High-dimensional systems are difficult to model and predict. The underlying mechanisms of such systems are too complex to be fully understood with limited theoretical knowledge and/or physical measurements. Nevertheless, redcued-order models have been widely used to study high-dimensional systems, because they are practical and efficient to develop and implement. Although model errors (biases) are inevitable for reduced-order models, these models can still be proven useful to develop real-world applications. Evaluation and validation for idealized models are indispensable to serve the mission of developing useful applications. Data assimilation and uncertainty quantification can provide a way to assess the performance of a reduced-order model. Real data and a dynamical model are combined together in a data assimilation framework to generate corrected model forecasts of a system. Uncertainties in model forecasts and observations are also quantified in a data assimilation cycle to provide optimal updates that are representative of the real dynamics. In this research, data assimilation is applied to assess the performance of two reduced-order models. The first model is developed for predicting prostate cancer treatment response under intermittent androgen suppression therapy. A sequential data assimilation scheme, the ensemble Kalman filter (EnKF), is used to quantify uncertainties in model predictions using clinical data of individual patients provided by Vancouver Prostate Center. The second model is developed to study what causes the changes of the state of stratospheric polar vortex. Two data assimilation schemes: EnKF and ES-MDA (ensemble smoother with multiple data assimilation), are used to validate the qualitative properties of the model using ECMWF (European Center for Medium-Range Weather Forecasts) reanalysis data. In both studies, the reduced-order model is able to reproduce the data patterns and provide insights to understand the underlying mechanism. However, significant model errors are also diagnosed for both models from the results of data assimilation schemes, which suggests specific improvements of the reduced-order models.

This thesis project focuses on algorithms that generate good sampling points for function approximation. In one dimension, polynomial interpolation using equispaced points is unstable, with high Oscillations near the endpoints of the interpolated interval. On the other hand, Chebyshev nodes provide both stable and highly accurate points for polynomial interpolation. In higher dimensions, optimal sampling points are unknown. This project addresses this problem by finding algorithms that are robust in various domains for polynomial interpolation and least-squares. To measure the quality of the nodes produced by said algorithms, the Lebesgue constant will be used. In the algorithms, a number of numerical techniques will be used, such as the Gram-Schmidt process and the pivoted-QR process. In addition, concepts such as node density and greedy algorithms will be explored.

Patients suffering from Retinitis Pigmentosa (RP), the most common type of inherited retinal degeneration, experience irreversible vision loss due to photoreceptor degeneration. The preservation of cone photoreceptors has been deemed medically relevant as a therapy aimed at preventing blindness in patients with RP. Cones rely on aerobic glycolysis to supply the metabolites necessary for outer segment (OS) renewal and maintenance. The rod-derived cone viability factor (RdCVF), a protein secreted by the rod photoreceptors that preserves the cones, accelerates the flow of glucose into the cone cell stimulating aerobic glycolysis. This dissertation presents and analyzes ordinary differential equation (ODE) models of cellular and molecular level photoreceptor interactions in health and disease to examine mechanisms leading to blindness in patients with RP.

First, a mathematical model composed of four ODEs is formulated to investigate the progression of RP, accounting for the new understanding of RdCVF’s role in enhancing cone survival. A mathematical analysis is performed, and stability and bifurcation analyses are used to explore various pathways to blindness. Experimental data are used for parameter estimation and model validation. The numerical results are framed in terms of four stages in the progression of RP. Sensitivity analysis is used to determine mechanisms that have a significant affect on the cones at each stage of RP. Utilizing a non-dimensional form of the RP model, a numerical bifurcation analysis via MATCONT revealed the existence of stable limit cycles at two stages of RP.

Next, a novel eleven dimensional ODE model of molecular and cellular level interactions is described. The subsequent analysis is used to uncover mechanisms that affect cone photoreceptor functionality and vitality. Preliminary simulations show the existence of oscillatory behavior which is anticipated when all processes are functioning properly. Additional simulations are carried out to explore the impact of a reduction in the concentration of RdCVF coupled with disruption in the metabolism associated with cone OS shedding, and confirms cone-on-rod reliance. The simulation results are compared with experimental data. Finally, four cases are considered, and a sensitivity analysis is performed to reveal mechanisms that significantly impact the cones in each case.

