John Snyder


About me

A native of So Cal, I take the sun and beach for granted, enjoy complaining about the amazing weather, discussing the best routes thru LA to avoid traffic. After studying physics at Stanford University, I came straight to UC Irvine to do more physics. What do Stanford and UCI have in common? Ridiculous (awesome) mascots. After recently defending my dissertation, I am slated to leave UC Irvine in January as I received a Humboldt postdoctoral fellowship to continue my research in Germany with Klaus-Robert Müller at T.U. Berlin and Hardy Gross at the Max Planck Institute for microstructure physics in Halle. Besides physics, I am a semi-professional dancer (mainly modern and contemporary ballet) and a licensed instructor of Gyrotonic. I play piano and guitar, and dabble in music composition. Also, I enjoy cooking (eating) delicious food, long backpacking trips in nature, and travel. My favorite anime is death note.

Group Likes

-The Burkies (aka the Burg) don't take themselves too seriously.
-We have cookies at group meeting.
-Kieron tries to extort cookie money from visiting professors.

Research Summary

Density functional theory (DFT) allows us to learn about the structure and properties of atoms and molecules on a computer, without having to do an experiment in the lab. For example, when two molecules react together, bonds between certain atoms or created and broken, forming a product. These reactions can be very complicated, with many intermediate structures formed along the way. DFT allows us to explore the stability of certain structures, and probe which pathways are possible for the reaction to occur. It’s a very powerful method that can be applied to any electronic system, and offers a great balance between computational efficiency and accuracy. Its being used in many scientific fields, such as solid state physics, geophysics, drug design, photochemistry, biophysics, soil science, astrophysics (star formation, properties of planets), etc.

In solving the equations of DFT, there is a few quantities (i’m focusing on one in particular: the kinetic energy) that are unknown. My job as a theoretical physicist is to use all my knowledge of quantum mechanic principals, together with physical intuition gained from studying the properties of certain fundamental and simple electronic systems, to try and learn about the kinetic energy and approximate it. However, this is a pretty difficult and has progress over the last few decades has been slow.

Machine learning is a powerful collection of methods from artificial intelligence that can learn the patterns in data. Here how it works: you give the computer a few examples (i.e. particular molecules and their properties) that you know are correct. Then the algorithm infers the properties of new molecules from the few examples given. Using this technique, I am able to learn about the kinetic energy of electronic systems very accurately by giving the machine a few examples to start from.

Ultimately, this will speed up the calculations of DFT, and allow us to apply DFT to much larger systems than is currently possible (so we could study a million atoms instead of just a thousand).

Publications with Kieron
[165] Nonlinear gradient denoising: Finding accurate extrema from inaccurate functional derivatives John C. Snyder, Matthias Rupp, Klaus-Robert Müller, Kieron Burke, International Journal of Quantum Chemistry 115, 1102--1114 (2015). [bibtex] [pdf] [doi]
[161] Understanding kernel ridge regression: Common behaviors from simple functions to density functionals Kevin Vu, John C. Snyder, Li Li, Matthias Rupp, Brandon F. Chen, Tarek Khelif, Klaus-Robert Müller, Kieron Burke, International Journal of Quantum Chemistry 115, 1115--1128 (2015). [bibtex] [pdf] [doi]
[153] Understanding machine-learned density functionals Li Li, John C. Snyder, Isabelle M. Pelaschier, Jessica Huang, Uma-Naresh Niranjan, Paul Duncan, Matthias Rupp, Klaus-Robert M\FCller, Kieron Burke, International Journal of Quantum Chemistry 116, 819--833 (2016). [bibtex] [pdf] [doi]
[144] Orbital-free Bond Breaking via Machine Learning John C. Snyder, Matthias Rupp, Katja Hansen, Leo Blooston, Klaus-Robert Müller, Kieron Burke, J. Chem. Phys. 139, 224104 (2013). [bibtex] [pdf] [doi]
[142] Kernels, Pre-Images and Optimization John C. Snyder, Sebastian Mika, Kieron Burke, Klaus-Robert Müller, Chapter in Empirical Inference - Festschrift in Honor of Vladimir N. Vapnik (2013). [bibtex] [pdf]
[137] Finding Density Functionals with Machine Learning John C. Snyder, Matthias Rupp, Katja Hansen, Klaus-Robert Müller, Kieron Burke, Phys. Rev. Lett. 108, 253002 (2012). [supplementary information] [bibtex] [pdf] [doi]
[128] Communication: Ionization potentials in the limit of large atomic number Lucian A. Constantin, John C. Snyder, John P. Perdew, Kieron Burke, The Journal of Chemical Physics 133, 241103 (2010). [bibtex] [pdf] [doi]

“I want you. I love you. I need you." by McCree O’Kelley. UC Irvine Dance Escape 2013. Photo by Skye Schmidt.

I want you. I love you. I need you. Choreographer: McCree O’Kelley