|james francis toy iv
|james the GIANT killer: Evaluating Locking Schemes in the FreeBSD Kernel
|Producing Better Output From Finite State Verification Systems
|Alexander T. Likos
|Training an Artificial Neural Network to Sail a Boat
|Carlon A. McPherson
|Union through Data: Aesthetic Information Visualizations
|David G. Sayles
|Evofabber: Evolutionary Fabrication
|Stand Alone Complex::Open Your Eyes
|Some links below may require MS Powerpoint 2007 or Adobe Reader
|Click on a student picture for a larger view.
james francis toy iv
Advisor: Professor Hemmendinger
This project is concerned with locking strategies in the FreeBSD unix kernel. Locks prevent race conditions in a multi-programming environment. At present the FreeBSD kernel employs a GIANT_LOCK to prevent race conditions by locking access to the entire kernel. This method may inhibit concurrency; however, it is a simple and easy way to ensure correctness. The goal of this project is to see if fine-grained locking (FGL), an alternative to the GIANT_LOCK, promotes concurrency and throughput by locking only critical shared resources. FGL was implemented for two kernel subsystems, and we measured the time that a thread was waiting to acquire a lock; FGL showed zero wait time and the GIANT_LOCK showed non-zero wait time.
Advisor: Prof. Cass
Finite State Verification Systems are powerful tools that people can use to find elusive bugs in their programs. As powerful as they are, all FSVs have different outputs, but none of them are designed to be user-friendly. We set out to find one coherent output that will be more effective than current outputs in assisting the users of FSVs.
Alexander T. Likos
Advisor: Kristina Striegnitz
The goal of this project was to train a back propagation artificial neural network how to Sail a small sail boat using information gathered from a realistic sailing simulator.
Carlon A. McPherson
Advisor: Prof. Orellana and Prof. Striegnitz
In this project, several visualizations relating to various types of data, including demographics, race, energy, and waste, were created. The data used was related to Union College. The point of these visualizations is to highlight significant and important trends in the data. The data is represented through different visual metaphors that are meant to engage the user on a more in-depth level than simply presenting the information in the most easily understandable way.
David G. Sayles
Advisor: Prof Striegnitz and Prof Rieffel
The successful construction of evolved object designs is challenging. Physics simulators are a means to replicate the real world virtually, but phenomena such as the fabrication gap prevent certain optimized blue prints from being constructed in the real world. To circumvent this problem, we propose removing the virtual aspect from object design optimization and evolve objects solely in the physical world. To accomplish this, we introduce Evofabber, a fully embodied evolutionary fabricator, which evolves objects instead of virtual designs.
Advisor: Prof. Striegnitz
Digital film is not a new concept for us, since we have been introduced to 3D animations and special effects in movies such as Wall-E and Star Trek. Digital film allows us to achieve an old goal of the filming industry by making a surreal world a real world. This research project stands to present a perspective, philosophy and visual rendition which will help to artistically argue my concepts. Finally, I will make a particle system and combine it with personally recorded footage to create a short film about human loneliness.