Ha’s friend, teammate and basketball buddy) received the award on Ha’s behalf as he was traveling. Congratulations Yunsoo!!
Shashaani received the ISE Faculty Scholar Award, which recognizes current ISE (tenure track) assistant and associate professors for excellence in research, teaching, and service over the past three calendar years and the potential for becoming a University Faculty Scholar.
Apr 24: Houser presents his research on robust and fast feature selection with screening and partitioning methods at the 2024 NC State Graduate Student Research Symposium. His poster will be added to the groups website!
Mar 24: Ha presented First Order Trust Region Methods with Adaptive Sampling at the 2nd INFORMS Optimization Society’s Conference in Houston, TX.
Mar 24: Paper “Iteration Complexity and Finite-Time Efficiency of Adaptive Sampling Trust-Region Methods for Stochastic Derivative-Free Optimization” accepted for publication in IISE Transactions.
Jan 24: New paper “Robust Variational Quantum Algorithms Incorporating Hamiltonian Uncertainty” submitted.
Jan 24: New paper “Simulation Model Calibration with Dynamic Stratification and Adaptive Sampling” submitted.
Dec 23: Join the Shashaani Research Group at 2023 Winter Simulation Conference in San Antonio. WSC starts with the SimOpt workshop, co-hosted by Shashaani.
Nov 23: Ha successfully defends his PhD dissertation and receives an unconditional pass from the committee. He will next be joining the National Laboratory of Renewable Energy as a postdoc working with a computational optimization group. Congratulations Dr. Ha!
Nov 23: Shashaani and colleagues from Biological & Agricultural Engineering and Marine, Earth and Atmospheric Sciences win a 1-year grant from NC Department of Justice to minimize impacts of climate variability on NC swine farms through the use of simulations and decision-making under uncertainty
Nov 2023: New paper “Statistical Inference on Simulation Output: Batching as an Inferential Device” with Y. Jeon and R. Pasupathy submitted.
Oct 23: Ha presents oracle complexity with adaptive sampling at the Midwest Optimization Meeting held at the University of Michigan:
Oct 2023: Shashaani’s group have 5 presentations at the 2023 INFORMS annual meeting:
Sep 2023: Shashaani selected as the Southern Cross University’s International Alumnus of the Year
Aug 23: Vahdat successfully defended her PhD dissertation and will be a data analyst in Liberty Mutual after graduate school. Congratulations Kimia!
Aug 23: Paper “On Common-Random-Numbers and the Complexity of Adaptive Sampling Trust-Region Methods” with Ha and Pasupathy now available on optimization-online.
Jul 23: Paper “Building Trees for Probabilistic Prediction via Scoring Rules” with Surer, Plumlee, and Guikema submitted.
Jun 23: Paper “Strata Design in Stochastic Simulations with Multivariate Inputs” with Park, Byon, and Ko submitted.
Jun 23: Paper “Wake Effect Parameter Calibration with Large-Scale Field Operational Data using Stochastic Optimization” published in Applied Energy! Here are some highlights:
- Engineering wake models have parameters that crucially affect their performance.
- Wake parameters can be calibrated as constants or functions of other wind variables.
- Stochastic optimization (SO) provides accurate and reliable wake calibrations.
- Both point and functional calibration of wake parameters can be done well with SO.
- A derivative-free trust-region SO method provides robust wake calibration.
- Efficiently implemented trust-region SO uses adaptive sampling and variance reduction.
- Stratifying data based on wind characteristics effectively reduces variance during SO.
- Choosing strata of wind data and dynamically sampling from each expedites calibration.
- Robust wake calibration helps understanding power deficit patterns in wind farms.
- Good prediction of power deficit due to wake enables optimal design of new wind farms.