octave-dev — GNU Octave developers' list

[OctDev] 3D visiualization and vrml tools w/ mini-howto

[Etienne G. <et...@is...>]

  Hello,

I'm working again on 3D visualization tools. Most conspicuous are
the functions :

  vmesh (x,y,z) : Somewhat like mesh(), but uses a VRML browser for
                  display. Allows to examine the function and save
                  snapshots.

  
  sel  = select_3D_points (x) : View a set of 3D points and select
                  some of them by using the mouse.

There are functions for doing lines, polyhedra and various other 3D
objects.

  This is derived from the vrml toolbox on my homepage + some
improvements and a mini-howto that I'm writing.

  Anyone interested to have this, e.g. in main/vrml or
main/3Dvisualization or wherever?

  Cheers,

  Etienne

PS : Requires FreeWRL (http://www.crc.ca/FreeWRL) to do the
     visualization. There's at least a Debian package of it.

-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] formula2mat : fill a matrix with values of formula

[Etienne G. <et...@is...>]

  Hello,

what about a function like this (in main/miscellaneous)? For people
too lazy to call linspace() and meshgrid().

  Cheers,

  Etienne

======================================================================
## z = formula2mat (f, ...)    - Fill matrix 'z' with values of formula 'f'
##
## INPUT ------
## f    :string: A formula written in octave code.
##
## OPTIONS ----
## bndx : 2x1 : Boundary  of x in which f is evaluated (overrides bnd)
## bndy : 2x1 : Boundary  of y in which f is evaluated (overrides bnd)
## bnd  : 2x2 : bnd = [bndx; bndy] Boundary in which f is evaluated
##                                                     Default = [-1,1;-1,1]
##                                                             
## sz   : 2x1 : Number of x and y values at which f is evaluated
##                                                     Default =   [11,11]
## 
## x    : vec : Values of x at which f is evaluated (overrides bndX and szX)
## y    : vec : Values of y at which f is evaluated (overrides bndY and szY)
##
## vec  : bool: If true, assume that z can be obtained by evaluating f when
##              x and y are matrices, rather than scalars.
## 
## OUTPUT -----
## z    : sz-sized matrix containing the values of f.
##
## EXAMPLES ---
##                                ## Fills a matrix z containing the values
##                                ## "x^2 - y^2" for x and y varying in
##                                ## [-1,1] by intervals of 0.2.
## z = formula2mat ("x^2 - y^2");
## 
##                                ## Same thing, quicker.
## z = formula2mat ("x.^2 + y.^2","vec",1);
##
function zz = formula2mat (f, ...)

op1 = " x y bndx bndy bnd sz szx szy vec ";

default = struct ("x", nan,\
		  "y", nan,\
		  "bnd",[-1,1;-1,1],\
		  "bndx",nan,\
		  "bndy",nan,\
		  "sz",[11,11],\
		  "szx",nan,\
		  "szy",nan,\
		  "vec",0);

if nargin > 1,
  opts = read_options (list (all_va_args), "default", default, \
		       "op1", op1);
else opts = default;
end

[x,y,bnd, bndx, bndy, sz, szx, szy, vec] = \
    getfield (opts, "x","y","bnd" ,"bndx", "bndy", "sz", "szx", "szy", "vec");

## szx overwrites sz (same for szy)
if ! isnan (szx), sz(1) = szx; end
if ! isnan (szy), sz(2) = szy; end

if length (sz) != 2
  error ("length of sz should be 2, not %i\n", length (sz));
end
bnd
## bndx overwrites bnd
if ! isnan (bndx(1)), bnd(1,1) = bndx(1); end
if length (bndx)>1 && ! isnan (bndx(2)), bnd(1,2) = bndy(2); end
if ! isnan (bndy(1)), bnd(2,1) = bndy(1); end
if length (bndy)>1 && ! isnan (bndy(2)), bnd(2,2) = bndy(2); end
bnd
if any (size (bnd) != [2,2])
  error ("size of bnd should be [2,2], not [%i,%i]\n", size (bnd));
end

## If x or y are not def'd, define them now
if isnan (x), x = linspace (bnd(1,1),bnd(1,2),sz(1)); end
if isnan (y), y = linspace (bnd(2,1),bnd(2,2),sz(2)); end


[xx,yy] = meshgrid (x,y);


## x and y will become scalar variables below


if ! vec
  zz = zeros (sz);
  eval (sprintf (["for ii = 1:sz(1),\n",\
		  "  for jj = 1:sz(2),\n",\
		  "    x = xx(ii,jj); y = yy(ii,jj);\n",\
		  "    zz(ii,jj) = %s;\n",\
		  "  end;\n",\
		  "end;\n"],\
		 f));
else
  x = xx;
  y = yy;
  eval (["zz = ",f,";"]);
end

