Message Boards Message Boards

0
|
8752 Views
|
2 Replies
|
1 Total Likes
View groups...
Share
Share this post:

FindFit question - find multiple solutions in one run?

Posted 11 years ago
Dear all,

is there a way that FindFit gives all the possible values for the specified parameters?

For example, in the file at the url http://www.risc.jku.at/people/merascu/FindFit.nb, I should basically have 4 values for each of the parameters a and b (3 for each a and b determining the the curves, and 1 a and 1 b for the isolated point at the left most of the graphic).
If is not possible with FindFit, is there another way?

Thanks
POSTED BY: Madalina Erascu
2 Replies
In the ListPlot it looks like you have data points falling along several curves. What you really have is a flat list of data points and what looks like curves are not made up of sequential points, but are quite interlaced together. Have a look at the same ListPlot with the option Joined -> True
ListPlot[lstQ, ImageSize -> {1200, 339}, AspectRatio -> Full , Joined -> True, PlotRange -> All]
The data list can be split up this way into four separate runs of data. The resulting split is not perfect.
ListPlot[Transpose[Partition[lstQ, 4, 4]], ImageSize -> {1200, 339}, AspectRatio -> Full , Joined -> True, PlotRange -> All]
I think that the fitting functions will not be happy, till the data for a single curve can be isolated out. Then you can fit the three or four separately.
If you generate the equations describing the fitting parameters, you could try solving them with FindRoot, using multiple starting points.  Given the number of data points, I don't think that NSolve would be able to handle the problem.
POSTED BY: Frank Kampas
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract