DSpace@RPI (Rensselaer Polytechnic Institute)
Not a member yet
6809 research outputs found
Sort by
Measurements and evaluation of 54-fe in the kev neutron energy region
No full textMay2025School of EngineeringAccurate neutron-induced reaction data is at the precipice of various nuclear applicationsthat require high accuracies to reduce uncertainties in calculations involving nuclear data
when simulating nuclear systems. These calculations can vary widely, ranging from burnup
calculations of nuclear power reactors, predictions of stellar abundances, radiation dose calculations behind shielding walls, and more. The nuclear data evaluation process ensures that
the latest high fidelity nuclear data are available for end users to make use of in their simulations and calculations. Iron is an important constituent of many nuclear-grade materials,
and relative to the major isotope of 56Fe, the cross-sections of the minor isotopes of Fe are
not as well-characterized and understood.
RPI has updated the RRR evaluation of 54Fe to improve upon the previously evaluated neutron radiative capture cross section as well as the energy-energy covariance of 54Fe
cross sections. To improve upon the existing 54Fe evaluation, new differential neutron TOF
measurements were performed at the Gaerttner LINAC Center. A radiative capture yield
measurement was performed using an enriched 0.021 a/b metallic 54Fe sample using an array
of C6D6 detectors to provide additional capture yield data below 1 MeV. This measurement
was in good agreement with a previous n TOF experiment, and both suggest that the capture cross section in ENDF/B-VIII.0 is too low. Additionally, a transmission experiment
was conducted using a 6Li glass detector on the same enriched 54Fe sample to constrain the
resonance evaluation. Full experimental nuclear data covariances were generated for both
experiments and were propagated to the fitting of resonance parameters with SAMMY. Together with these experiments and other relevant EXFOR data, an improved set of resonance
parameters were generated along with a resonance parameter covariance matrix (MF=32).
The updated RPI 54Fe evaluation improved the agreement to RPI and EXFOR differential data. Additionally, the RPI evaluation was tested against integral critical experiments
and other shielding experiments and found to have maintained the agreement between calculated and experimental values that was observed with previous evaluations.Ph
The v-generic filter is in meinong's jungle, and that's (probably) ok
No full textMay2025School of Humanities, Arts, and Social SciencesIn this paper, we offer some analysis of Joel David Hamkins's work 'The Set-Theoretic Multiverse' and take a look at some of the views concerning the existence of mathematical objects (and the kind of existence which they have). In particular, we will discuss the forcing, the method employed by Cohen to show that the negation of the Continuum Hypothesis is consistent with the axioms of ZFC, an object involved with a certain type of forcing, the so-called V-generic filter and what various schools of philosophers might say about this object and forcing proper.M
Converging-diverging nozzle simulations at varying nozzle pressure ratios
No full textMay2025School of EngineeringCompressible converging-diverging nozzles have a wide range of applications, including aerospacepropulsion, flow control and microfluidic flows commonly found in deposition, coating and
cooling techniques. When the ambient pressure is low, the continuum assumption does not
hold. With this aim in mind, compressible flow simulations near the continuum-rarefied
boundary are modeled using open-source software to increase the efficiency of converging-
diverging nozzles. This thesis employs OpenFOAM’s density-based solver, rhoCentralFoam,
to simulate nozzle flow dynamics under a range of operating conditions. Close comparisons
were made with the existing experimental and numerical data in the literature. Furthermore,
a comparative study is conducted between a unsteady Reynolds-Averaged Navier-Stokes
(URANS) based k − ω SST model and a Large Eddy Simulation (LES) k-equation model
to evaluate their accuracy in capturing turbulent structures and shock phenomena. Simu-
lations cover both free jet and impinging jet scenarios over a nozzle pressure ratio (NPR)
range of 8.78 to 296. For the impinging jet simulations, multiple distances from the wall
were simulated, specifically at 2.08 and 3.08 X/D. A grid convergence study was conducted
for the k − ω SST and k-equation model geometries to demonstrate mesh independence.
