Mac binaries

[jabaws.git] / website / archive / binaries / mac / src / disembl / Tisean_3.0.1 / source_f / predict.f

diff --git a/website/archive/binaries/mac/src/disembl/Tisean_3.0.1/source_f/predict.f b/website/archive/binaries/mac/src/disembl/Tisean_3.0.1/source_f/predict.f
new file mode 100644 (file)
index 0000000..0e8d161
--- /dev/null
+++ b/website/archive/binaries/mac/src/disembl/Tisean_3.0.1/source_f/predict.f
@@ -0,0 +1,104 @@
+c===========================================================================
+c
+c   This file is part of TISEAN
+c 
+c   Copyright (c) 1998-2007 Rainer Hegger, Holger Kantz, Thomas Schreiber
+c 
+c   TISEAN is free software; you can redistribute it and/or modify
+c   it under the terms of the GNU General Public License as published by
+c   the Free Software Foundation; either version 2 of the License, or
+c   (at your option) any later version.
+c
+c   TISEAN is distributed in the hope that it will be useful,
+c   but WITHOUT ANY WARRANTY; without even the implied warranty of
+c   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+c   GNU General Public License for more details.
+c
+c   You should have received a copy of the GNU General Public License
+c   along with TISEAN; if not, write to the Free Software
+c   Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+c
+c===========================================================================
+c   simple nonlinear prediction, fast neighbour search
+c   see  H. Kantz, T. Schreiber, Nonlinear Time Series Analysis, Cambridge
+c      University Press (1997,2004)
+c   author T. Schreiber (1998)
+c===========================================================================
+      parameter(nx=1000000)
+      dimension x(nx), y(nx)
+      character*72 file, fout
+      data eps/0./, frac/0./, ifc/1/
+      data iverb/1/
+
+      call whatido("prediction with locally constant fits",iverb)
+      id=imust("d")
+      m=imust("m")
+      eps=fcan("r",eps)
+      frac=fcan("v",frac)
+      ifc=ican("s",ifc)
+      nmaxx=ican("l",nx)
+      nexcl=ican("x",0)
+      jcol=ican("c",0)
+      isout=igetout(fout,iverb)
+      if(eps.eq.0.and.frac.eq.0.) call usage()
+
+      do 10 ifi=1,nstrings()
+         call nthstring(ifi,file)
+         nmax=nmaxx
+         call readfile(nmax,x,nexcl,jcol,file,iverb)
+         if(file.eq."-") file="stdin"
+         if(isout.eq.1) call addsuff(fout,file,"_pred")
+         call rms(nmax,x,sc,sd)
+         if(frac.gt.0) eps=sd*frac
+         iun=istdout()
+         if(fout.eq." ") iun=istderr()
+         write(iun,*) "err: ", fcerror(nmax,x,y,m,id,ifc,eps), 
+     .      " "//file(1:index(file," ")-1)
+ 10      call writefile(nmax,y,fout,iverb)
+      end
+
+      subroutine usage()
+c usage message
+
+      call whatineed(
+     .   "-d# -m# [-r# | -v#]"//
+     .   " [-s# -o outfile -l# -x# -c# -V# -h] file(s)")
+      call ptext("either -r or -v must be present")
+      call popt("d","delay")
+      call popt("m","embedding dimension")
+      call popt("r","absolute radius of neighbourhoods")
+      call popt("v","same as fraction of standard deviation")
+      call popt("s","time steps ahead forecast (one step)")
+      call popt("l","number of values to be read (all)")
+      call popt("x","number of values to be skipped (0)")
+      call popt("c","column to be read (1 or file,#)")
+      call pout("file_pred")
+      call pall()
+      stop
+      end
+
+      function fcerror(nmax,y,yp,m,id,ifc,eps)
+      parameter(im=100,ii=100000000,nx=1000000) 
+      dimension y(nmax),yp(nx),jh(0:im*im),jpntr(nx),nlist(nx)
+
+      if(nmax.gt.nx) stop "fcerror: make nx larger."
+      call base(nmax-ifc,y,id,m,jh,jpntr,eps)
+      fcerror=0
+
+      call rms(nmax,y,sx,sd)
+      do 10 n=1,(m-1)*id+ifc
+ 10      yp(n)=sx
+      do 20 n=(m-1)*id+1,nmax-ifc           
+         call neigh(nmax,y,y,n,nmax,id,m,jh,jpntr,eps,nlist,nfound)
+         av=0
+         do 30 nn=1,nfound            
+ 30         if(nlist(nn).ne.n) av=av+y(nlist(nn)+ifc) 
+         if(nfound.gt.1) then
+            yp(n+ifc)=av/(nfound-1)
+         else
+            yp(n+ifc)=sx
+         endif
+ 20      fcerror=fcerror+(y(n+ifc)-yp(n+ifc))**2
+      fcerror=sqrt(fcerror/(nmax-ifc-(m-1)*id))
+      end
+