I’m currently a 5th year Ph.D. student in physics at UT Austin receiving Provost’s graduate excellence fellowship. Recently, I joined the center for theoretical and computational neuroscience as a research assistant under the supervision of Dr. Thibaud Taillefumier. The two main focuses of my interests in the field of computational neuroscience, broadly speaking, are:
I use mathematical tools such as stochastic calculus and functional differential equation to study the behavior and extract properties of neural networks; and tropical algebra to study problems in bioinformatics such as phylogenetic trees. Meanwhile, for the second part, I’m also looking into the optimization over tensor networks that has a wide range of applications emerging in machine learning.
I enjoy using computational tools to solve problems either from research or in life. As a passionate believer and active advocate of Wolfram Mathematica, I use it almost daily for research and pleasure. I also use (C++ | Python | Rust) when they best suit my needs. I’m also into parallel computing, more specifically, CUDA for GPU computing. As a side hobby, I casually do web development; for example, this site and Jazzify.AI.
Characterizing metastable neural dynamics in finite-size spiking networks remains a daunting challenge. We propose to address this challenge in the recently introduced replica-mean-field (RMF) limit. In this limit, networks are made of infinitely many replicas of the finite network of interest, but with randomized interactions across replica. Such randomization renders certain excitatory networks fully tractable at the cost of neglecting activity correlations, but with explicit dependence on the finite size of the neural constituents. However, metastable dynamics typically unfold in networks with mixed inhibition and excitation. Here, we extend the RMF computational framework to point-process-based neural network models with exponential stochastic intensities, allowing for mixed excitation and inhibition. Within this setting, we show that metastable finite-size networks admit multistable RMF limits, which are fully characterized by stationary firing rates. Technically, these stationary rates are determined as solutions to a set of delayed differential equations under certain regularity conditions that any physical solutions shall satisfy. We solve this original problem by combining the resolvent formalism and singular-perturbation theory. Importantly, we find that these rates specify probabilistic pseudo-equilibria which accurately capture the neural variability observed in the original finite-size network. We also discuss the emergence of metastability as a stochastic bifurcation, which can also be interpreted as a static phase transition in the RMF limits. In turn, we expect to leverage the static picture of RMF limits to infer purely dynamical features of metastable finite-size networks, such as the transition rates between pseudo-equilibria.
The set of ultrametrics on $[n]$ nodes that are $\ell^\infty$-nearest to a given dissimilarity map forms a $(\max,+)$ tropical polytope. Previous work of Bernstein has given a superset of the set containing all the phylogenetic trees that are extreme rays of this polytope. In this paper, we show that Bernstein’s necessary condition of tropical extreme rays is sufficient only for $n=3$ but not for $n\geq 4$. Our proof relies on the exterior description of this tropical polytope, together with the tangent hypergraph techniques for extremality characterization. The sufficiency of the case $n=3$ is proved by explicitly finding all extreme rays through the exterior description. Meanwhile, an inductive construction of counterexamples is given to show the insufficiency for $n\geq 4$.
Magnetism in topological insulators (TIs) opens a topologically nontrivial exchange band gap, providing an exciting platform for manipulating the topological order through an external magnetic field. Here, we show that the surface of an antiferromagnetic thin film can magnetize the top and the bottom TI surface states through interfacial couplings. During the magnetization reversal, intermediate spin configurations are ascribed from unsynchronized magnetic switchings. This unsynchronized switching develops antisymmetric magnetoresistance spikes during magnetization reversals, which might originate from a series of topological transitions. With the high Néel ordering temperature provided by the antiferromagnetic layers, the signature of the induced topological transition persists up to $\sim 90 K$.
A balloon will cool down as air escapes from inside. This article establishes a model that explains its causes in detail from the perspective of continuum mechanics and thermodynamics. The cooling of the balloon is primarily attributed to the work done by the rubber sheet during its contraction process. Subsequently, a series of experiments and simulations are also carried out, which show good consistency with the theory.
We systematically study the localization effect in discrete-time quantum walks on a honeycomb network and establish the mathematical framework. We focus on the Grover walk first and rigorously derive the limit form of the walker’s state, showing it has a certain probability to be localized at the starting position. The relationship between localization and the initial coin state is concisely represented by a linear map. We also define and calculate the average probability of localization by generating random initial states. Further, coin operators varying with positions are considered and the sufficient condition for localization is discussed. We also similarly analyze another four-state Grover walk. Theoretical predictions are all in accord with numerical simulation results. Finally, our results are compared with previous works to demonstrate the unusual trapping effect of quantum walks on a honeycomb network, as well as the advantages of our method.