A time convergence study was performed for the k − ω SST model geometry and a batch
convergence study was performed for the k equation model geometry to verify temporal ac-
curacy. Experimental data from previous studies using an identical nozzle geometry validate
the numerical results, highlighting features such as mach disks, expansion plumes, and shock
structures across the continuum regime and near the continuum-rarefied boundary. The
findings provide insights into the benefits and limitations of each turbulence model, offering
guidelines for their application in high-performance nozzle design and aeroacoustics. Future
work on simulating impinging supersonic underexpanded jets can yield novel results on the
shock tones and aeroacoustic features present in unsteady impinging jets especially within
the rarefied regime.M
Probabilistic scene graph generation and its applications
No full textMay2025School of EngineeringScene graphs encode relationships between image entities as triplets (subject-relationship-object), where nodes represent grounded entities and directed edges define relationships from the subject to the object. The Scene Graph Generation (SGG) task faces significant challenges, including difficulty detecting small or occluded entities and classifying entities and relationships due to imbalanced class distributions and ambiguous annotations. As a result, SGG models often suffer from low accuracy and a bias toward frequently occurring classes. Existing methods employ techniques such as re-weighting training samples or post-processing inference results to mitigate the bias. However, these approaches often compromise overall accuracy, as they trade off general model performance for a more balanced class distribution. In this thesis, we leverage prior knowledge of scene graph triplets to enhance accuracy and mitigate bias in trained SGG models in a principled manner. We propose a Bayesian Network (BN) to capture the stable within-triplet prior and a Conditional Random Field (CRF) to model the between-triplet prior of scene graph triplets. BN inference, when applied to uncertain evidence from a biased SGG model, improves the overall accuracy as well as mitigates bias. The CRF further refines predictions by integrating unary potentials derived from the BN posterior with pairwise potentials, representing the between-triplet prior learned from triplet co-occurrence statistics. Beyond improving performance in static scene graphs, we explore the challenge of integrating both static and temporal potentials in Dynamic Scene Graph (DSG) generation. Existing methods implicitly assume that all relationships in DSG are purely temporal, neglecting their static components. To address this, we propose a Transformer-based CRF model that effectively captures both static and long-term temporal potentials, demonstrating its superiority over traditional Transformer-based approaches. Finally, we showcase the effectiveness of scene graphs as a bridge for Visual Question Answering (VQA). Prior works on SG-based VQA assume that every question can be answered solely from the perfect scene graph, leading to poor performance on questions unrelated to the scene graph. To overcome this limitation, we introduce an uncertainty-guided approach that combines predictions from two Bayesian ensembles: one for image-based VQA and another for SG-based VQA, ensuring more robust and accurate question answering.Ph
Current limitation of ai in education
No full textMarch2025Information Technology and Web Science ProgramThis study examines the effectiveness and limitations of artificial intelligence (AI) in education, focusing on the subjects where AI can enhance student learning and the barriers to its widespread adoption. Using a semantic review methodology, the research synthesizes existing literature to assess AI's impact across different academic domains. Findings indicate that AI shows promise in subjects that benefit from personalized learning and data-driven insights, such as mathematics and language learning. However, significant challenges remain, including ethical concerns, data privacy risks, and the potential to exacerbate educational inequalities. These findings underscore the need for a balanced approach to AI integration, ensuring that technological advancements align with principles of equity and ethics. Further research, incorporating empirical methods, is recommended to deepen understanding and address these challenges.M
Interplay between resin media, feed constituents, and process conditions in aav affinity chromatographic systems
No full textDecember 2024School of EngineeringThe emergence of gene therapies has provided a viable approach for treating previously incurable diseases and disorders. Adeno-associated virus (AAV) vector-based therapies have gained attention in clinical trials due to their efficient delivery of therapeutic genes. However, high costs of therapy, ranging from 0.8 to 3 million dollars per dose, render these treatments difficult to access for many patients.A major contributor to these exorbitant costs is the expenses incurred during the complex AAV manufacturing process. These costs arise from various sources, such as the use of expensive raw materials, low-scale production units, and poor yields during upstream processing. Additionally, the generation of high levels of impurities during the manufacturing process has led to substandard purification platforms that can lose up to 70% of the product.
Downstream processing ubiquitously relies on affinity chromatography as the AAV capture step. The expensive affinity resins used in this step are typically limited to single use in a cGMP process before needing replacement due to risks to AAV product quality in subsequent cycles. To cut down material costs, resin reuse is advantageous. However, the lack of a mechanistic understanding relating to the behavior of AAV products and feedstock constituents in existing affinity systems poses a major challenge in achieving an extended column lifetime. To fill this gap, this thesis presents the underlying factors that influence AAV affinity capture chromatography performance in relation to resin reusability.
