It is proposed to carry out an experimental and theoretical program of research related to the dynamical modelling, identification and feedback control of optical tweezers. By the proper use of feedback control and identification techniques, we plan to increase the accuracy and bandwidth of the optical tweezer as a sensor by orders of magnitude. The testbed application for these techniques is to use the tweezer to sense, in real time, mechanical or chemical changes to a biological molecule; for example, the mechanical ``unzipping'' of DNA. Current techniques for such sensing are limited by the effects of thermal noise (Langevin forces) and the internal dynamics of the molecule itself. Techniques are proposed that incorporate dynamical models of the molecule as well as recursive system identification algorithms to estimate the parameters of trap-molecule interaction in the presence of significant background noise. The proposed research program will consist of developing appropriate models for molecule-trap interaction, feedback control and identification, and experimental testing of the algorithms on the existing and operational optical tweezer instrument in the PIs laboratory at UCSB.