-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] new pchip.m

[<Kai...@t-...>]

Hello,

I have commited a new pchip.m function (main/splines). It uses the 
SLATEC/PCHIP function dpchim.f to calculate the derivatives. I did some 
testing and it should work correctly now. So, if you do 1-d interpolation 
give it a test. Suggestions and bug reports are welcome

Bye Kai

[OctDev] Re: small change to leasqr.m

[Etienne G. <et...@is...>]

  Hello,

alternatively, you could do as in the other optimization functions and
allow 'pin' (or 'x') in

    leasqr(x,y,pin,F,{stol,niter,wt,dp,dFdp,options})

to be a list : pin = list (p1,p2,...). 'F' would be called with x, p1,
p2 ... as arguments and p1 is the adjusted argument. I don't know how
this fits w/ matlab's syntax. I guess a patch that doesn't break
anything would be accepted.

  Cheers,

  Etienne

On Wed, Aug 21, 2002 at 09:12:13AM -0600, E. Joshua Rigler wrote:
# I want to recommend, and perhaps eventually submit a small change to the
# leasqr.m function in octave-forge that should be completely transparent
# and backward compatible.  Feedback on my idea, and instructions for
# submitting this change would be appreciated.
# 
# In the eval calls to the function "F" (i.e. the function who's
# parameters are being estimated), and "dFdp" (either the numerical, or
# user-written analytic derivative function), I suggest that a structure
# is passed that contains miscellaneous information that is neither an
# input or output vector, nor one of the parameters.  It will be the job
# of the user-written function to use the information in that structure
# appropriately.
# 
# This, of course, will require an additional input to leasqr.m, but if
# the structure is placed as the last input variable, no previously
# written code should be affected.  It also seems to me that this should
# be compatible with Matlab, which seems to have been another goal of the
# authors of leasqr.m.
# 
# My only option now (aside from changing my local copy of leasqr.m) is to
# set a global variable, or variables, that define the options I want, but
# I always feel a little tainted using global variables (a consequence of
# working on too much bad FORTRAN code I suppose).  If someone has a
# better option, please let me know.  Thanks.
# 
# -EJR


-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] inline.m goes to main/miscellaneous ?

[Etienne G. <et...@is...>]

  or somewhere else?

  Etienne

-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] Re: inline functions (Corrected source)

[Etienne G. <et...@is...>]

  Hello,

On Sat, Aug 10, 2002 at 04:04:24PM -0400, Paul Kienzle wrote:
# Sure.  Add it to octave forge.  I simplified the while loop a bit, and let

  Good, but "if ! exist (fname), break; endif" is better.

# you use a default parameter of "x".  I wouldn't go much beyond this in
# terms of compatibility --- too many assumptions required about what makes a
# variable and what order the variables should go in. Maybe inline(expr,n) as
# shorthand for inline(expr,"x","P1","P2",...,"Pn") is useful.

  Or "x1", "x2", ... ? Does Matlab do such a thing? I would keep it as
it is. I'll check in your version w/ "x" as default but won't do more
refinements.

  Cheers,

  Etienne

# Paul Kienzle
# pki...@us...
# 
# On Sat, Aug 10, 2002 at 06:37:46PM +0100, Etienne Grossmann wrote:
# > 
# >   Sorry, there was a part missing from the first attached func. Try
# > the following.
# > 
# >   Etienne
# > 
# > On Sat, Aug 10, 2002 at 01:32:42PM +0000, Steffan Berridge wrote:
# > # 
# > # Hi,
# > # 
# > # I have just started using Octave, having used Matlab for a
# > # while.  In Matlab I have often used inline functions, they
# > # are very convenient for example when passing a function as
# > # an argument.
# > # 
# > # For example, the following is a function returning the
# > # square of the argument:
# > # 
# > # fun = inline('x^2','x');
# > # 
# > # One can then pass this function as an argument to another
# > # function:
# > # 
# > # x = foo(fun,...);
# > # 
# > # So my question is, is there an equivalent construction in
# > # Octave?  If not, what is the best way to pass functions as
# > # arguments?
# > # 
# > # Thanks,
# > # Steffan
# > -- 
# > Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne
# 
# ## [fname,fcode] = inline (str,arg1,...) - Define a function from a string
# ##
# ## INPUT : -----------
# ## str   : string    : String defining the result of the function
# ## argN  : string(s) : Names of the arguments of the function
# ##
# ## OUTPUT : ----------
# ## fname : string    : Name of the new function, which can e.g. be called w/
# ##                     feval() 
# ## fcode : string    : The code of the function
# ##
# ## EXAMPLE : ---------
# ##
# ## fn = inline ("x.^2 + 1","x");
# ## feval (fn, 6)        
# ##   ans = 36
# ##
# function [fname,fcode] = inline (str,varargin)
# 
# ## Choose a name (naive way : won't work zillions of times)
# while 1
#   fname = sprintf ("inline_func_%06i",floor (rand()*1e6));
#   if exist (fname), break; end
# end
# 
# ## construct arg list
# if nargin == 1
#   argstr = "x";
# else
#   argstr = sprintf ("%s,",all_va_args);
#   argstr = argstr(1:length(argstr)-1);
# end
# 
# fcode = sprintf (["function r = %s (%s)\n",\
# 		  "  r = ",str,";\n",\
# 		  "endfunction;"],\
# 		 fname, argstr);
# eval (fcode);
# endfunction		  
# 
# 
# 

