<!doctype html public "-//W3C//DTD W3 HTML//EN">
<html><head><style type="text/css"><!--
blockquote, dl, ul, ol, li { padding-top: 0 ; padding-bottom: 0 }
--></style><title>Re: [proFit-list] Pro Fit and
Applescript</title></head><body>
<div>That's correct. The LM method needs to take derivatives and
invert matrices, things that can be sensitive to initial conditions.
The "Robust" methods tries to follow a landscape to find a
minimum by sensing the environment.</div>
<div><br></div>
<div>This offers choice and control: If you get singularities in LM it
is because your initial guesses for the parameters are too far away,
or because two parameters have similar effects on the function (or are
redundant).</div>
<div><br></div>
<div>In such a case, you can change the initial guess by hand, or you
can use the Robust algorithm to improve the guess.</div>
<div><br></div>
<div>The Robust algorithm will quickly deliver a better fit which can
then be used as a starting point for the LM algorithm (LM is
faster).</div>
<div><br></div>
<div>Most often than not, the Robust algorithm may even already take
you directly to the best fit.</div>
<div><br></div>
<div>In any case, run the LM algorithm at the end, so that you can get
more statistical information on the fit (things like parameter errors,
etc.)</div>
<div><br></div>
<div>ivan</div>
<div><br></div>
<blockquote type="cite" cite>Usually, when the LM algorithm gives a
singularity, the robust method works. As far as I understand it, the
robust method is for that purpose.</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>Cheers</blockquote>
<blockquote type="cite" cite>Chris<br>
</blockquote>
<blockquote type="cite" cite>On Aug 22, 2009, at 10:20 AM, Nic Meyer
wrote:</blockquote>
<blockquote type="cite" cite><br>
<blockquote type="cite" cite>thanks to everyone for their
replies.</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>I have been trying the recording option
and it sure helps.</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>that said, I am running into problems
of singularities when fitting my curves (which IGOR handles
without complaint) so that is a new hurdle to overcome (in particular
i am trying to fit lognormal curves and am wondering whether I should
specify my own function for this)</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>given that i do need results soon i will
take them from IGOR, but as a Mac user i am happy that their could
well be a possibility where everything could be scripted.</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>cheers</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>Nic</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>
<hr size="1"></blockquote>
<blockquote type="cite" cite><font face="Tahoma"
size="-1"><b>From:</b> pro Fit Support <</font><a
href="mailto:profit@quansoft.com"><font face="Tahoma"
size="-1">profit@quansoft.com</font></a><font face="Tahoma"
size="-1">><br>
<b>To:</b> all things about pro Fit <</font><a
href="mailto:profit-list@quantum-soft.com"><font face="Tahoma"
size="-1">profit-list@quantum-soft.com</font></a><font face="Tahoma"
size="-1">><br>
<b>Sent:</b> Friday, August 21, 2009 5:37:44 PM<br>
<b>Subject:</b> Re: [proFit-list] Pro Fit and Applescript<br>
</font><br>
In addition to the comments provided by Chris and Davide:</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>One way to learn about controlling pro
Fit through AppleScript or a Pascal script is by exploiting the fact
that pro Fit is "recordable": When you hit the "record"
button in AppleScript Editor or one of pro Fit's function windows, and
then execute some commands manually within pro Fit (such as fitting a
function, or binning data), most of these operations will be recorded
as Apple Script or Pascal statements for you. This usually gives a
good starting point for writing a script. Give it a try.<br>
</blockquote>
<blockquote type="cite" cite><br>
Best regards</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>Kurt Sutter</blockquote>
<blockquote type="cite" cite>QuantumSoft</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>On 21 Aug 2009, at 7:43, Nic Meyer
wrote:</blockquote>
<blockquote type="cite" cite><br>
<blockquote type="cite" cite>Good morning,</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>I found Pro Fit yesterday and am really
impressed that this useful tool could be completely controlled via
applescript. (which I am already using extensively). I am trialling
Profit for scientific data analysis and curve fitting and the program
compares well with equivalent packages but with the added bonus of
(apparently) being fully scriptable</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>That said, very few examples exist
showing the manner in which one can script ProFit. In particular I am
looking to run Levenburg fits and would also like to test the bin data
functionality. I would also like to be able to perform calculations on
the fit curves which are generated.</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>I have access to the Pro Fit dictionary
via Text Editor but after an hour or so trial an error guessing what
information the script required with no success I am hoping to gather
the experiences of others.</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>If any one could send me some example
apple scripts or point me in the right direction i would be most
appreciative</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>(on another note, having had a snoop
through the post history, I am impressed that Python functionality
will be included in the next version. Looking forward to checking this
out!)</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>cheers</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>Nic</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>from Oz</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite>in Zurich</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br>
_______________________________________________<br>
proFit-list mailing list<br>
<a
href="mailto:proFit-list@quantum-soft.com"
>proFit-list@quantum-soft.com</a></blockquote>
<blockquote type="cite" cite><a
href=
"http://quantum-soft.com/mailman/listinfo/profit-list_quantum-soft.com"
>http://quantum-soft.com/mailman/listinfo/profit-list_quantum-soft.com</a
><br>
</blockquote>
</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br>
_______________________________________________<br>
proFit-list mailing list<br>
<a
href="mailto:proFit-list@quantum-soft.com"
>proFit-list@quantum-soft.com</a><br>
<a
href=
"http://quantum-soft.com/mailman/listinfo/profit-list_quantum-soft.com"
>http://quantum-soft.com/mailman/listinfo/profit-list_quantum-soft.com</a
><br>
</blockquote>
</blockquote>
<blockquote type="cite" cite><br></blockquote>
<blockquote type="cite" cite><br>
_______________________________________________<br>
proFit-list mailing list<br>
proFit-list@quantum-soft.com<br>
http://quantum-soft.com/mailman/listinfo/profit-list_quantum-soft.co<span
></span>m</blockquote>
<div><br></div>
<div><br></div>
<x-sigsep><pre>--
</pre></x-sigsep>
<div
>--------------------------------------------------------------------<span
></span>-----<br>
QuantumSoft <span
></span
> <span
></span
> <span
></span>
profit@quansoft.com<br>
<span
></span
> <span
></span
> <span
></span
> <span
></span> http://www.quansoft.com/<br>
</div>
</body>
</html>