To join the Applied Mathematics Colloquium mailing list visit https://groups.google.com/a/njit.edu/forum/?hl=en#!forum/math-colloquium/join (Google Profile required). To join the mailing list without a Google Profile, submit the seminar request form.

*******************************************************************************************************

September 4

Roy Goodman, NJIT

Host: Amir Sagiv

Massive vortices in quantum fluids: Hamiltonian Mechanics, Averaging, Normal Forms, and Slow Manifolds

The point vortex model is a classical ODE system that describes the interaction of idealized localized vortices in a planar, inviscid fluid. It has found significant application in quantum fluids such as liquid helium and Bose-Einstein condensates. In several physical systems, vortices are better modeled as having mass, in sharp contrast to classical point vortices, which are massless. We study a model of massive point vortices proposed by Richaud. Its inertial terms constitute a singular perturbation and have profound consequences for the dynamics. Fast oscillations superimposed on the vortices' slower motion are immediately apparent, but careful mathematical analysis may reveal more. Mathematically, the vortex motion satisfies a system of Hamiltonian differential equations, and the inertial terms introduce a singular perturbation. We apply methods from Hamiltonian mechanics, most notably Lie transform perturbation theory, to derive a normal-form approximation, construct a slow manifold, and reveal the behavior of solutions.

*******************************************************************************************************

September 11

Yuri Antipov, Louisiana State University

Host: Daniel Massatt

Carleman problem for a strip and crack propagation in a power-law graded material

The steady-state subsonic and transonic models of an interfacial crack moving along the interface between two power-law graded half-planes under conditions of plane strain are considered. The problems are governed by the Navier equations with variable coefficients. The solution method consists of two steps. First, the Navier equations are transformed into  a  Carleman problem wit two shifts of two meromorphic functions in a strip that is recast as a system of singular integral equations with a fixed singularity solved numerically. At the second stage, on satisfying the boundary conditions, a vector Riemann-Hilbert (RH) problem of two functions with a piece-wise constant matrix coefficient is derived. Its entries are expressed through the solution of the Carleman problem. An exact solution to the RH problem is determined. Asymptotics of the displacements jumps and the stresses at the crack tip are recovered. A Griffith-type criterion of crack propagation is proposed.  An application to the Heun differential equation is also discussed.

*******************************************************************************************************

September 18

Naoki Masuda, University of Michigan

Host: James MacLaurin

Anticipating bifurcations in network dynamics: temporal and spatial early warning signals

Complex dynamical systems often undergo sudden major changes, or tipping points, as conditions vary gradually. Examples discussed include mass extinctions of species, deforestation, and aggressive disease progression. Early warning signals (EWSs), often based on "critical slowing down", aim to anticipate such transitions. In fact, many systems of interest are heterogeneous networks arising from interactions among nodes, where constructing reliable EWSs is challenging. I will discuss temporal and spatial approaches to this problem. For temporal EWSs, we present heuristic and mathematically grounded methods (based on linear stochastic differential equation theory) for selecting small sets of sentinel nodes that can anticipate transitions as well as, or better than, monitoring the entire network. Because temporal EWSs demand many observations per node in general, spatial EWSs instead use a single snapshot and compute statistics across nodes, but network heterogeneity can obscure dynamical warning patterns. Therefore, we introduce a preprocessing step called baseline referencing, which compares each node’s state with its own far-from-transition baseline before computing spatial statistics. Numerically, this substantially improves variance-based signals. Finally, we develop theory explaining how equilibrium heterogeneity across nodes, network eigenvectors, and bifurcation type govern the success or failure of spatial EWSs in networks.

*******************************************************************************************************

September 25

Hector Baños, California State University - San Bernardino 

Host: Kristina Wicke

Modeling Ancestral Protein Evolution: Overparameterization, Model Misspecification, and a New Implementation

Modeling ancestral evolutionary relationships that occurred billions of years ago presents many challenges. Profile mixture models approximate the dynamics of protein sequence evolution over billion-year timescales, enabling accurate inference of ancient relationships across the tree of life. Recently, questions have been raised about whether these models are overparameterized. Drawing on classical statistical theory, extensive simulations, and empirical examples, we show that overparameterization is indeed a problem. We further identify issues arising from model misspecification and introduce a new implementation that alleviates these problems, yielding substantially better estimates. This talk is intended for a general audience.

