topic Re: Fortran library or subroutine needed for curve fitting in Intel® Fortran Compiler

Fortran library or subroutine needed for curve fitting

eriehml — Tue, 23 Jun 2009 23:51:14 GMT

Hello all.
I am physics grad. student and I am returning to coding after 15 years so have some patience with me. I have experimental data points, X and Y which need to be fitted to an exponential or forth order polynomial (no other order is necessary and I could use forth order polynomial only if it easier). The purpose of the fitting is to subtract background. These X and Y are photoemission data (Valence Band) to be precise. So it is only necesarry to fit the curve (exponential or forth order polynomial) to the firsts and lasts data points. After the fit is obtained it will correspond to the background. Then I can subtract at any X, Y the background just calculated. Online I found this NAG library but I can not download anything but pdf files. In the NAG library there is a subroutine called E02ADF which might help me do the fiiting but I can not download it, any body knows how access NAG downloads? Either this subroutine or any other you might know I can give it a try as long as I can download it. As far as the minimization algorithm I am really not picky, LM or Simplex will do for me. I am coding in fortran using G77 compiler, windows Vista, plataform Force 2 in my laptop. My code is really simple I just need to handle too many data to fit manually each one of them. I appreciate your time, everyone's. Thanks for your help.

Re: Fortran library or subroutine needed for curve fitting

DavidWhite — Wed, 24 Jun 2009 00:15:36 GMT

Quoting - eriehml

The NAG libraries are probably only available via license. You may want to try Netlib (www.netlib.org) for a range of Fortran and other sources for many statistical and other operations.

Re: Fortran library or subroutine needed for curve fitting

eriehml — Wed, 24 Jun 2009 00:57:13 GMT

Thanks. I already had that done and netllib.org libraries where no help, they have a very good search engine so looking for stuff you need is not so painful. Such elementary stuff like what I need I hope is not on license but if NAG has to sell me the subroutine I need then I will be glad to pay. Thanks for your time.

Re: Fortran library or subroutine needed for curve fitting

Peter — Wed, 24 Jun 2009 02:01:50 GMT

Quoting - eriehml

NAG libraries are very expensive, if you are just looking for curve fitting.Intel Visual Fortran's (IVF) MKL libraries might be of help. But you need to buy a copy of IVF.

Alternatively try excel 2007!

Re: Fortran library or subroutine needed for curve fitting

eos_pengwern — Wed, 24 Jun 2009 08:47:37 GMT

Quoting - Peter

NAG libraries are very expensive, if you are just looking for curve fitting.Intel Visual Fortran's (IVF) MKL libraries might be of help. But you need to buy a copy of IVF.
Alternatively try excel 2007!

Another very good source of Fortran source code for a host of scientific applications is Alan Miller's page. Whereas most of what's on NETLIB is in Fortran 77, Alan Miller's resources are generally in Fortran 90 so are a bit more user-friendly. This may not help you if you're using g77, however.

Re: Fortran library or subroutine needed for curve fitting

Jugoslav_Dujic — Wed, 24 Jun 2009 13:02:32 GMT

Quoting - eriehml

For polynomial fitting see:

http://rosettacode.org/wiki/Polynomial_Fitting#Fortran

LAPACK is part of Intel MKL, shipped with Intel Fortran Professional 11.0 onwards.

Re: Fortran library or subroutine needed for curve fitting

anthonyrichards — Wed, 24 Jun 2009 16:03:09 GMT

Quoting - Jugoslav Dujic

For polynomial fitting see:

http://rosettacode.org/wiki/Polynomial_Fitting#Fortran

LAPACK is part of Intel MKL, shipped with Intel Fortran Professional 11.0 onwards.

Or find a friend with access to MathCad and ask to borrow it. It is very good for curve fitting.

Re: Fortran library or subroutine needed for curve fitting

Greg_T_ — Wed, 24 Jun 2009 17:28:30 GMT

Hello,
Another source for Fortran routines is the NIST "Guide to Available Mathematical Software": http://gams.nist.gov/
I think there are a number of routines related to polynomials and curve fitting. The web site has a search tool and directory tree to find what you're looking for. I've used a few routines from that web site with good success.

