DSpace@RPI (Rensselaer Polytechnic Institute)
Not a member yet
6809 research outputs found
Sort by
Application of wave turbulence theory to surface gravity-internal waves & fermi-pasta-tsingou chains
No full textDecember 2021School of ScienceWe present new results in the field of Wave Turbulence Theory, applying our results in theanalysis of two specific systems: interacting surface and internal waves, and β-Fermi-Patsa-Ulam-Tsingou chains. In chapters 4–6 we consider interactions between surface and internal waves in a modelproblem: a two layer system. We analyze this simplified model with the goal of building a
tractable framework to understand how energy flows between the surface and internal wave
modes from first principles. We consider an approach using standard Wave Turbulence techniques, based on theHamiltonian structure of the equations of motion. We include the general procedure for
diagonalization of the quadratic part of the Hamiltonian with two wave types, a non-trivial
question, with our transformation being applicable to a other Hamiltonians which may share
a similar structure of nondiagonal terms. We derive the interaction coefficients between the surface and internal waves, obtainingthe coupled kinetic equations which describe the evolution of the total spectral energy of the
system. Notably, our derivation allows for both resonant and near-resonant interactions, the
latter which are important in some parameter regimes when the system is no longer weakly
nonlinear. We consider the case when the surface waves follow an ocean JONSWAP spectrum.We find that energy transfer from surface to internal waves occurs along a timescale of
hours; for our choice of parameters, we find that the energy transfers are dominated by the
specific class III resonances. We also note that internal waves oblique to the direction of the
wind generated, with a specific lobed spectrum of internal waves developing for some initial
conditions. In chapter 7–9 we consider general Hamiltonian nonlinear dispersive wave systems withcubic nonlinearity, investigating the common Wave Turbulence assumption of the Random
Phase Approximation. We show that such systems can develop anomalous correlators between phases during their evolution, despite the prescribed randomness in the initial data. To explain this phenomena theoretically, we follow a prior generalization of the Wick’sdecomposition and Wave Turbulence theory, showing that stationary solution with nonzero
anomalous correlations exists. We also describe the relationship between the anomalous correlator and the appearance
of “ghost” excitations in the spatiotemporal spectrum of the system, i.e. excitations given
by negative frequencies, in addition to the usual positive frequencies given by the linear
dispersion relation and the standard Wave Turbulence theory. We test our theory on extensive simulations of the celebrated β-Fermi-Pasta-UlamTsingou chain; numerically measured values of the anomalous correlators agree with ourtheory in the weakly nonlinear domain. We hypothesize that such excitations exist in other
systems dominated by nonlinear interactions, such as surface-gravity waves. In chapter 10 we outline a derivation for the collision integral of a kinetic equation of asystem of nonhomogenous wave turbulence. We provide a specific physical example of how
this kinetic equation may prove useful in investigating the nonlinear Schrodinger equation
with a varying potential.Ph
Comprehensive assessment of cellular drug delivery and efficacy via noninvasive functional and time-resolved molecular optical imaging
No full textDecember 2022School of EngineeringDuring drug development, preclinical in vivo validation of a viable drug candidate is a necessary unavoidable step before successful Food and Drug Administration’s (FDA) approval process for clinical trial. Conventional preclinical in vivo studies are performed using immunocompromised mouse models – including cell-line-based and patient-derived xenograft (PDX) models. Further, the validation of a drug’s delivery efficacy is performed ex vivo – that is, via animal sacrifice, tissue extraction and subsequent biomolecular assays. This method is time/financially consuming, does not permit longitudinal study, requires many animal subjects, and does not accurately reflect in vivo physiology. For any viable preclinical imaging platform, especially ones associated with the assessment of drug delivery and efficacy, it is crucial to be able to image the whole body of small animal models. Hence, any proposed imaging approach should be able to image a large field of view with high sensitivity. In addition, biological investigations are greatly enhanced by the ability to image multiple biomarkers simultaneously. This can be done by multiplexing the fluorescence probes spectrally and via lifetime sensing. However, increasing the dimensionality of the acquired data leads to significantly increased complexity in the imaging apparatus, data quantification and experimental protocols. Förster Resonance Energy Transfer (FRET) imaging can sense protein-protein interaction events at the distance of 2-10nm, which is the range in which binding of antibodies/protein ligands to their respective receptors often occur. Indeed, we have demonstrated the capability of