Research Diary

For the longest time I am waiting for someone to create an USB key format network adapter that has builtin firewall, antivirus and VPN software and is activated by fingerprint authentication. For corporations this kind of device would help a lot with edge control and also help preventing outbreaks when non-corporate devices are attached to [...]

Okay everybody is talking about Biofuels. All big companies are investing in biofuels. Some do it to look nice and use it as a PR stunt (most oil companies), knowing that the current attempts will not even make a dent in their market share or profits, some others do it because it might be profitable [...]

Tidbits from behavioral economics: What counts as a gain or a loss is relative to prior prices, not considered as independent of an investor’s attention. A stock acquired at $50 and and acheives a year-end value of $100. A few months into the new year it is sold at $75. While this is objectively a [...]

For a long time I was thinking about identifying models in multiscale domain rather than converting time domain models into multiscale domain, but how can this be done? There are certain requirements:

* Model should be consistant accross scales
* Models at each scale should only contain relevant information for that scale
* Statistically models should be sound.
* Scales should represent either special or temporal resolutions.

Current multiscale modeling trends are having totally different model at each scale and matching the model at the interfaces by some boundry conditions. This would not describe real work problems as well since models are continous in real time.

Okay this has nothing to do with Biology or Genetics. Perhaps on the fringes.Loopshaping is very powerful technique in linear controller design in frequency domain. Open or Closed loop transfer functions are subjected to a filter or controller that only allows certain frequencies to exit the system. Effectively eleminating  or better supressing unwanted frequencies. This [...]

An interesting site and associated blog on the less known and somehow lost work of Alan Turing. Especially interesting is the section on Turing’s work on mathematical biology. More can be found on the site. Here is the link.

I have more than 2 machines that I use to surf the web. On all of these machines I am running Foldershare to sync documents. I use Firefox on all of these machines and I use only one machine at a time. All of the machines have very similar bookmark structure and I want them to be the same. I want the changes to propagate.

More details will follow but the general idea is to drop data points in a random (Monte Carlo) way to detect outliers and robustness of the estimation. Instead of random drops one can compartmentelize data using tree structure to identify local mean and variation changes.