Stevan White

Achievements at CTF Systems

I am proud of my work on several projects at CTF Systems. Here they are, arranged in roughly anti-chronological order.

Neural sequencing software

My final assignment put me in charge of neural stimulus sequencing software. I researched, designed and wrote the code.

The purpose of the software was to produce audio, visual, and other stimuli in very precisely timed sequences while informing a data collection computer of when these stimuli occur.

Two problems were especially difficult: One was how to accommodate an open-ended definition of a “sequence of stimuli”; the other was how to manage large amounts of data resources that, for performance reasons, have to be in memory before the stimuli occur.

I solved the first problem by dividing the package into two parts, a sequence player written in C++ and a set of simple sequence generating programs written in Java.

I overcame the second problem by writing code that predicts when resources will be needed for the stimuli, and loads them just before the stimuli occur.

Internet connection

In the Summer of ‘96, CTF had only a dial-up phone connection to an Internet provider. This situation had become very expensive and inconvenient for the company.

I proposed that we arrange a direct connection with the phone company, and become a proper Internet site. I was made part of a team formed to put the plan into effect. In addition to installing and configuring a Sun workstation as a web and ftp server, I was the primary liaison with the phone company.

Our team discovered a bug in the phone company’s software that reduced throughput by 50%, in addition to other problems that rendered our ftp service useless. I collected data for the phone company and hounded them until they recognized and fixed the problems. In effect, our team debugged the phone company’s system.

The result is that CTF now enjoys the immediate communication of the Internet, unified local and global e-mail, a web site under their immediate control, and a fast means of exchanging large amounts of data with clients.

Research

I have been the source of aid and information on mathematical physics and numerical problems. Some recent examples are: implementation of second-order nonlinear optimization methods, solution of underdetermined Least-Squares methods, and inverse methods including dipole modelling.

I researched and presented talks on several methods for imaging brain activity. I coded one, Mosher’s method, to distinguish two electric dipoles that vary in a statistically uncorrelated way.

I also did research on data noise analysis and filtering, including Fourier, Wavelet, FIR and IIR filters. Most importantly, I invented a new spatio-temporal method for filtering neuromagnetic data.

Object-oriented design and programming

In addition to the sequencing software, I wrote a large package implementing Linear Algebra in C++, and a program that did filtering and imaging for 3-D tracking data. I designed a package to efficiently manage and process huge amounts of data (on the order of gigabytes) in a very flexible way.