No description available.

Thesis (M.S.)--Virginia Polytechnic Institute and State University, 1991. / Vita. Abstract. Includes bibliographical references (leaves 204-212). Also available via the Internet.

The problem treated in this thesis is to create a system to monitor multiple interacting simulators. The problem is encountered in an attempt to simulate urban growth. The resulting system has a supervisor which controls all of the components of the system, namely the I/O routines, the graphics routine, the command routines, and the simulators. The main task of the supervisor is to display whichever data values the user desires. This involves executing some simulators, and extracting the data values from the data base. The extraction process first finds an association path between the files in the data base in order to relate the variables being displayed. Then using the association path the physical values are extracted from the data base. The data values are then passed to the graphics routine to be displayed for the user. / Science, Faculty of / Computer Science, Department of / Graduate

This paper presents SIMQ, a method of simulation program generation through questionnaires. Simulation is a powerful technique for analyzing complex systems; Hut its, acceptance has been impeded by the high cost of computer simulations. The objective of SIMQ is to enhance the usefulness of computer simulation by reducing the cost of developing simulation programs. Related research is reviewed in order to outline the development of the concept of program generation and to provide a standard of comparison for SIMQ. The structure and use of SIMQ is explained and then illustrated through application to a demonstration problem. Initially, the discussion centers on a basic version of SIMQ. Subsequently an extension of SIMQ is presented which permits stochastic branching within the model. Finally, an interactive graphical front end for the SIMQ system is described. Several experiments were conducted to test SIMQ's efficiency and characteristics. Results of field tests are also reported. Possible further extensions to SIMQ are outlined in the final chapter. The major conclusion reached is that the concept of program generation through questionnaires is feasible and that the structure of the SIMQ system is valid and useful. / Business, Sauder School of / Graduate

No description available.

Graduation date: 1983

M.Sc. (Physics) / A computer program called RAUPV was developed to simulate one-dimensional p-n structures and photo-voltaic cells. In order to simulate multilayer structures, the device is devided into a large number of discrete points with variable spacings. The physical parameters are calculated at each point, subjected to given external boundary conditions at the endpoints of the device. The physical processes are formulated from first principles, in such a way that they can be handled by numerical methods. A Newton-Raphson iteration technique is used to solve the large number of coupled, linear equations. The simulation is formulated in such a way that the equations must be solved for three variables at each point: the electron potential, the quasi-Fermi level for electrons and the quasi-Fermi level for holes. For the case of thermodinamic equilibrium, Poisson's equation is solved. A formulation is developed to handle the equation numericaly for variable intervals. Expressions for the free carrier concentrations are obtained using Fermi-Dirac statistics. Expressions for the charge density in traps are also obtained and several types of boundary conditions are considered. The program is able to calculate the band structure, charge density, internal electric field and free carrier concentrations for any multi-layer device. For the non-equilibrium case, Poisson's equation is solved simultaneously with the two continuity equations for electrons and holes. A special formulation for the current densities was developed, to assure convergence during the iteration process. Recombination is formulated in terms of capture cross-sections of trap states within the gap. Several types of boundary conditions are considered. The program is able to calculate the current densities of electrons and holes within the device and yield as output the net current through the device for a given external applied voltage. A technique was developed for the Newton-Raphson iteration to work only with the diagonals of the matrix containing the partial derivatives. This technique saves much computing time and memory. Various techniques are built into the program to assure convergence and to decrease computing time. The solar spectrum is processed in order to calculate the optical exitations within the device. Multiple reflections are taken into account and an anti-reflection layer is also simulated. The program can thus calculate current-voltage curves for a photo-voltaic cell for any given spectrum. The program runs on a PC and is able to analise p-n structures in detail. It can be used to design photo-voltaic cells using fundamental physical principles as point of departure.

LD2668 .T4 1966 W368 / Master of Science

There are many examples of earthquakes whose surface expressions are much more complicated than the seismologically derived faulting models. Seismologists also have found seismic source complexity and improved seismicity data have shown that rupture may occur on irregular or multiple shear surfaces. To simultaneously map both geometrical and temporal variation of the seismic sources for a complex rupture history from observed seismograms, it is possible to use a time dependent moment tensor (TDMT) inversion. The TDMT inversion algorithm has been tested with three synthetic data examples with varying degrees of complexity. The first example demonstrates that a multiple source with no focal depth change can be recovered, and the source parameters of each of the subevents can be accurately determined. In the second case we allowed the depth to vary for subevents (9-km and 13-km source depth, respectively). The two subevents can be identified on the basis of simultaneous shape-change of the moment tensor elements along with non-causality and the size of the CLVD component. The third example introduced source complexity by having two subevents which overlapped in time. The overlapped period could be seen in the moment tensor elements as unusually abrupt changes in the time function shape. The TDMT inversion was also performed on long-period body waves for three earthquakes: the 1982 Yemen earthquake, the 1971 San Fernando earthquake, and the 1952 Kern County earthquake. The Yemen earthquake was mapped as two simple, normal-slip subevents (with onset timing of the second subevent 5 seconds after the first) without a significant component of left- or right-lateral displacement or source depth change. The San Fernando earthquake is interpreted as two shear dislocation sources with changing source depths, possibly indicating upward rupture propagation (from 13-km to 7-km). The interpretation of the TDMT inversion for the Kern County earthquake was also a double point source which propagates upward from 20-km to 5-km. The resultant moment tensor functions from inversion of the synthetic waveforms, a combination of isotropic and tectonic release, demonstrated that the tectonic release associated with underground nuclear explosion can be separated and identified if the source depth between the explosions and tectonic release are different.

A dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of requirements for the degree of Master of Science. Johannesburg, 2015. / Crowd simulation can be thought of as a group of entities interacting with one another. Traditionally, an animated entity would require precise scripts so that it can function in a virtual environment autonomously. Previous studies on crowd simulation have been used in real world applications but these methods are not learning agents and are therefore unable to adapt and change their behaviours. The state of the art crowd simulation methods include flow based, particle and strategy based models. A reinforcement learning agent could learn how to navigate, behave and interact in an environment without explicit design. Then a group of reinforcement learning agents should be able to act in a way that simulates a crowd. This thesis investigates the believability of crowd behavioural simulation via three multi-agent reinforcement learning methods. The methods are Q-learning in multi-agent markov decision processes model, joint state action Q-learning and joint state value iteration algorithm. The three learning methods are able to produce believable and realistic crowd behaviours.