Regards,
Greg

Re: Fortran library or subroutine needed for curve fitting

eriehml — Wed, 24 Jun 2009 21:44:21 GMT

Quoting - Peter

Peter
Thanks for your reply. When I read the "try excel 2007!" I thought, Oh man, people with good sense of humor at this forums, so thanks for that laugh. I had no idea about the Intel Visual Fortran you mentioned so I really appreciate you said it. I am looking more into it as we speak and found only some info at
http://software.intel.com/sites/products/collateral/hpc/compilers/fwin_brief.pdf
but maybe you can answer some of my questions. Is this a stand alone plataform/program so that I can code and compile?, do you know if the it also has some ploting libraries?, can this be installed in a laptop?, any idea of the price? I'll keep looking into it but I would appreciate your comments. I just found out the cost is $699, Oh my!, that is so totally out my budget unless I hit Powerball tonight.
Thanks.

Re: Fortran library or subroutine needed for curve fitting

eriehml — Wed, 24 Jun 2009 21:50:10 GMT

Quoting - greg_thorwald

Greg
Thanks a lot. I am looking into it.
Erie

Re: Fortran library or subroutine needed for curve fitting

anthonyrichards — Thu, 25 Jun 2009 08:42:00 GMT

Quoting - eriehml

Greg
Thanks a lot. I am looking into it.
Erie

I still think that if you want a quick solution with graphical display of results, MathCad is easily the quickest route. I frequently use MathCad 11 to test and debug my algorithms (and for quick display of intermediate results)before reprogramming them in Fortran for speed. You can buy MathCad 11 on E-bay for a pittance compared to getting a complete Fortran compiler + library of math functions. I attach a screen shot showing a sample fitting function that is available to a MathCad user.

Re: Fortran library or subroutine needed for curve fitting

eliosh — Thu, 25 Jun 2009 08:46:59 GMT

Quoting - Peter

Quoting - eriehml

Thanks. I already had that done and netllib.org libraries where no help, they have a very good search engine so looking for stuff you need is not so painful. Such elementary stuff like what I need I hope is not on license but if NAG has to sell me the subroutine I need then I will be glad to pay. Thanks for your time.

NAG libraries are very expensive, if you are just looking for curve fitting.Intel Visual Fortran's (IVF) MKL libraries might be of help. But you need to buy a copy of IVF.
Alternatively try excel 2007!

You can use a free non-commercial copy of the Intel compiler for Linux.
(In this case you do not have to pay for Excel 2007 either ;) )

Re: Fortran library or subroutine needed for curve fitting

Peter — Fri, 26 Jun 2009 16:30:40 GMT

Quoting - eriehml

Hmm...I wasn't sarcastic about excel 2007, but it made you laugh !.

Excel can not be underestimated, because it is very powerful tool for most engineering and scientific applications (not powerful than fortran and c++). But based on your requirement it looks like you want fortran subroutines for curve fitting. I did not understand exactly what your project is about, but if you get the data from fortran and then transfer to excel 2007, you might be able to curvefit.

Regarding IVF, its a fortran 90 /03 compiler, which you can code and compile !. For most of my data applications I don't rely on fortran plotting routines. I export the data to matlab or excel which is very easy and time-saving !!

Re: Fortran library or subroutine needed for curve fitting

Paul_Curtis — Fri, 26 Jun 2009 18:50:46 GMT

See the attached, what the OP actually asked for.

Re: Fortran library or subroutine needed for curve fitting

Paul_Curtis — Fri, 26 Jun 2009 19:06:20 GMT

My sample program was carved out of some real code, needs a minor correction in the output section, only apparent after uploading:

[cpp]DO j = n1, n2
    xx = x(j)
    yy = pxpand (norder, xx, a)
    WRITE (3,*) xx, y(j), yy, yy-y(j)
END DO
[/cpp]

Re: Fortran library or subroutine needed for curve fitting

eriehml — Sat, 27 Jun 2009 16:08:51 GMT

Quoting - Paul Curtis

See the attached, what the OP actually asked for.