Macroscopic Fluorescence Lifetime Imaging (MFLI)-FRET for noninvasive, whole-body quantitative monitoring of receptor-ligand binding (e.g. Transferrin-Transferrin Receptor, [Tf-TfR]) both in tumor xenografts and for monitoring pharmacokinetic activity in mice. The capability to monitor drug efficacy noninvasively over an entire live intact animal model would offer many significant improvements to the current paradigm: true longitudinal studies, decrease of the number of animal models needed (minimizing intra-animal variances), intact physiological context and, if imaging can be performed fast enough, pharmacokinetic monitoring post-injection. Most importantly, a technique capable of whole-body imaging as described above should result in higher success rates upon reaching human trials, given the significantly richer information that can be collected and used to decide whether to proceed with a candidate. The objective of the project presented herein is to further develop a cost-effective, noninvasive, and user-friendly whole-body optical imaging workflow capable of meeting these needs. Hence, the results of this project will act as a giant leap forward with regards to propelling optical molecular imaging into conventional preclinical research and development.Ph
Deep learning and optimization based interferometric and phaseless imaging
No full textAugust2023School of EngineeringInterferometric imaging involves reconstructing an image from the cross-correlations of the scattered waves reflected from the illuminated medium. Phaseless imaging on the other hand involves reconstructing an image from the auto-correlated scattered signal measurements. The objective of this thesis is to address these two inverse problems in imaging in a unified mathematical framework and develop theory, methods and algorithms. Specifically, our objective is to design provably convergent and exact deep learning (DL) based techniques for the interferometric and phaseless image reconstruction problems. Towards this end, we first introduced a DL-based phaseless imaging approach, which we refer to as DL-Wirtinger flow or DL-WF, and theoretically established sufficient conditions on the network parameters to guarantee exact recovery under deterministic imaging geometries. We designed a corresponding deep imaging network composed of three DL-based network elements: an encoder that produces an optimal initial encoded representation from the spectral estimation, a recurrent neural network (RNN) that generates the encoded image estimation, and a decoder that produces the final image from the RNN output. The RNN is designed using the unrolling technique from the non-convex optimization based Wirtinger flow (WF) algorithm, while the learned decoding operator helps integrate prior information during the reconstruction process by restricting the recovered signals to its range space. Recovery in the lower dimensional encoded representation space leads to improved sample complexity, and all three trained network components jointly contribute towards the faster convergence rate and improved reconstruction quality compared to the optimization-based methods. Furthermore, we extended the existing mathematical tools used to study the WF algorithm under general deterministic forward maps to account for our encoded representation and decoding prior based inversion approach. We additionally designed a computationally efficient decoding prior based phaseless imaging approach that does not require prior information of the forward map unlike DLWF, and instead acquires this knowledge indirectly during a supervised training phase. The improvement in computational efficiency is accomplished, firstly, by a simple DL-based initialization step that replaces the computationally expensive spectral initialization used in DL-WF, and secondly, by addressing a modified optimization problem using the kernel PCA concept and a DL-based sample processing function that simplifies the gradient calculation at each iterative update stage. This new optimization problem necessitates an alternative mathematical tool compared to DL-WF, that accounts for the network architecture and the training set sizes while establishing theoretical performance guarantee of this approach. Later, we generalized the nature of learned prior information integrated during the recovery process by adopting imaging schemes that apply general DL-based prior information in the Bayesian sense for interferometric and phaseless imaging problems. Additionally, unlike the previous two DNs that are designed to implement a least square minimization type algorithm whose convergence to the ground truth relies heavily on the accuracy of the initial point, we instead devised a denoising scheme for the steps of the power method with the goal of optimally searching for the leading eigenvector of the back-projected measurements. The leading eigenvector of this matrix resembles the true unknown for sampling vectors offering high redundancies in the quadratic measurements. Moreover, we introduced a novel strategy for designing a denoiser formulation under certain assumptions on the unknown signals that leads to theoretical recovery guarantees. We further presented a DL-based implementation of this denoiser that offers improved computational cost, necessary for its application to high dimensional signal recovery problems, as well as enables learned regularization for attaining faster convergence. Finally, we extended our quadratic inversion technique using deep decoding prior tothe challenging wave-based imaging configuration for which the Born approximation related linearization of the sampling vectors are not valid. We introduced an RNN type network which is empirically observed to offer improved reconstruction quality and faster convergence rate compared to the state-of-the-art.Ph
