Physical Sciences and Mathematics Commons^™

Experimental Evidence That Shear Bands In Metallic Glasses Nucleate Like Cracks, Alan A. Long, Wendelin Wright, Xiaojun Gu, Anna Thackray, Mayisha Nakib, Jonathan T. Uhl, Karin A. Dahmen

Faculty Journal Articles

Highly time-resolved mechanical measurements, modeling, and simulations show that large shear bands in bulk metallic glasses nucleate in a manner similar to cracks. When small slips reach a nucleation size, the dynamics changes and the shear band rapidly grows to span the entire sample. Smaller nucleation sizes imply lower ductility. Ductility can be increased by increasing the nucleation size relative to the maximum (“cutoff”) shear band size at the upper edge of the power law scaling range of their size distribution. This can be achieved in three ways: (1) by increasing the nucleation size beyond this cutoff size of the …

Recall Distortion In Neural Network Pruning And The Undecayed Pruning Algorithm, Aidan Good, Jiaqi Lin, Hannah Sieg, Mikey Ferguson, Xin Yu, Shandian Zhe, Jerzy Wieczorek, Thiago Serra

Faculty Conference Papers and Presentations

Pruning techniques have been successfully used in neural networks to trade accuracy for sparsity. However, the impact of network pruning is not uniform: prior work has shown that the recall for underrepresented classes in a dataset may be more negatively affected. In this work, we study such relative distortions in recall by hypothesizing an intensification effect that is inherent to the model. Namely, that pruning makes recall relatively worse for a class with recall below accuracy and, conversely, that it makes recall relatively better for a class with recall above accuracy. In addition, we propose a new pruning algorithm aimed …

Acid Mine Drainage In The Shamokin Creek Watershed: A Spatial Analysis Of Economic And Environmental Consequences Of Coal Mining, Ben Shimer

Student Project Reports

No abstract provided.

Training Thinner And Deeper Neural Networks: Jumpstart Regularization, Carles Riera, Camilo Rey, Thiago Serra, Eloi Puertas, Oriol Pujol

Faculty Conference Papers and Presentations

Neural networks are more expressive when they have multiple layers. In turn, conventional training methods are only successful if the depth does not lead to numerical issues such as exploding or vanishing gradients, which occur less frequently when the layers are sufficiently wide. However, increasing width to attain greater depth entails the use of heavier computational resources and leads to overparameterized models. These subsequent issues have been partially addressed by model compression methods such as quantization and pruning, some of which relying on normalization-based regularization of the loss function to make the effect of most parameters negligible. In this work, …

Strengthening A Linear Reformulation Of The 0-1 Cubic Knapsack Problem Via Variable Reordering, Richard Forrester, Lucas Waddell

Faculty Journal Articles

The 0-1 cubic knapsack problem (CKP), a generalization of the classical 0-1 quadratic knapsack problem, is an extremely challenging NP-hard combinatorial optimization problem. An effective exact solution strategy for the CKP is to reformulate the nonlinear problem into an equivalent linear form that can then be solved using a standard mixed-integer programming solver. We consider a classical linearization method and propose a variant of a more recent technique for linearizing 0-1 cubic programs applied to the CKP. Using a variable reordering strategy, we show how to improve the strength of the linear programming relaxation of our proposed reformulation, which ultimately …

Positivity Among P-Partition Generating Functions, Nathan R. T. Lesnevich, Peter R. W. Mcnamara

Faculty Journal Articles

We seek simple conditions on a pair of labeled posets that determine when the difference of their (P,ω)-partition enumerators is F-positive, i.e., positive in Gessel's fundamental basis. This is a quasisymmetric analogue of the extensively studied problem of finding conditions on a pair of skew shapes that determine when the difference of their skew Schur functions is Schur-positive. We determine necessary conditions and separate sufficient conditions for F-positivity, and show that a broad operation for combining posets preserves positivity properties. We conclude with classes of posets for which we have conditions that are both necessary and …

Environmental Racism In Baltimore: A Geographical Study Into The Connections Between Environmental Toxins And Public Health, Genevieve Block

Honors Theses

An investigation into the relationship between environmental toxins and environmental racism in Baltimore City, Maryland.

Evaluating The Relationship Between Floodplain Topography And Channel Avulsion: Evidence From The Devonian Catskill Formation, North-Central Pennsylvania, Usa, Molly O'Halloran

Honors Theses

Topographic complexity on floodplains can route flow, control sediment dispersal, and influence channel behavior, but studying floodplain-channel interactions in modern rivers is challenging because of human modifications and the short timescales of observable data. This project assesses the link between different types of floodplain microtopography and avulsion style in the Devonian Catskill Formation, north-central Pennsylvania, where thick stacks of fluvial strata provide a lengthy record of channel-floodplain interaction. Using a combination of field observations and computer modeling, this study identifies sedimentary features indicative of floodplain complexity and analyzes their impact on avulsion style at fourteen Catskill Formation outcrops.

Based on …

A Contraction Based Approach To Tensor Isomorphism, Anh Kieu

Honors Theses

Tensor isomorphism is a hard problem in computational complexity theory. Tensor isomorphism arises not just in mathematics, but also in other applied fields like Machine Learning, Cryptography, and Quantum Information Theory (QIT). In this thesis, we develop a new approach to testing (non)-isomorphism of tensors that uses local information from "contractions" of a tensor to detect differences in global structures. Specifically, we use projective geometry and tensor contractions to create a labelling data structure for a given tensor, which can be used to compare and distinguish tensors. This contraction labelling isomorphism test is quite general, and its practical potential remains …

Indoor Air Quality Through The Lens Of Outdoor Atmospheric Chemistry, Jonathan P.D. Abbatt, Douglas B. Collins

Faculty Contributions to Books

Outdoor atmospheric chemistry and air quality have been the topic of research that intensified in earnest around the mid-20^th century, while indoor air quality research has only been a key focus of chemical researchers over the last 30 years. Examining practices and approaches employed in the outdoor atmospheric chemistry research enterprise provides an additional viewpoint from which we can chart new paths to increase scientific understanding of indoor chemistry. This chapter explores our understanding of primary chemical sources, homogeneous and multiphase reactivity, gas-surface partitioning, and the coupling between the chemistry and dynamics of indoor air through the lens of …