Striving for a better world.

HISTORY | INTERESTS | PROJECTS
Society faces complex multidisciplinary challenges, requiring rigorous contemporary approaches.
PROFESSIONAL HISTORY DOCUMENTS
Overview of skills and accomplishments.
I am now a NSF postdoctoral fellow!
In 2022, I was awarded the NSF Mathematical and Physical Sciences Ascending Postdoctoral Research Fellowship (NSF MPS-Ascend). My tenure began Fall 2022 and will last until Fall 2025.
This award is unique: alongside strong research goals, the award supports engagement of underrepresented groups (URGs) in Mathematical and Physical Sciences (MPS).
Below are some materials from my application. Along with research activities, my plan for engaging URGs in MPS fields is included in the project description.
Success does not come without challenge!
Below are some materials from my application to the National Research Council's Research Associateship Program (NRC RAP), which I submitted in February, 2021. I applied to a postdoctoral position in the Material Measurement Lab at the National Institute of Standards and Technology (NIST). Ultimately, I was not offered the position. Still, this helps illustrate my broad interests and adaptability.
PROFESSIONAL INTERESTS
Apply scientific expertise to problems critical for future society.
My interests are broad. I am particularly interested in working towards any goal focused on building a more intelligent, healthy, and sustainable society.
There are many types of open questions related to such goals, all quite different though not all independent or mutually exclusive. Within any area, many details must be considered. For example, carbon-free sources of energy are becoming increasingly popular around the world, with growing options for energy capture and storage.
But the existence of efficient green energy technologies does not address the issue of climate change by itself -- such technologies are only beneficial if they are used on large scales, which translates to questions in economics, politics, and social acceptance. Even in ideal circumstances, e.g. plasma fusion devices reaching high efficiency at relatively low cost and gaining wide public approval, it would take a considerable amount of time and money to incorporate such devices across the global infrastructure.
Contemporary challenges require multidisciplinary work and a holistic mindset. Technology is one important area, but it must be addressed in parallel with sociopolitical and economic question. Analogy: one cannot solve a Rubik's cube by solving each side in isolation -- instead, all sides must be solved together, self-consistently.
More generally, I am interested in applying my technical skills to advance science and technology in any way to the benefit of society at large.
Since I am trained as a physicist and have a passion for Markov networks and Bayesian inference, I believe this can go in many directions. I principally utilize Python for mathematical and statistical analysis, and I use Bash/zsh for automation and cluster computing. I also use GPU computing (e.g. using PyOpenCL) when the situation calls for it.

INDEPENDENT RESEARCH THEME
Modeling Social Interactions for an Intelligent Data Driven Future.
Since 2017
The spirit and tools of physics, such as formulating mathematical definitions to represent physical reality, are traditionally perceived as being confined to particle accelerators and advanced laboratories. However, there are many ideas and concepts from physics which have direct application in other contexts, including economics and politics.
Using techniques from the physics of many interacting individual parts, we can approach many questions related to sociology, economics, and politics. Such questions must be addressed to enable an intelligent, well planned, data driven society.
Examples include:
Can a minority opinion decide society's net behavior? [Yes, depending on details.]
How can emergency evacuation times be minimized by design? [Counter-intuitively, install barriers.]
Have international treaties and trade agreements made large-scale conflict less likely? [Maybe not.]
Primers for this relatively young field:
Claudio Castellano, et al., Statistical physics of social dynamics (review article)
Dirk Helbing, Quantitative Sociodynamics (book)
PUBLISHED WORK
Physics-Based Machine Learning Trains Hamiltonians and Decodes the Sequence-Conformation Relation in the Disordered Proteome
November 2024
Lilianna Houston, Michael Phillips, Andrew Torres, Kari Gaalswyk, Kingshuk Ghosh
Journal of Chemical Theory and Computation 20, 10266–10274
Theoretical treatments of IDPs generally capture charge patterning and electrostatic interactions very well, but are not able to encode non-electrostatic interactions from first principles. Here we use machine learning to get non-electrostatic interactions from sequence, and we combine it with electrostatic physics to obtain accurate predictions of the entire disordered proteome in humans (comprising 28,058 proteins). Because the machine learning prediction is embedded within a physical theory, we are able to make predictions that most machine learning systems cannot, e.g. full distance maps and response to conditions.