The main objective of mathematical modeling is to connect mathematics with other scientific fields. Developing predictable models help to understand the behavior of biological systems. By testing models, one can relate mathematics and real-world experiments. To validate predictions numerically, one has to compare them with experimental data sets. Mathematical modeling can be split into two groups: microscopic and macroscopic models. Microscopic models described the motion of so-called agents (e.g. cells, ants) that interact with their surrounding neighbors. The interactions among these agents form at a large scale some special structures such as flocking and swarming. One of the key questions is to relate the particular interactions among agents with the overall emerging structures. Macroscopic models are precisely designed to describe the evolution of such large structures. They are usually given as partial differential equations describing the time evolution of a density distribution (instead of tracking each individual agent). For instance, reaction-diffusion equations are used to model glioma cells and are being used to predict tumor growth. This dissertation aims at developing such a framework to better understand the complex behavior of foraging ants and glioma cells.

This dissertation develops a second order accurate approximation to the magnetic resonance (MR) signal model used in the PARSE (Parameter Assessment by Retrieval from Single Encoding) method to recover information about the reciprocal of the spin-spin relaxation time function (R2*) and frequency offset function (w) in addition to the typical steady-state transverse magnetization (M) from single-shot magnetic resonance imaging (MRI) scans. Sparse regularization on an approximation to the edge map is used to solve the associated inverse problem. Several studies are carried out for both one- and two-dimensional test problems, including comparisons to the first order approximation method, as well as the first order approximation method with joint sparsity across multiple time windows enforced. The second order accurate model provides increased accuracy while reducing the amount of data required to reconstruct an image when compared to piecewise constant in time models. A key component of the proposed technique is the use of fast transforms for the forward evaluation. It is determined that the second order model is capable of providing accurate single-shot MRI reconstructions, but requires an adequate coverage of k-space to do so. Alternative data sampling schemes are investigated in an attempt to improve reconstruction with single-shot data, as current trajectories do not provide ideal k-space coverage for the proposed method.

The dynamics of a fluid flow inside 2D square and 3D cubic cavities

under various configurations were simulated and analyzed using a

spectral code I developed.

This code was validated against known studies in the 3D lid-driven

cavity. It was then used to explore the various dynamical behaviors

close to the onset of instability of the steady-state flow, and explain

in the process the mechanism underlying an intermittent bursting

previously observed. A fairly complete bifurcation picture emerged,

using a combination of computational tools such as selective

frequency damping, edge-state tracking and subspace restriction.

The code was then used to investigate the flow in a 2D square cavity

under stable temperature stratification, an idealized version of a lake

with warmer water at the surface compared to the bottom. The governing

equations are the Navier-Stokes equations under the Boussinesq approximation.

Simulations were done over a wide range of parameters of the problem quantifying

the driving velocity at the top (e.g. wind) and the strength of the stratification.

Particular attention was paid to the mechanisms associated with the onset of

instability of the base steady state, and the complex nontrivial dynamics

occurring beyond onset, where the presence of multiple states leads to a

rich spectrum of states, including homoclinic and heteroclinic chaos.

A third configuration investigates the flow dynamics of a fluid in a rapidly

rotating cube subjected to small amplitude modulations. The responses were

quantified by the global helicity and energy measures, and various peak

responses associated to resonances with intrinsic eigenmodes of the cavity

and/or internal retracing beams were clearly identified for the first time.

A novel approach to compute the eigenmodes is also described, making accessible

a whole catalog of these with various properties and dynamics. When the small

amplitude modulation does not align with the rotation axis (precession) we show

that a new set of eigenmodes are primarily excited as the angular velocity

increases, while triadic resonances may occur once the nonlinear regime kicks in.

