Help with motoglobalfit

Share icon

Fri, 09/23/2011 - 01:51 am

I have used motofit successfully in the past, but it seems to be giving me a strange error now that I have reinstalled everything:

Function Execution Error

While executing GenCurveFit, the following error occurred: Non-Existent Wave.

I am using a fresh copy of Igor Pro 6.22A and motofit 3.2 (latest package from source forge) on an Intel mac running OS X Lion 10.7.1.

The console reads:

*** Doing Global fit ***
Global Fit converged normally.

Global fit results
Fit converged normally
V_chisq = 0 V_npnts= 142 V_numNaNs= 0 V_numINFs= 0
Number of iterations: 8

... information about various constants.

The Levenberg option works fine, but the genetic algorithm is giving me this error.

Any help would be appreciated!
Could you post a packed experiment file so that I can see what is going wrong?
The latest beta version of Motofit (are you fitting reflectometry data) can be found at:
https://motofit.svn.sourceforge.net/svnroot/motofit/branches/dev/instal…
or
https://motofit.svn.sourceforge.net/svnroot/motofit/branches/dev/instal…

this contains the latest version of the MOTOFIT_Global fit 2.ipf file. Alternatively you can get it from the gencurvefit project on IGORexchange. In fact, can you check if you have the latest version of Gencurvefit.xop installed (from IGORexchange).

I think you may be pressing cancel when you've just said the limits, but I can say for sure if you send the expt file.

September 23, 2011 at 02:37 am - Permalink

I can share the file with you, but it works fine on an older version of motofit (still called 3.2). I will try the latest beta and see what happens. Thanks for your reply!

EDIT: I am fitting impedance data.

September 28, 2011 at 05:42 am - Permalink

I have solved the issue by reverting to an earlier version of motofit. I'm still not sure what the problem is.

September 28, 2011 at 08:56 am - Permalink

I have attached an example experiment. It is much simpler than the fit I want to do, but it fails in the same way. Note that this experiment works just fine with an older version of motofit. I am currently using the latest version of motofit, igor 6.22A and mac os X 10.7.1.

EDIT: motofit ->about shows rev. 409.
example.pxp.zip

September 29, 2011 at 08:40 am - Permalink

It should work now. Please use the version from the Gencurvefit IGORexchange site.

cheers,
Andrew.

September 29, 2011 at 06:02 pm - Permalink

The installer for gencurvefit on igorexchange isn't working. I get the following error:
Igor not installed? GenCurvefit will only install in ~/Documents/Wavemetrics/Igor Pro 6 User Files/Igor Extensions

The reason for this is the line in distribution.dist that checks for a file in the following location:

'/Users/andrew/Documents/WaveMetrics/Igor Pro 6 User Files/About Igor Pro 6 User Files.txt'

modifying that line to match my system allowed me to install.

Upon launching Igor, I get several errors from compiling MOTOFIT_Global Fit 2.ipf
expected comma Error in MOTOFIT_Global Fit 2.ipf:MOTO_DoNewGlobalFit GEN_setlimitsforGENcurvefit(root:packages:MotofitGF:NewGlobalFit:MasterCoefs, newGF_HoldString)

I have attached a third screenshot of my Igor Pro User Files directory. Perhaps something is out of place?

Thanks again!
Installer.png Function Compilation Error.png Finder.png

October 1, 2011 at 12:49 am - Permalink

curses.
I've checked the installer and it works now (home vs work computer). To use the code you have to:
#include "geneticoptimisation"
#include "MOTOFIT_Global Fit 2"

-or-
put the files in IGOR procedures instead.

October 2, 2011 at 03:11 am - Permalink

Andy, thank you so much! Everything seems to be working great with the latest installer package (October 2). I am occasionally getting the NaN or INF errors, but I think that is an issue with my function and limits. I'd also like to add that logarithmic spacing is working perfectly now, which was a problem for me in the past. Three cheers!

October 2, 2011 at 07:55 am - Permalink

I have one more issue... it always quits after 200 iterations, and will not go further. In the past, I was able to let it chug along for thousands of iterations without interruption. Am I missing something? The "Max Iterations" field seems to have no effect on this.

EDIT: It seems the number of iterations is hard-coded. I don't think this was the case in previous versions. I can change the value in motofit_Global Fit 2 to a larger value and it allows me to increase the iterations.

