self portrait

Todd Mytkowicz

Graduate Student in Computer Science
B.S. Computer Science, Syracuse University '01

email:mytkowit @ colorado.edu
office:  ECCR B121
CV:  resume.pdf
Dynamical Systems & Performance Analysis

Computer Dynamics Lab


Research

Neural Net

Physics of Multicore Computation
Collaborators

Whether we like it or not, multicore computation is here to stay. Along with multicore comes the increased complexity of parallel computation. Our approach to this problem is a pragmatic one; we are building tools to accommodate the increased complexity in understanding where performance bottlenecks arise in multicore computing.

Neural Net

Chaos in computer systems
Collaborators

Computation on modern hardware is a complex system. This work uses traditional nonlinear time series analysis techniques to understand the dynamics of real hardware. Our efforts have shown the first experimental evidence of chaos in computer systems and as such provides a deeper understanding of the inner workings of modern hardware. We hope these techniques will allow us to collect high-fidelity data from a running system without perturbing the system's true behavior.

Neural Net

Dynamics of Social Collaborative Systems
Collaborators

This work was done while I was a summer intern at Xerox' PARC. We used information theory as a means to analyze the dynamics of social tagging systems. Our results show that over time, a social tagging site like del.icio.us is becoming harder and harder to navigate. The tagging environment is becoming saturated.

We generated a bit of press:


Publications

Conference Publications

Workshop Publications

Conference Abstracts, Presentations and Posters


Downloads


Fun

A pet project of mine has been to create artwork using dynamical systems as templates for creative, interesting pictures. My most recent attempts have been using the Lorenz Attractor and a raytracing tool named povray. Here is my first cut at this attractor. I will add more as time allows. Enjoy!