In silico design of integrated chromatographic purification processes for therapeutic proteins
No full textMay 2019School of EngineeringDesigning non-affinity downstream processes for biologics poses a significant challenge due to the broad range of design space available for resin selection and buffer conditions. Imposing the design constraint of integrated manufacturing can help to prune this space and improve efficiency at the cost of markedly increasing the complexity during process design. To address this challenge, we developed an in silico-based approach to quickly design and rank a fully inclusive list of integrated downstream processes for their ability to remove impurities using only orthogonally selective multimodal and ion exchange resins. This approach involves the one-time characterization of an impurity database and considers both impurity profiles patterns and product retention behavior to generate and score all possible integrated purification trains. This database was then employed in concert with an in silico process development tool which generates and ranks all possible integrated chromatographic sequences for their ability to remove orthogonal impurities. Top-ranking outputs are then used to guide the experimental development and refinement of purification processes, significantly expediting the development of downstream processes. This approach represents a platformable strategy for rapidly designing purification processes for non-platform molecules. In this work, impurities found in null cell culture fluids were characterized on sets of multimodal, HCIC, and ion exchange resins using fast, high-resolution UPLC assays. This strategy was employed to generate one-time process-related impurity databases for both Pichia pastoris and CHO cell cultures; two industrially relevant cell lines with very different HCP burdens and properties. To demonstrate the effectiveness and versatility of this approach, the in silico tool was tasked with solving purification challenges in both of these systems. For Pichia pastoris, two non-mAb products, hGH and G-CSF, were successfully purified from cell culture fluid resulting in high purity and product recovery. In order to extend the utility of this in silico tool, we performed modifications to our process design approach in order to account for both process-related impurities and product-related impurities. This modified strategy was first shown to be successful for purifying IFN produced in Pichia pastoris, and was then applied to the non-affinity-based purification of a mAb-aggregate challenge in CHO, an expression system with a much higher HCP burden. Although this in silico tool is effective at identifying successful purification processes for specific products, we hypothesized the existence of a small set of optimally orthogonal resins which could be successful for purifying most protein products. As a result, using the concepts and lessons learned from this process development work, we wanted to gain a deeper insight into the nature of orthogonal selectivity and identify resin sets which operate orthogonally in a product-agnostic manner. The extent of an orthogonal separation in preparative chromatography is a concept that many practicing chromatographers intuitively and heuristically understand yet cannot easily quantify. In particular, the extent of orthogonality in multimodal separations can be particularly challenging to intuitively predict due to the complexity and synergy of different modes of interactions. To understand and quantifying these nonintuitive orthogonality trends, we developed a novel mathematical framework to characterize and quantify orthogonality in multimodal systems. Using this, we observed several interesting and unexpected results including the existence of a highly orthogonal pair of resins belonging to the same class/family. Taken together with the in silico tool, the work presented in this thesis is not only instrumental for use in improving the efficiency of process development, but can also have significant utility for the design of next-generation multimodal ligands.Ph
Finite element formulation of laser material interaction accounting for geometry evolution
No full textAugust 2023School of EngineeringLaser subtractive and additive manufacturing holds tremendous potential to create intricate parts in small batches. However, due to the complexity of the process physics, it is not currently possible to predict a priori the accuracy of the part geometry or material quality. Numerical simulation with the capability of providing insightful prediction to reassure partprecision and quality in advance is desired, where numerical accuracy and computational performance are equally important aspects to consider.
