# Colin K

less info
reputation
5
bio website location age 29 member for 2 years, 11 months seen Oct 9 at 20:19 profile views 19

I've got a BS in Physics from the University of Rochester, and I'm over half-way through my masters in optical engineering. I work at the Laboratory for Laser Energetics, where I develop and maintain technology for the Omega EP laser.

http://en.wikipedia.org/wiki/Laboratory_for_Laser_Energetics

# 24 Actions

 Sep26 awarded Talkative Sep26 asked When an analytical Jacobian is available, is it better to approximate the Hessian by $J^TJ$, or by finite differences of the Jacobian? Jun1 comment From confidence interval to standard deviation - what am I missing? Yeah, this is great. May20 comment What is the difference between “priors” and “likelihood”? Awesome answer. May18 comment Why/When might it be preferable to fit to the integral of the data, rather than to the data itself? @whuber: Have a second to chat? May18 comment Why/When might it be preferable to fit to the integral of the data, rather than to the data itself? Just to clarify, since I wasn't explicit about it, I've got an analytical expression for the shape of the integral of the pulse, so the pulse height, Gaussian width, and exponential decay time can be pulled out of the fit and used in the expression for the shape of the pulse itself. May18 awarded Commentator May18 comment Why/When might it be preferable to fit to the integral of the data, rather than to the data itself? @whuber: To address your points: The noise in the raw data should have no temporal correlation afaik. Thus I suppose your (a) and (b) will hold for me. The integral is monotonically increasing (ignoring any violation of that due to noise), but the underlying response is definitely not monotonic. I looks, qualitatively, like a Gaussian. The exponential decaying component adds a slightly longer tail. Thus, the integral looks like an erf(). I'm doing my fitting by nonlinear least squares. Specifically, I'm minimizing the sum squared error between the data and my fit using fminsearch() in Matlab. May18 asked Why/When might it be preferable to fit to the integral of the data, rather than to the data itself? May1 comment What is probability to get through to the value You've asked a very unclear question, and you certainly are not telling us the whole story. This makes it very hard to provide an answer. You would get better results if you explain what actual process is producing this data set, and what problem you are trying to solve. Apr14 comment Estimating Frechet distribution parameter in R I get a general sense of unease and doom when people talk about choosing a distribution simply because it "best fits the data." Apr14 comment Goodness of fit for 2D histograms Wow. You answered the heck out of this thing. If there was a "best of stackexchange," this would be in it. Nov4 accepted What is a good reference for compound Poisson processes? Nov2 asked What is a good reference for compound Poisson processes? May17 comment Categorization/Segmentation techniques I've bought a copy of "Pattern Recognition" by D, H, and S. It really is spectacular, and the whole thing is useful and easy to follow. Thanks again. This is exactly how SE is supposed to work :) May10 awarded Scholar May10 awarded Supporter May10 accepted Categorization/Segmentation techniques May10 comment Categorization/Segmentation techniques Best. Answer. Ever. (bazinga!) May9 comment Categorization/Segmentation techniques @deps_stats: The only one of those I am familiar with is k-means, which I have used in the past for auto-thresholding images. In this case, looking at scatter plots of my data has convinced me that the data does not cluster into well defined populations. At least, not in a way that well reflects the human-defined classification. I've already left my office for the day, but I think scatter plots would be acceptable for me to share, soI'll make some edits tomorrow.