October 8, 2011 at 04:52 pm - Permalink

Fellows, thank you for this thread. :-) I am reasonably new to Igor: half a year, and probably serious about using it as my new does-everything data platform. I'm just about to try Motofit as a tool to assist my synchrotron beamline users. I also encountered the "expected comma" compile error from the first file I downloaded, though the link in this thread seemed to lead to a file that compiled.

Up to now I've been writing C-like code in the Spec/Cplot fitting package to analyze data. Looking forward to investigating Motofit.

October 10, 2011 at 09:46 am - Permalink

ElaineDiMasi wrote:
Fellows, thank you for this thread. :-) I am reasonably new to Igor: half a year, and probably serious about using it as my new does-everything data platform. I'm just about to try Motofit as a tool to assist my synchrotron beamline users. I also encountered the "expected comma" compile error from the first file I downloaded, though the link in this thread seemed to lead to a file that compiled.

Up to now I've been writing C-like code in the Spec/Cplot fitting package to analyze data. Looking forward to investigating Motofit.


Dear posters,
I am the author of the motofit package and appreciate all the feedback. Based on Elaines post I discovered that the current stable installer of Motofit (from the sourceforge download site) wasn't working. Thank you so much for the feedback, I don't know how long it was like that for - the horror. Can I just say that I use the package extensively myself to analyse data and need the thing to work, and it is a priority for me that everything works well. Compile errors are not acceptable - please contact me if this problems occur.

Anyway, the code from the download page is going to be replaced with a version currently in development - available from https://motofit.svn.sourceforge.net/svnroot/motofit/branches/dev/instal…

This has a _lot_ of improvements and I strongly suggest that people use this version (although it cannot do repeating multilayers at the moment).

October 10, 2011 at 01:55 pm - Permalink

andyfaff wrote:
... version currently in development - available from https://motofit.svn.sourceforge.net/svnroot/motofit/branches/dev/instal…

This has a _lot_ of improvements and I strongly suggest that people use this version (although it cannot do repeating multilayers at the moment).


Sounds good - I unloaded the existing Motofit package, ran the motofitInstaller.exe from the above link (windows 7 on a lenovo x201), loaded the package, and have the Motofit menu to work with now.

If I have more questions I will be happy to ask for help, thanks,

Elaine

October 11, 2011 at 08:44 am - Permalink

Hi all,

I am still having issues using the development version. It seems to work for me on the mac, but on the PC, I am getting the "non existent wave" issue again. Any thoughts? I am using a fresh copy of 6.22A and the development version Andy suggested.

November 7, 2011 at 01:45 pm - Permalink

mtaylor wrote:
Hi all,

I am still having issues using the development version. It seems to work for me on the mac, but on the PC, I am getting the "non existent wave" issue again. Any thoughts? I am using a fresh copy of 6.22A and the development version Andy suggested.


The latest version is up to date. However, I ruined the installer program again (I have a version at home and at work). The problem is that when the wrong installer is used it puts the code into C:/Program Files/Wavemetrics/Igor Pro Folder/....
This is ok if you have XP, but when you have Win>XP then the OS performs some ugly tricks.
Please read the following link
http://www.wavemetrics.com/search/viewmlid.php?mid=21778

Try and remove GeneticOptimisation.ipf, MOTOFIT_Global fit 2.ipf, Gencurvefit.xop, etc from "C:\Users\\AppData\Local\VirtualStore". You may have to look around a bit.
You can tell if there is a problem in the code, or if you have an older version, if line 628 is NOT commented out.

The trouble is that IGOR will always loads the code in "C:\Users\\AppData\Local\VirtualStore" first before looking in $DOCUMENTS\Wavemetrics\Igor Pro 6 User Files\ .

November 7, 2011 at 05:53 pm - Permalink

Andy, thanks as always.

I got the latest version of gencurvefit working by realizing that there were conflicts in the user's wavemetrics folder with that of the program.

As you mentioned there are still some wonky things about the installer... I noticed that one of the installers, not sure which, on my mac installed the following in my home folder:

~/Users/anz/Documents/Andy/MOTOFIT/XOP\ Toolkit\ 5/IgorXOPs5/MultiDimensionalGenCurvefit/ents/Wavemetrics/Igor\ Pro\ 6\ User\ Files/User\ Procedures/motofit/ExampleExperiment.pxp

Yes, the folder is in my home folder... Quite an odd place to put that file.

Cheers!

November 9, 2011 at 05:42 am - Permalink