This thesis focuses on developing a generalized finite element framework to simulate the laser-material interaction process, including additive manufacturing (powder bed fusion) and subtractive manufacturing (pulsed laser ablation). In this work, evolution of the part geometry is predicted. For the laser grooving process, material ablation is calculated and the moving material front is tracked under a pulsed laser source. For the Laser Powder Bed Fusion (LPBF) process, the melt pool dimension is computed and its boundary is tracked as the new powder layers are added to the build plate. State variable fields are introduced to solve the phase change physics. Numerical parameters are defined as a priori by a calibration step. Powder consolidation is considered and deformation is accounted by updating the reference configuration. A method is implemented to track the moving material front for multilayer simulation. LPBF experiment samples with multiple powder
layers are imaged and quality is correlated to processing parameters based on an analytical model, and further used to validate the finite element simulation result. A three-dimensional transient parallel finite element model is developed for laser grooving and LPBF processes, which may be extended to the entire manufacturing process that enables direct comparison to experiments.Ph
Understanding the response of glasses to sharp contact loading via classical molecular dynamics simulation
No full textDecember 2022School of EngineeringGlass materials have enabled many products for a wide range of applications. One downside of oxide glasses is that they are very susceptible to crack under sharp contact loading. There have been extensive studies in experiments to understand the deformation mechanisms and to improve crack resistance of glass. However, conducting in-situ characterizations in real-time experimentally is challenging, which has mostly limited our understanding of glass deformation and cracking behaviors under sharp contact loading. This study conducted nanoindentation tests using classical molecular dynamics simulation to understand how the stress/strain fields and the structure of glass evolve under sharp contact loading from an atomistic perspective. First, we used the 2.5D nanoindentation method in sodium aluminosilicate and sodium aluminoborate systems to illustrate the structural origin of the improved crack resistance of boron-containing glass. To generate stress/strain fields and deformation patterns in glass that can be directly compared with instrumented experimental studies, we developed a 3-D nanoindentation protocol to mimic real-life loading conditions, and applied it in a model metallic glass favoring shear flow to understand the shear band activation/interaction mechanism. After validating our results with multiple experimental studies, we conducted more 3-D nanoindentation simulations in metallic and silica glasses with different densification abilities. The comparisons between silica and metallic glasses suggest that balancing shear deformation with a combination of instantaneous and permanent densification can provide multiple pathways to dissipate energy under indentation, thus increasing the load to initiate cracks and improving the damage resistance of glass.Ph
About measuring neutron lifetime
No full textMay 2023School of ScienceParticle accelerators are scientific instruments used to accelerate subatomic particles to extremely high speeds and energies, obtaining important perspectives into the fundamental properties of matter and the universe, exploring the fundamental forces of nature, and developing new materials and technologies. Neutrons are subatomic particles found in the nucleus of atoms. One important use of particle accelerators is the generation of free neutron for neutron lifetime measurements. Neutron properties are challenging to research because they are neutral particles that do not easily interact with matter. However, scientists are able to analyze their behavior using techniques such as neutron lifetime measurements, neutron scattering and neutron diffraction. For research utilizing neutrons, the decay rate of neutrons, which is unaffected by their energy level, is typically not a factor.This thesis provides background information on neutrons and reviews research on the measurement of neutron lifetime using the beam technique and the bottle technique, which are commonly used today. In addition, it describes the particle accelerator facilities currently available and in development for such research, such as the Gaerttner Linear Accelerator (LINAC) Center at Rensselaer Polytechnic Institute (RPI) and the Turkish Accelerator and Radiation Laboratory (TARLA) in Turkey. These facilities and their capabilities for neutron generation utilizing the Bremsstrahlung technique are discussed.M
Analyzing and modeling human behavioral dynamics in social networks
No full textDecember 2021School of ScienceUnderstanding, analyzing, and modeling human behavior within the context of social networks is a constantly ongoing study, being one of the central focuses of many scientific disciplines, include social and network science. Human behavioral dynamics are complex and multifaceted, but the advent of digital communications and online social media has made researching some of the driving mechanics and underlying trends more feasible. And while concepts such as value homophily, polarization, and relationship dynamics are integral to how people interact with each other, only recently have we begun to analyze, model, and forecast such behavior in a quantifiable way. And even with the advance of this kind of research, many aspects of human behavior are still not accurately modeled beyond simulations. In this thesis we seek to elucidate some behavioral dynamics through comprehensive analysis and modeling with empirical validation. In