Re: 2D Gaussian fit
- To: mathgroup at smc.vnet.net
- Subject: [mg126316] Re: 2D Gaussian fit
- From: Cisco Lane <travlorf at yahoo.com>
- Date: Mon, 30 Apr 2012 04:44:04 -0400 (EDT)
- Delivered-to: email@example.com
You might consider a polynomial fit to the logarithm of your data to the logarithm of the Gaussian, (which is a polynomial). A direct fit of the data to a Gaussian will minimize the square of the differences between the data and the Gaussian. This will tend to give large relative errors in the wings, and small relative errors in the spot. A logarithmic fit will minimize the relative errors, rather than the absolute errors. The "percent error" will be roughly the same over the entire field. I don't know if this is acceptable to you or not.