To understand the effect of AAV capsid design, feed material constituents, and column recycling on AAV purification behavior, the elution behavior of model AAV9 vectors with varying viral protein (VP) ratio was investigated on AAV9-specific POROS CaptureSelect AAV9 and pan-serotype selective POROS CaptureSelect AAVX affinity resin columns. It was found that in the pure state, the vector types displayed consistent elution profiles with 75% product recovery. However, clarified lysate purification resulted in only 50-65% product recovery with inconsistent chromatographic profiles from column recycling. This suggested that the impurities in the feed stream significantly impacted the affinity chromatography performance and column reusability.
To discern the consequence of the in-process variabilities induced due to impurities on AAV product quality and column performance, individual fractions obtained during column recycling were examined using a combination of analytical techniques to assess purity, aggregation and process-related impurity profiles. The results showed that the critical quality attributes that were most affected by column reuse were functional product titer, aggregates, and host-related dsDNA and chromatin impurities. Moreover, the impurity content in the product eluate progressively increased from AAV9-specific affinity column for the least stable capsid types. These findings collectively demonstrated that feed impurities, affinity resin characteristics, elution pH, column clean-in-place (CIP), and vector stability impact affinity column recyclability and AAV product quality.
The next area of research investigated column fouling in affinity resins applying a series of complementary techniques, testing the hypothesis that the increased impurity co-elution in the product eluate during column recyling was likely due to residual carryover caused by incomplete column CIP, leading to fouling of columns. Confocal laser scanning microscopy (CLSM) was employed to visualize and characterize resin beads post-CIP, identifying the location and extent of fouling due to feed remnants across multiple cycles of column reuse. These studies were complemented with structural imaging and analysis of resin architecture and associated transport mechanics using nanoscale X-ray computed tomography (CT). The confocal micrographs of used resin materials confirmed fouling of POROS affinity resin materials. Fouling occurred due to blockage of resin pores by both vector-related and process-related contaminants. Further investigations revealed that although the high impurity burden in the feed contributed to reducing the lifetime of POROS AAV affinity columns, the primary factors affecting reusability were the characteristics of the resin base matrix and the ligand.
All in all, this thesis provides a comprehensive understanding of the challenges and promise of affinity-based downstream processing for AAV vector purification, offering essential insights that can help advance capture technologies and drive innovation in gene therapy vector purification for both academic and industrial applications.Ph
Molecularly-induced effects on the synthesis and properties of thin film inorganic/molecular-nanolayer interfaces and their multilayers
No full textMay2025School of EngineeringInserting molecular nanolayers (MNLs) at inorganic thin film interfaces has been shown to enhance chemical and mechanical stability, and access unexpected electrical/thermal transport and mechanical responses. Stacking inorganic nanolayers and MNLs offer the potential for crafting new classes of high-interface-fraction multilayered composites with emergent responses arising from the superposition of effects from multiple MNL interfaces. This work demonstrates studies on the synthesis of metal-oxide/MNL multilayers and metal/MNL/metal sandwiches, and their mechanical and acoustic properties. Synthesis techniques used include low-temperature atomic layer deposition (ALD) or sputter deposition combined with MNL formation from vapor-phase molecular flux exposures. Results of experiments combining multiple spectroscopy, microscopy, and diffraction techniques unveil different correlations between MNL structure and chemistry on inorganic nanolayer growth kinetics, chemistry, morphology, phase stability, and oxidation, as well as provide insights into their atomistic mechanisms. Ab initio molecular dynamics simulations were used to reveal MNL-induced strain-hardening and toughening in metal/MNL/metal sandwiches, with atomistic insights on the effects of MNL molecular chain length and terminal chemistry. Pump-probe time-domain Brillouin spectroscopy unveiled unusual enhancements in optoacoustic transmission in titania/MNL multilayers at selected sub-terahertz frequencies. This is attributed to MNL-induced global optical effects and interference of acoustic trains reflected from MNL interfaces, and hence, sensitive to and tunable via MNL structure and chemistry. Such tunable MNL-induced emergent responses in inorganic/MNL multilayers could open new vistas for viscoelastic bandgap engineering and phononic laser development.Ph
Towards emotional reasoning by dialogue agents
No full textMay2025School of ScienceEmotions play an important role in human interactions. They determine how people feel, express or respond in different situations. Therefore, emotions are an essential part of building human-like dialogue agents in various applications like healthcare, customer service, or psychotherapy. Despite major advancements in building emotion detection models, this remains an active research area as human emotions are very specific to people and domains. A significant limitation of state-of-the-art deep learning models for emotion detection is their inability to generalize across different application areas. These models rely on domain-specific fine-tuning data and predefined emotion labels that often fail to capture the nuanced emotional state when applied to new domains and situations. Our research addresses these limitations by developing the context-aware and generalizable emotion detection framework that enables dialogue agents to understand and respond to the user emotions without domain-specific fine-tuning. The major contributions of our research are:- Development of a generalizable emotion detection framework that is adaptable to unknown domains without re-training on domain-specific training data