particular, we investigate the behavioral aspects mentioned above: value homophily, polarization, and relationship dynamics (in the context of tie strength analysis), with the addition of conversational dynamics. First, we analyze value homophily and how it drives the creation, evolution, and dissolution of groups. Using insights gained from a unique dataset, we formulate a measure of utility that individuals gain from groups based on their alignment with the opinions of other group members. We hypothesize that group membership changes (as well as opinion changes) are driven by a need to maximize this utility. We empirically validate this hypothesis by analyzing how many actual group membership changes occur within the data, and how many of them improve utility of the individuals making the changes, improve the utility of the changing individuals' like-minded peers, and improve the utility of all affected members. We then analyze how membership changes are affected by the popularity of the opinions held by those making the changes, and how utility gains differ as a consequence. Finally, we implement the mechanic of utility maximization in a predictive analytical model, showing how such a model can accurately forecast group membership and opinion changes. Next, we investigate political polarization on a comprehensive scale, using a massive longitudinal Twitter dataset covering the 2016 and 2020 U.S. presidential elections. Enhanced with media classifications allowing us to identify fake news, extremely biased news, and traditional news propagated within our data, we extract Twitter users that are ``super-spreaders'' of these types of content. We analyze these influencers, measuring their shifts in degree of influence and affiliations (or lack thereof) from 2016 to 2020. We then analyze how the user base of our Twitter data polarize themselves with respect to our media classifications, comparing this analysis between our two elections. We find that users have become increasingly polarized over time on this platform. We expand upon this result by also analyzing how influencers group together based on the similarity of the users that propagate their content to see if influencers inherently separate based on this basis, finding that, between the two elections, ever-tightening ``echo-chambers'' are formed around media classification, similarly increasing polarization. We then introduce a novel framework for continuously predicting tie strengths of dyadic connections over long periods of time. With a suite of analytical and machine learning models implemented within this framework for a pair of longitudinal datasets, we explore the upper bounds of prediction for both single-dataset predictions and cross-domain predictions. Then, taking advantage of the uniquely continuous nature of our generated tie strengths, we analyze these values over the length of our data, augmenting them with relationship classifications to investigate long-term relationship dynamics, observing strong trends that both reinforce, and expand upon, previous related literature. Finally, we design an ensemble of unique, text-agnostic measures to characterize conversational dynamics as they occur within forum-like online social media, exploring the relationship between conversational structure and content topics. We use a Support Vector Machine to accurately classify different content based on their genre using only our measures. We then cluster content using our measures, finding that the resultant groupings effectively delineate topical divisions inside genres, with finer clusters even capturing subtle semantic differences within singular topics. We find that the distance between clusters also correlates to difference in content, with farther clusters being less related in content than closer clusters. We conclude our cluster analysis by using an outlier detection algorithm to identify content that is, according to our measures, substantially different from the content of inliers. We investigate these identified outliers with their associated text and discuss the different aspects that contribute to their differences from the majority.Ph
Defect engineering for tuning electronic and optical properties of materials
No full textAugust 2023School of ScienceAll pristine nanostructured materials are susceptible to disruptions in their chemical composition and expected crystal structure, where these disruptions are commonly referred to as defects. At first glance, defects may seem to be detrimental to the quality of a material, but they can actually improve the usability of the material and cause the material to perform even better for specific applications. In this thesis, density functional theory (DFT) calculations are applied to various defects across multiple nanostructured materials to identify the defects which are potentially responsible for observations found in experiments, and what makes those specific defects special and how those defects can optimize specific target applications. Defects in transition metal dichalcogenides (TMDs), hexagonal boron nitride (hBN), perovskites, and TMD alloys are investigated. For TMDs, an oxygen interstitial (Oins) defect in monolayer WSe2 is found to likely be responsible for single photon emission (SPE) and this Oins defect can assist in enhancing the adsorption strength for the sensing of ammonia (NH3) molecules. Additionally, transition metal defective MoS2 can assist in the sensing of neurotransmitter biomolecules. In hBN, the antisite defect (NBVN) is the most likely defect responsible for SPEs and its SPE qualities are enhanced in multilayer hBN when compared to monolayer hBN. The NBVN defect also assists in the migration of ions within an hBN anode/LiNiMnCoO2 cathode system. For perovskites, multiple dopants are introduced into the BaZrS3 perovskite to determine the best candidates to tune the bandgap