Tyler H. Chang

Argonne National Laboratory,
Mathematics & Computer Science (MCS) Division

About Me:

I am currently a Postdoc @ Argonne.

Before that, I was the Cunningham fellow @ Virginia Tech.

I develop software and algorithms at the intersection of optimization, approximation theory, and machine learning for scientific applications such as material design, computational physics, and HPC tuning. My active areas of work are blackbox optimization, machine learning, computational geometry, analysis of algorithms, and open-source scientific software. Read more about my research, and check out some of my software below or on GitHub.

Don't hesitate to reach out or connect at an upcoming event!

If you are a US-based student see several opportunities to visit Argonne.


^top

News:

Upcoming Events

May 2023: Giving talk Exploiting problem structure with ParMOO -- A Python library for parallel multiobjective simulation optimization at SIAM OP in Seattle

Aug 2023: Giving talk Data sampling for surrogate modeling and optimization at ICIAM in Tokyo

Recent Events

May 2023: Presented poster A framework for fully autonomous design of materials via multiobjective optimization and active learning: challenges and next steps at ICLR ML4Materials Workshop in Rwanda (virtual)

Apr 2023: Released ParMOO ver. 0.2.2 on GitHub

Apr 2023: The ParMOO Solver Farm is now publicly available at GitHub

Mar 2023: Our paper A framework for fully autonomous design of materials via multiobjective optimization and active learning: challenges and next steps has been accepted at ICLR 2023 Workshop on ML4Materials

Mar 2023: Gave talk ParMOO: A Python library for parallel multiobjective simulation optimization at SIAM CSE in Amsterdam

Feb 2023: Paper ParMOO: A Python library for parallel multiobjective simulation optimization accepted for publication in JOSS

All Talks


^top

Research and Publications:

I combine techniques from optimization, approximation theory, and machine learning in order to guide scientific experimentation and model complex real-world phenomena. My work facilitates reliable, interpretable, and structure-exploiting machine learning to drive scientific discovery and engineering system design.

Some examples of my past and ongoing work include:

  • multiobjective optimization and active learning problems involving time-consuming simulations and experiments,
  • predictive modeling with limited amounts of high-dimensional scientific and engineering data,
  • design and analysis of algorithms for numerical modeling in complex domains, and
  • development of software packages for optimization and geometric interpolation.

Representative Publications

All Papers


^top

Publicly Available Software:


2023. ParMOO: Python library for parallel multiobjective simulation optimization
Release: 0.2.2
Devs: T. H. Chang (lead), S. M. Wild, and H. Dickinson
Primary Prog. Lang: Python 3



2022. VTMOP: Solver for blackbox multiobjective optimization problems
Devs: T. H. Chang (sole)
Primary Prog. Lang: Fortran 2008



2020. DelaunaySparse: Interpolation via a sparse subset of the Delaunay triangulation
Devs: T. H. Chang (lead) and T. C. H. Lux
Primary Prog. Lang: Fortran 2003



2019. QAML: Quantum annealing math library
Devs: T. C. H. Lux (lead), T. H. Chang, and S. S. Tipirneni
Primary Prog. Lang: Python 3


^top

Information for Students:

Opportunities for Students

Former Summer Student Collaborators


Jun 2022 - Aug 2022. Manisha Garg (Urbana-Champaign), NSF MSGI at Argonne
Jun 2022 - Aug 2022. Hyrum Dickinson (Urbana-Champaign), DOE SULI at Argonne


^top

Teaching:

Courses Taught

  • Jan 2022 - Present. Adjunct Professor: College of DuPage, Dept. of Computer and Info. Science
    • Spring 2022. CIS 2531: Introduction to Python Programming (online)
    • Summer 2023. CIS 2531: Introduction to Python Programming (in-person)
  • Jan 2020 - May 2020. Instructor of Record: Virginia Tech, Dept. of Computer Science
    • Spring 2020. CS 3114: Data Structures and Algorithms (half in-person, half online)

Lecture Notes


^top




Last update: 2023-05-18

Copy this style