- A generative approach to emotion detection and reasoning, inspired by the intuitive emotional reasoning process of humans in specific situations
- We focus on a task-oriented domain to demonstrate how emotion state tracking across dialogues can improve interactions in goal-driven dialogues. We show that adapting dialogue strategies based on the emotional state of the user significantly enhances the probability of task success, particularly in persuasion dialogues, where emotions influence decision making. By incorporating emotional reasoning into dialogue agents, this thesis contributes to the development of more emotion-aware adaptive dialogue agents for use across many real-life applications.Ph
Molecular mechanisms underlying the commitment to division in budding yeast
No full textMay2025School of ScienceThe regulation for cell size homeostasis is evolutionarily conserved and is vital for organismal survival across the tree of life. The coordination between cell growth and division for cell size homeostasis occurs primarily at the G1/S phase transition, termed Start in budding yeast. The transcriptional activation of ≈200 genes in the G1/S regulon by key transcription factor complexes, SBF and MBF, is essential for the progression of cell cycle post Start. This study focuses on the quantitative assessment of the molecular mechanisms of cell size regulation through investigation into the absolute copy number, concentration, localization and coordination of the Start machinery. Our lab has previously shown that the cell size dependent accumulation of SBF and MBF, and their subsequent titration of the target promoters throughout G1 are critical determinant of Start (Dorsey et al. 2018). This work expands on that study by investigating the dependence of Start on the copy number of Swi4, the DNA binding subunit of SBF, under different genetic contexts using single molecule fluctuation microscopy. We found that Swi4 transcription is partly SBF-mediated and is essential for Start transition. Further, a threshold level of about 170 Swi4 molecules titrating SBF binding sites in G1/S promoters predicted the effects of nutrients, ploidy, and G1/S regulatory mutations on cell size. Additionally, this work reveals the expression dynamics of the extremely low abundance and high turnover G1 cyclins, responsible for the phosphorylation of the SBF repressor, Whi5. Building on the past finding in our lab that the G1/S transcription factors assemble in increasing numbers of discrete clusters in G1 (Black et al. 2020), which suggest a possible clustering of their target promoters as well, we provide some preliminary results into the investigation of the super-resolution localization of the G1/S promoters in budding yeast. In combination, our work provides quantitative insights into the mechanisms of cell size regulation in budding yeast.Ph
An investigation of usability factors in designing a companion system for an immersive environment
No full textDecember 2024School of ArchitectureDialogue systems have become a popular research medium as recent advances in task-orientedand open-domain systems combined with deep learning technologies have increased the po-
tential for practical applications across many disciplines. One such domain of applications
involves multimodal dialogue systems deployed in interactive spaces that seek to provide
an immersive experience for participants. To take advantage of the opportunities allowed
by advances in the hardware and software aspects of immersive environments, this thesis
proposes a design concept for companion system deployed in the EMPAC Research Lab at
Rensselaer Polytechnic Institute and conducts two experiment aimed to demonstrate the
practicality of the audiovisual capabilities of the system. The companion system, named
Scalable Cognitive Immersive Learning & Analysis Room (SCILAR), will utilize a two-part
combination system that provides users with their own local device that is connected to a
shared global space. This shared global space is capable of spatial awareness with a multi-
modal immersive dialogue system and an array of audio and visual sensors that can track
multiple participants within this interactive space across both physical and virtual mediums.
The system will also provide a transcription of conversations that can be used for enabling
further functionality and is provided as-is to track conversation history. Testing on the us-
ability and practicality of this system was conducted via two user studies. The first study
involved a collaborative data analytics scenario that has participants analyze New York City
taxi trips using a made-to-task Google Maps web app and the Google Maps website. The
second study investigated the effect of an acoustic pointer to help with locating alphanumeric
pairs located in a virtual environment displayed on a panoramic screen. The findings of both
of these experiments show statistical significance in the reduction of task completion time
when compared to baseline conditions. This thesis aims to provide insight into interactive
spaces for education and analyst applications and demonstrate the potential capabilities of
combining spatial awareness with a multimodal dialogue system.Ph