Paul
Thanks for sending me the code. My plataform is Force 2.0 and reads only G77 or similar. I am going to download (hopefully I find it for free) a Fortran90 plataform so that I can check your code out. I have too many sets of data that I need to fit and the fit is only for background subtraction and just a part of the data massaging I have to perform. I really appreciate your help. I'll get to work right now and try to find a free fortran 90 compiler and editor to check your code out. Thanks a lot.

Re: Fortran library or subroutine needed for curve fitting

eriehml — Sat, 27 Jun 2009 20:45:37 GMT

Quoting - Paul Curtis

My sample program was carved out of some real code, needs a minor correction in the output section, only apparent after uploading:

[cpp]DO j = n1, n2
    xx = x(j)
    yy = pxpand (norder, xx, a)
    WRITE (3,*) xx, y(j), yy, yy-y(j)
END DO
[/cpp]

Paul
Thanks. I managed to download a F95 plataform called PLATO, from Silverfrost which seems to work well. I performed the modifications in the DO LOOP you mentioned. When I compile though I still have problems
in line 434

DO j = order, 2, -1
polyval = xv*(polyval + coeff(j))
END DO

I am not sure what the problem would be, do you?

For starters I am only using

INTEGER, PARAMETER :: nvals = 5 ! YOUR POINTCOUNT
REAL, DIMENSION(nvals) :: x = (/ 1.0, 2.0, 3.0, 4.0, 5.0 /) !(/ YOUR x DATA HERE /)
REAL, DIMENSION(nvals) :: y = (/ 1.0, 16.0, 80.9, 256.5, 625.0 /) !(/ YOUR y DATA HERE /)

! polynomial order for fitting
norder = 4 ! YOUR FIT ORDER

Please let me know. Thanks.

Re: Fortran library or subroutine needed for curve fitting

Paul_Curtis — Sat, 27 Jun 2009 22:14:55 GMT

This code compiles and runs perfectly under MS5.5, DVF, all CVF through 6.6, all IVF through v11. The svdfit routines are an F90 translation from the original F77 which I think may have come from Numerical Recipes or similar; very old (pre-dates PCs by decades), completely portable and with no vendor-specific features.

If your compiler can't handle this well-proven code and doesn't provide informative error messages, one solution might be to upgrade to a better compiler, maybe from a company that has invested in hosting a user forum for compiler support and language discussions.

FYI, here is the output from this program, using your data, as created by IVF (I have added some column headers):

Fit order is 4
1 -42.02048
2 79.64155
3 -48.84335
4 11.91672
x y polyfit diff.
1.000000 1.000000 0.6944389 -0.3055611
2.000000 16.00000 17.22299 1.222988
3.000000 80.90000 79.06549 -1.834511
4.000000 256.5000 257.7222 1.222229
5.000000 625.0000 624.6935 -0.3065186

Re: Fortran library or subroutine needed for curve fitting

mavlik — Sat, 27 Jun 2009 22:59:32 GMT

Quoting - eriehml

<....> In the NAG library there is a subroutine called E02ADF which might help me do the fiiting but I can not download it, any body knows how access NAG downloads?

As I understand, E02ADF is least-squares polynomial approximation.

Try to use this link
http://iranyahoo.nag.com/numeric/fl/manual20/examples/source/e02adfe.f

If you need polynomial interpolation or other algorithms you can find it in "Numerical Recipes in Fortran"
(http://www.nrbook.com/a/)

Re: Fortran library or subroutine needed for curve fitting

eriehml — Sun, 28 Jun 2009 22:53:14 GMT

Quoting - Paul Curtis

Paul
Thanks. I see it looks like IVF is necessary, I am stuck on it though because it costs $699 and that is not an option at the moment. Gosh, I didn't think it was going to be hard to find a routine for polynomial fitting using any algorithm. I appreciate your patience and time. Enjoy.