A continuously and stably stratified fluid contained in a square cavity subjected to harmonic body forcing is studied numerically by solving the Navier-Stokes equations under the Boussinesq approximation. Complex dynamics are observed near the onset of instability of the basic state, which is a flow configuration that is always an exact analytical solution of the governing equations. The instability of the basic state to perturbations is first studied with linear stability analysis (Floquet analysis), revealing a multitude of intersecting synchronous and subharmonic resonance tongues in parameter space. A modal reduction method for determining the locus of basic state instability is also shown, greatly simplifying the computational overhead normally required by a Floquet study. Then, a study of the nonlinear governing equations determines the criticality of the basic state's instability, and ultimately characterizes the dynamics of the lowest order spatial mode by the three discovered codimension-two bifurcation points within the resonance tongue. The rich dynamics include a homoclinic doubling cascade that resembles the logistic map and a multitude of gluing bifurcations.

The numerical techniques and methodologies are first demonstrated on a homogeneous fluid contained within a three-dimensional lid-driven cavity. The edge state technique and linear stability analysis through Arnoldi iteration are used to resolve the complex dynamics of the canonical shear-driven benchmark problem. The techniques here lead to a dynamical description of an instability mechanism, and the work serves as a basis for the remainder of the dissertation.

Earth-system models describe the interacting components of the climate system and

technological systems that affect society, such as communication infrastructures. Data

assimilation addresses the challenge of state specification by incorporating system

observations into the model estimates. In this research, a particular data

assimilation technique called the Local Ensemble Transform Kalman Filter (LETKF) is

applied to the ionosphere, which is a domain of practical interest due to its effects

on infrastructures that depend on satellite communication and remote sensing. This

dissertation consists of three main studies that propose strategies to improve space-

weather specification during ionospheric extreme events, but are generally applicable

to Earth-system models:

Topic I applies the LETKF to estimate ion density with an idealized model of

the ionosphere, given noisy synthetic observations of varying sparsity. Results show

that the LETKF yields accurate estimates of the ion density field and unobserved

components of neutral winds even when the observation density is spatially sparse

(2% of grid points) and there is large levels (40%) of Gaussian observation noise.

Topic II proposes a targeted observing strategy for data assimilation, which uses

the influence matrix diagnostic to target errors in chosen state variables. This

strategy is applied in observing system experiments, in which synthetic electron density

observations are assimilated with the LETKF into the Thermosphere-Ionosphere-

Electrodynamics Global Circulation Model (TIEGCM) during a geomagnetic storm.

Results show that assimilating targeted electron density observations yields on

average about 60%–80% reduction in electron density error within a 600 km radius of

the observed location, compared to 15% reduction obtained with randomly placed

vertical profiles.

Topic III proposes a methodology to account for systematic model bias arising

ifrom errors in parametrized solar and magnetospheric inputs. This strategy is ap-

plied with the TIEGCM during a geomagnetic storm, and is used to estimate the

spatiotemporal variations of bias in electron density predictions during the

transitionary phases of the geomagnetic storm. Results show that this strategy reduces

error in 1-hour predictions of electron density by about 35% and 30% in polar regions

during the main and relaxation phases of the geomagnetic storm, respectively.

Inverse problems model real world phenomena from data, where the data are often noisy and models contain errors. This leads to instabilities, multiple solution vectors and thus ill-posedness. To solve ill-posed inverse problems, regularization is typically used as a penalty function to induce stability and allow for the incorporation of a priori information about the desired solution. In this thesis, high order regularization techniques are developed for image and function reconstruction from noisy or misleading data. Specifically the incorporation of the Polynomial Annihilation operator allows for the accurate exploitation of the sparse representation of each function in the edge domain.

This dissertation tackles three main problems through the development of novel reconstruction techniques: (i) reconstructing one and two dimensional functions from multiple measurement vectors using variance based joint sparsity when a subset of the measurements contain false and/or misleading information, (ii) approximating discontinuous solutions to hyperbolic partial differential equations by enhancing typical solvers with l1 regularization, and (iii) reducing model assumptions in synthetic aperture radar image formation, specifically for the purpose of speckle reduction and phase error correction. While the common thread tying these problems together is the use of high order regularization, the defining characteristics of each of these problems create unique challenges.

Fast and robust numerical algorithms are also developed so that these problems can be solved efficiently without requiring fine tuning of parameters. Indeed, the numerical experiments presented in this dissertation strongly suggest that the new methodology provides more accurate and robust solutions to a variety of ill-posed inverse problems.