*******************************************************************************************************

October 9

Fioralba Cakoni, Rutgers University 

Host: Thi Phong Nguyen

Spectral Problems in Inverse Scattering

Inverse scattering is a central theme in applied mathematics, with rich mathematical structure and a wide range of applications. The problem is inherently nonlinear and ill-posed, giving rise to fundamental challenges in uniqueness, stability, and algorithmic reconstruction. A paradigm shift in this area has emerged through the study of the relative scattering operator, which maps incoming to outgoing waves. This framework yields mathematically justified and computationally efficient reconstruction methods that exploit properties of the linear scattering operator to recover nonlinear information about the scattering object. In the frequency domain, this approach leads to interior eigenvalue problems posed on the support of the scatterer. These eigenvalues correspond to parameter values for which there exist nontrivial incident waves that generate no scattered field and are determined solely from scattering data. A prominent example for the inhomogeneous media is the transmission eigenvalue problem, a non-self-adjoint spectral problem with a perplexing mathematical structure. This presentation surveys the significance of such spectral problems in uniqueness and reconstruction and highlights many open theoretical and computational questions.

*******************************************************************************************************

October 16

Youssef Marzouk, Massachusetts Institute of Technology [MIT]

Host: Amir Sagiv

Designing dynamic measure transport for sampling and quantization

Sampling or otherwise summarizing complex probability distributions is a central task in applied mathematics, statistics, and machine learning. Many modern algorithms for this task introduce dynamics in the space of probability measures and design these dynamics to achieve good computational performance. We will discuss several aspects of this broad design endeavor.

First is the problem of optimal scheduling of dynamic transport, i.e., with what speed should one proceed along a prescribed path of probability measures? Though many popular methods seek trajectories with zero acceleration in a Lagrangian frame, we show how a specific class of “curved” trajectories can improve approximation and learning. We then present extensions of this idea which seek not only schedules but paths that improve spatial regularity of the underlying velocity.

Second, we discuss the problem of weighted quantization, i.e., summarizing a complex distribution with a small set of weighted samples. We study this problem from the perspective of minimizing maximum mean discrepancy via gradient flow in the Wasserstein–Fisher–Rao geometry. This perspective motivates a new fixed-point algorithm, called mean shift interacting particles (MSIP), which outperforms state-of-the-art methods for quantization. We show how MSIP can be used not only to quantize an empirical distribution, but to sample given an unnormalized density.

*******************************************************************************************************

October 23

David Nicholls, University of Illinois Chicago

Host: Xinyu Zhao

Scattering of Electromagnetic Waves by a Quasiperiodic Grating: A High-Order Numerical Method

In many applications of scientific and engineering interest, the accurate modeling of linear waves scattered by quasiperiodic media plays a crucial role. The ability to numerically simulate such configurations robustly and rapidly is of overwhelming importance in photonics applications. In this talk we will discuss the specific problem of electromagnetic radiation interacting with a two-dimensional multiply layered diffraction grating with quasiperiodic interfaces. We describe how the classical boundary perturbation method of Field Expansions can be extended to this two-dimensional problem, and with specific numerical experiments we will show the remarkable efficiency, fidelity, and high-order accuracy one can achieve with an implementation of this algorithm.

*******************************************************************************************************

October 30

John Zweck, New York Institute of Technology

Host: Roy Goodman

Stability of Short Pulse Lasers: Modeling, Numerics, and Analysis

Fiber lasers generate ultra-short pulses of light for applications such as the accurate measurement of time and frequency, the detection of exo-planets, and precision surgery. In the first part of the talk, we use a lumped model to study the time-periodic evolution of pulses that circulate through the optical devices that make up the laser ring. We develop numerical methods to solve the equations in the model, an optimization method to discover time-periodic pulses, and Floquet theory to determine their stability.

In the second part of the talk, we discuss a novel computational method to determine the spectral stability of stationary pulse solutions of the cubic-quintic complex Ginzburg-Landau equation We prove that the point spectrum of the linearization of the equation about a pulse is given by the zero set of the regular Fredholm determinant of a trace-class Birman-Schwinger operator. We adapt a method of Bornemann to numerically approximate the Fredholm determinant by a matrix determinant, and we quantify the error in this approximation. Numerical results show excellent agreement with existing methods. This approach avoids the challenge of solving the numerically stiff system of equations for the matrix-valued Jost solutions, and paves the way for the rigorous spectral analysis of breather solutions of nonlinear wave equations.

This is joint work with Vrushaly Shinglot, Erika Gallo, Yuri Latushkin, Chris Jones, Jeremy Marzuola, and Curtis Menyuk.