into the Shockley-Queisser limit to potentially maximize photovoltaic efficiency. From the dopant dataset, a machine learning (ML) model is created which can accurately predict formation energies and bandgaps, and Ti and Ca substituting at the Zr and Ba sites, respectively, are deduced as the best dopants for potential photovoltaic devices. For alloys, multiple possible configurations are constructed in a WxMo1-xS2 alloy to determine the behavior of the electronic transitions when a single sulfur vacancy (VS) is introduced with models based on the geometry and chemical composition of the alloy being used to predict the energies for the A exciton and defect-mediated transitions. This thesis attempts to deduce the best defects in its respective nanomaterial to operate for its specific target application and answer why those defects are the best choice across all the aforementioned applications.Ph
First-principles electron transport and electron-phonon coupling in thin films and interfaces
No full textDecember 2022School of EngineeringElectron-phonon interactions have implications in some of the most important current and future technologies like integrated circuits, solar cells, superconductors, spintronics, and quantum information. Though the study of electron-phonon scattering is a decades-old problem, up until recently it has mostly evaded a rigorous and fully first-principles treatment. In this work, we use parameter-free ab initio techniques for modeling electron transport and electron-phonon coupling in metallic films and nanostructures. We focus our efforts on two specific applications—plasmonics and electrical interconnects. In the field of plasmonics, we model the optical response of metallic nanoparticles in pump-probe spectroscopy. Generation of hot carriers followed by thermalization through electron-electron and electron-phonon scattering is a complex phenomenon. However, a multiscale approach which is able to capture the coupling between individual electronic states and phonon modes is able to predict the optical response with quantitative accuracy. We look at a long-standing puzzle in the ultrafast dynamics field which has evaded explanation for many years. The ultrafast pump-probe measurements of aluminum show a slow rise time and no decay in stark contrast to gold, even though carriers relax faster in aluminum by both electron-electron and electron-phonon scattering. We identify strong electron-phonon coupling and insensitivity of probe response to electron temperature as the solution to this long-standing puzzle. In our second research thrust, we study electron-phonon coupling in thin metallic films. A major hurdle in scaling down the size of integrated circuits is the increased resistivity of metallic Back-End-Of-Line (BEOL) interconnects at smaller dimensions. Enhanced surface and grain boundary scattering in the narrow limit leads to dramatic increase in the resistivity of the metal nanowires. Research in the last decade has focused on the search for highly conductive elemental metals like Co, Ru, Ir and Rh which could potentially replace the ubiquitously used Cu. Here, we explore the use of new classes of materials including intermetallics, metallic carbides, oxides, nitrides and topological metals and semimetals as next-generation interconnect materials. We show that metals with suitably anisotropic Fermi velocity distributions can strongly suppress electron scattering by surfaces and outperform isotropic conductors such as copper in nanoscale wires. We derive a corresponding descriptor for the resistivity scaling of anisotropic conductors, screen thousands of metals using first-principles calculations of this descriptor and identify the most promising materials for nanoscale interconnects. Previously-proposed layered conductors such as MAX phases and delafossites show promise in thin films, but not in narrow wires due to increased scattering from side walls. We find that certain intermetallics (notably CoSn) and borides (such as YCo3B2 ) with one-dimensionally anisotropic Fermi velocities are most promising for narrow wires. Combined with first-principles electron-phonon scattering predictions, we show that the proposed materials exhibit 2-3× lower resistivity than copper at 5 nm wire dimensions. For its potential as low-resistance interconnects, we pursue a fundamental understanding of electron transport properties of thin films of topological metals/semimetals. Studies have shown that in the case of Cu, the resistance-area RA product remains constant for pristine films and increases for films with defects with decreasing thickness. Our study show that the RA product decreases with film thickness for a Weyl semimetal—NbAs. This is attributed to the disproportionately large number of surface conduction states which dominate the ballistic conductance by up to 70%. The results presented here underscore the promise of topological semimetals as a future BEOL interconnect metal. Lastly, we find that some of the candidates we shortlisted for interconnects have significant advantages for efficient hot carrier harvesting. We show that the optical response of film-like conductors, PtCoO2 and Cr2AlC, resemble that of 2D metals, while that of wire-like conductors, CoSn and YCo3B2 , resemble that of 1D metals, which can lead to high mode confinement and efficient light collection in small dimensions, while still working with 3D materials with high carrier densities. Carrier lifetimes and transport distances in these materials, especially in PtCoO2 and CoSn, are competitive with noble metals. Most importantly, we predict that carrier injection efficiency from all of these materials into semiconductors can exceed 10% due to the small component of carrier momentum parallel to the metal surface, substantially improving upon the typical < 0.1% injection efficiency from noble metals into semiconductors.Ph