A cyclin-dependent kinase-mediated phosphorylation switch of disordered protein condensation
October 2023
Juan Manuel Valverde, Geronimo Dubra, Michael Phillips, Austin Haider, Carlos Elena-Real, Aurélie Fournet, Emile Alghoul, Dhanvantri Chahar, Nuria Andrés-Sanchez, Matteo Paloni, Pau Bernadó, Guido van Mierlo, Michiel Vermeulen, Henk van den Toorn, Albert J. R. Heck, Angelos Constantinou, Alessandro Barducci, Kingshuk Ghosh, Nathalie Sibille, Puck Knipscheer, Liliana Krasinska, Daniel Fisher, Maarten Altelaar
Nature Communications 14, 6316
Cells phosphorylate some proteins at particular points in their division cycle. In some cases, phosphorylation enhances propensity to phase separate, while in others phosphorylation diminishes it. Experiments show this clearly even in live cells. Theoretical modeling with mathematically intensive techniques reveals the physical origin of those behaviors.
Rules of Physical Mathematics Govern Intrinsically Disordered Proteins
May 2022
Kingshuk Ghosh, Jonathan Huihui, Michael Phillips, Austin Haider
Annual Review of Biophysics 51, 335-376
A review of advances and tools for analyzing Intrinsically Disordered Proteins based upon polymer physics. Notably, these methods are built upon first-principles physics and include sequence specificity, meaning the specific patterning and placements of amino acids rather than simply their counts. Models include ways to calculate thermodynamically averaged size, complete distance maps between any two amino acids, and propensity to undergo liquid-liquid phase separation.
Kondo Screening in Two-Dimensional p-Type Transition-Metal Dichalcogenides
February 2017
Michael Phillips and Vivek Aji
Physical Review B 95, 075103
The effect of Kondo screening of a magnetic orbital in monolayer transition-metal dichalcogenides is analyzed using techniques of variational states and the numerical renormalization group. Due to the strong spin-orbit coupling, lack of inversion symmetry, and topological nature of the material, an unusual Kondo bound state is formed at low temperatures. Its nature can also be tuned by the application of light to partially spin-polarize the bound state.
Beyond monopole electrostatics in regulating conformations of intrinsically disordered proteins
September 2024
Michael Phillips, Murugappan Muthukumar, Kingshuk Ghosh
In real physiological conditions within a cell, proteins are surrounded by ions. Most treatments of disordered proteins (IDPs) begin by modeling charged side-chains with full ionization, having charge of +1 or -1. In a bath of ions, side-chains will actually have some effective charge of less magnitude, due to the condensation of solution ions. Here, for the first time, we incorporate the process of ion condensation into the theory of IDP size/conformation, including emergent interactions from the formation of dipoles.
MaxCal can infer models from coupled stochastic trajectories of gene expression and cell division
July 2023
Andrew Torres, Spencer Cockerell, Michael Phillips, Gábor Balázsi, Kingshuk Ghosh
Biophysical Journal 122, 2623-2635
We develop a model of Coupled Stochastic Trajectories (CST) under the Maximum Caliber framework. We apply the concept to the problem of cell division, which is a stochastic process influenced by cellular properties such as protein abundance, which involves a second stochastic process from the production and degradation of proteins in the cell. We are able to disentangle the two stochastic processes and obtain either a full (joint) distribution, or focus on isolated (conditional) distributions. We also obtain effective transition rates for underlying processes.
Hysteresis Effects in Social Behavior with Parasitic Infection
June 2020
Michael Phillips
Journal of Statistical Physics 181, 293–304
Under what conditions, in principle, can a general parasitic infection control system-wide social behaviors? We model a simple two-population, two-behavior system under the Master equation in the large N limit. Main result: the net social behavior can be controlled by aberrant behavior stemming from infected individuals, if uniformity-seeking interactions are substantial.
Tunable Line Node Semimetals
September 2014
Michael Phillips and Vivek Aji
Physical Review B 90, 115111
The electronic and transport characteristics are analyzed for a layered heterostructure. The individual layers are alternating topological and normal insulators, giving rise to unique and tunable topological properties in the heterostructure. Band structure, an estimate of the anisotropic conductivity, and predictions of the de Haas-van Alphen effect are calculated.