Historically, numerical analysis has formed the backbone of supercomputing for decades by applying mathematical models of first-principle physics to simulate the behavior of systems from subatomic to a galactic scale. Recently, scientists have begun experimenting with a new approach to understanding complex systems using machine learning (ML) predictive models, primarily Deep Neural Networks (DNN), trained by the virtually unlimited data sets produced from traditional analysis and direct observation. Early results indicate that these “synthesis models” combining ML and traditional simulation, can improve accuracy, accelerate time to solution and significantly reduce costs.

In this thesis, we study how to enhance the usability of machine learning models to accelerate HPC applications. We first study an application, the Eulerian fluid simulation. The Eulerian fluid simulation is an important HPC application. The current methods that accelerate the fluid simulation with Neural Networks (NNs) lack flexibility and generalization. In this application, we tackle the above limitation and aim to enhance the applicability of NNs in the Eulerian fluid simulation. We introduce Smart-fluidnet, a framework that automates model generation and application. Given an existing NN as input, Smart-fluidnet generates multiple NNs before the simulation to meet the execution time and simulation quality requirement. During the simulation, Smart-fluidnet dynamically switches the NNs to make best efforts to reach the user’s requirement on simulation quality. Evaluating with 20,480 input problems, we show that Smart-fluidnet achieves 1.46x and 590x speedup comparing with a state-of-the-art NN model and the original fluid simulation respectively on an NVIDIA Titan X Pascal GPU, while providing better simulation quality than the state-of-the-art model.

Our scheme for the tagless purification of water soluble complexes. 10 g of protein from a crude bacterial extract is first fractionated by ammonium sulfate precipitation and then by a series of chromatographic steps: anion exchange (IEX), hydrophobic interaction (HIC), and finally size exclusion (Gel Filtration). Fractions from the last chromatography step are trypsin digested and peptides labeled with iTRAQ reagents to allow multiplexing and quantitation during mass spectrometric analysis. Elution profiles of identified proteins are then subjected to clustering analysis.

During the COVID-19 pandemic, inmates at USP Lompoc were subjected to increased exposure of the virus due to inaction from staff, guards, and the Bureau of Prisons (BOP). This paper focuses on the history behind prisons and incarceration as a punishment, the timeline of the COVID-19 pandemic in relation to events happening in Lompoc, and includes interviews from two former inmates at USP Lompoc, Bernd Appleby and Ron Shehee. Looking at the progression of the COVID-19 pandemic and how it affected inmates at Lompoc, the conscious refusal by staff and guards to follow instructions given by the Attorney General, the absence of personal protective equipment, and the lack of structure and testing led to the death and health issues of inmates. Appleby and Shehee highlight the apathy, power imbalance, and inhumane conditions they faced during their time at USP Lompoc. They share specific details into accounts from other prisoners and the health and administrative problems that inmates faced.

Incarceration not only affects inmates but their communities as well and more visibility around the injustices at USP Lompoc is needed to show that the ineffectiveness of prisons to handle a pandemic comes at the cost of inmates and their lives.

Tetralogy of Fallot (TOF) is one of the more complicated types of congenital heart defects. TOF a ects 3-5% of all infants born with congenital heart defects and is a leading cause of cyanosis, which occurs when insu cient amount of oxygen are transported to the rest of body. Four major defects contribute to cyanosis in patients with TOF.