You are here
Search results
(1  4 of 4)
 Title
 Computational developments for ab initio manybody theory
 Creator
 Lietz, Justin Gage
 Date
 2019
 Collection
 Electronic Theses & Dissertations
 Description

Quantum manybody physics is the body of knowledge which studies systems of many interacting particles and the mathematical framework for calculating properties of these systems. Methods in manybody physics which use a first principles approach to solving the manybody Schrodinger equation are referred to as ab initio methods, and provide approximate solutions which are systematically improvable. Coupled cluster theory is an ab initio quantum manybody method which has been shown to provide...
Show moreQuantum manybody physics is the body of knowledge which studies systems of many interacting particles and the mathematical framework for calculating properties of these systems. Methods in manybody physics which use a first principles approach to solving the manybody Schrodinger equation are referred to as ab initio methods, and provide approximate solutions which are systematically improvable. Coupled cluster theory is an ab initio quantum manybody method which has been shown to provide accurate calculations of ground state energies for a wide range of systems in quantum chemistry and nuclear physics. Calculations of physical properties using ab initio manybody methods can be computationally expensive, requiring the development of efficient data structures, algorithms and techniques in highperformance computing to achieve numerical accuracy.Many physical systems of interest are difficult or impossible to measure experimentally, and so are reliant on predictive and accurate calculations from manybody theory. Neutron stars in particular are difficult to collect observational data for, but simulations of infinite nuclear matter can provide key insights to the internal structure of these astronomical objects. The main focus of this thesis is the development of a large and versatile coupled cluster program which implements a sparse tensor storage scheme and efficient tensor contraction algorithms. A distributed memory data structure for these large, sparse tensors is used so that the code can run in a highperformance computing setting, and can thus handle the computational challenges of infinite nuclear matter calculations using large basis sets. By validating these data structures and algorithms in the context of coupled cluster theory and infinite nuclear matter, they can be applied to a wide range of manybody methods and physical systems.
Show less
 Title
 Design of a high performance, high availability, distributed file system
 Creator
 Ahuja, Chetan
 Date
 2001
 Collection
 Electronic Theses & Dissertations
 Title
 Largescale and high performance computations of complex turbulent reacting flows
 Creator
 Afshari, Asghar
 Date
 2006
 Collection
 Electronic Theses & Dissertations
 Title
 Scheduling for CPU Packing and node shutdown to reduce the energy consumption of high performance computing centers
 Creator
 Vudayagiri, Srikanth Phani
 Date
 2010
 Collection
 Electronic Theses & Dissertations
 Description

During the past decade, there has been a tremendous growth in the high performance computing and data center arenas. The huge energy requirements in these sectors have prompted researchers to investigate possible ways to reduce their energy consumption. Reducing the energy consumption is not only beneficial to an organization economically but also to the environment. In this thesis, we focus our attention on high performance scientific computing clusters. We first perform experiments with the...
Show moreDuring the past decade, there has been a tremendous growth in the high performance computing and data center arenas. The huge energy requirements in these sectors have prompted researchers to investigate possible ways to reduce their energy consumption. Reducing the energy consumption is not only beneficial to an organization economically but also to the environment. In this thesis, we focus our attention on high performance scientific computing clusters. We first perform experiments with the CPU Packing feature available in Linux using programs from the SPEC CPU2000 suite. We then look at an energyaware scheduling algorithm for the cluster that assumes that CPU Packing is enabled on all the nodes. Using simulations, we compare the scheduling done by this algorithm to that done by the existing, commercial Moab scheduler in the cluster. We experiment with the Moab Green Computing feature and based on our observations, we implement the shutdown mechanism used by Moab in our simulations. Our results show that Moab Green Computing could provide about an 13% energy savings on average for the HPC cluster without any noticeable decrease in the performance of jobs.
Show less