*******************************************************************************************************

November 6

Michael Goldberg, University of Cincinatti

Host: Amir Sagiv

Tools for Proving Sharp Dispersive Estimates

Solutions of the free Schrodinger equation in $R^n$ satisfy an $L^2$ conservation law while higher-power $L^p$ norms decrease over time. This combination is possible if the solution evolves toward smaller pointwise values over a larger volume of support. The $L^p$ bounds with time decay are known as dispersive estimates. They occur because wave packets with different wavelength travel with different group velocity and eventually separate from one another.

The same dispersive estimates are often true for the Schrodinger equation with a (possibly large) short-range potential, once bound states are projected away. It is a useful shorthand to think of "short-range" as decaying faster than $O(|x|^{-2})$. The relevant comparison is that the Laplacian has exactly the same scaling properties as multiplying a function pointwise by $\frac{1}{|x|^2}$.

In the talk we will explore functional analysis tools that make it possible to prove Schrodinger dispersive estimates with a potential at the scaling limit of being short-range.

*******************************************************************************************************

November 13

Chris Rasmussen, San Diego State University

Host: Michelle Cirillo

Insights and Lessons Learned from Department Initiatives to Improve Student Success
in Calculus: A Look Across Four National Research Projects

In any given year, approximately 700,000 students are enrolled in a college level calculus course. Calculus is thus a major footprint for our community. It is also, unfortunately, a major sore spot for our community. Passing rates and student interest continue to be problematically low, especially for students from underserved populations. On the positive side, a number of different national projects over the last decade have been making progress on this complicated challenge. In this talk, I will reflect on and highlight the findings and progress from four, multi-year national studies focused on student success in calculus. These findings and success stories offer a rich source of ideas and practices for other departments to adapt to improve their students’ experiences and success in calculus.

*******************************************************************************************************

November 20

Anirudh Sridhar, NJIT

Host: Michael Siegel

New perspectives on learning networks from dynamics

Suppose that a continuous-time, stochastic diffusion (i.e., the Susceptible-Infected process) spreads on an unknown graph. We only observe the time at which the diffusion reaches each vertex, i.e., the set of infection times. What can be learned about the unknown graph from the infection times? While there is far too little information to learn individual edges in the graph, we show that certain high-level properties -- such as the number of vertices of sufficiently high degree, or super-spreaders -- can surprisingly be determined with certainty. To achieve this goal, we develop a suite of algorithms that can efficiently detect vertices of degree asymptotically greater than sqrt(n) from infection times, for a natural and general class of graphs with n vertices. To complement these results, we show that our algorithms are information-theoretically optimal: there exist graphs for which it is impossible to tell whether vertices of degree larger than n^{1/2 - \epsilon} exist from vertices' infection times, for any \epsilon > 0. Finally, we discuss the broader implications of our ideas for change-point detection in non-stationary point processes. This talk is based on joint work with Anna Brandenberger (MIT) and Elchanan Mossel (MIT). 

*******************************************************************************************************

December 2 [Special Time - Wednesday 10AM]

Miguel Onorato, University of Turin

Host: Roy Goodman

Wave-wave resonant interactions in ocean wind waves

Ocean wind waves are a fundamental manifestation of complex dynamics in geophysical fluid systems, arising from the interplay between dispersion and nonlinearity. While linear wave theory provides a first-order description of wave motion, nonlinear interactions play a crucial role in the evolution of ocean wave fields by mediating the transfer of energy among waves of different scales. According to established theoretical frameworks, four-wave resonant interactions provide the primary mechanism for energy transfer among wave components in deep-water surface wave fields. Although the existence and efficiency of such interactions have been demonstrated in controlled laboratory experiments, their direct identification in the ocean, where a large number of random waves interact simultaneously, has remained elusive. 

In this talk, after a short review on the dynamics of ocean waves, I will present experimental evidence of four-wave resonant interactions in ocean wind waves. The result is based on the use of  a stereoscopic imaging system that measures the ocean surface elevation simultaneously in space and time. The observations reproduce the well-known figure-eight pattern predicted by Phillips and, more importantly, reveal a continuum of resonant configurations in close agreement with theoretical predictions. These findings support the validity of third-generation ocean wave models, strengthening their ability to accurately capture wave dynamics in the ocean.

*******************************************************************************************************

December 11

Pierre-Emmanuel Jabin, Penn State University

Host: James MacLaurin

Title Forthcoming

Abstract forthcoming

*******************************************************************************************************

 

Last Updated: November 19, 2026