-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] Re: inline functions (Corrected source)

[Paul K. <pki...@ja...>]

Sure.  Add it to octave forge.  I simplified the while loop a bit, and let
you use a default parameter of "x".  I wouldn't go much beyond this in
terms of compatibility --- too many assumptions required about what makes a
variable and what order the variables should go in. Maybe inline(expr,n) as
shorthand for inline(expr,"x","P1","P2",...,"Pn") is useful.

Paul Kienzle
pki...@us...

On Sat, Aug 10, 2002 at 06:37:46PM +0100, Etienne Grossmann wrote:
> 
>   Sorry, there was a part missing from the first attached func. Try
> the following.
> 
>   Etienne
> 
> On Sat, Aug 10, 2002 at 01:32:42PM +0000, Steffan Berridge wrote:
> # 
> # Hi,
> # 
> # I have just started using Octave, having used Matlab for a
> # while.  In Matlab I have often used inline functions, they
> # are very convenient for example when passing a function as
> # an argument.
> # 
> # For example, the following is a function returning the
> # square of the argument:
> # 
> # fun = inline('x^2','x');
> # 
> # One can then pass this function as an argument to another
> # function:
> # 
> # x = foo(fun,...);
> # 
> # So my question is, is there an equivalent construction in
> # Octave?  If not, what is the best way to pass functions as
> # arguments?
> # 
> # Thanks,
> # Steffan
> -- 
> Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

## [fname,fcode] = inline (str,arg1,...) - Define a function from a string
##
## INPUT : -----------
## str   : string    : String defining the result of the function
## argN  : string(s) : Names of the arguments of the function
##
## OUTPUT : ----------
## fname : string    : Name of the new function, which can e.g. be called w/
##                     feval() 
## fcode : string    : The code of the function
##
## EXAMPLE : ---------
##
## fn = inline ("x.^2 + 1","x");
## feval (fn, 6)        
##   ans = 36
##
function [fname,fcode] = inline (str,varargin)

## Choose a name (naive way : won't work zillions of times)
while 1
  fname = sprintf ("inline_func_%06i",floor (rand()*1e6));
  if exist (fname), break; end
end

## construct arg list
if nargin == 1
  argstr = "x";
else
  argstr = sprintf ("%s,",all_va_args);
  argstr = argstr(1:length(argstr)-1);
end

fcode = sprintf (["function r = %s (%s)\n",\
		  "  r = ",str,";\n",\
		  "endfunction;"],\
		 fname, argstr);
eval (fcode);
endfunction		  

[OctDev] Re: inline functions (Corrected source)

[Etienne G. <et...@is...>]

  Sorry, there was a part missing from the first attached func. Try
the following.

  Etienne

On Sat, Aug 10, 2002 at 01:32:42PM +0000, Steffan Berridge wrote:
# 
# Hi,
# 
# I have just started using Octave, having used Matlab for a
# while.  In Matlab I have often used inline functions, they
# are very convenient for example when passing a function as
# an argument.
# 
# For example, the following is a function returning the
# square of the argument:
# 
# fun = inline('x^2','x');
# 
# One can then pass this function as an argument to another
# function:
# 
# x = foo(fun,...);
# 
# So my question is, is there an equivalent construction in
# Octave?  If not, what is the best way to pass functions as
# arguments?
# 
# Thanks,
# Steffan
-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] Re: inline functions

[Etienne G. <et...@is...>]

  Hello again,

On Sat, Aug 10, 2002 at 01:32:42PM +0000, Steffan Berridge wrote:
#=20
# Hi,
#=20
# I have just started using Octave, having used Matlab for a
# while.  In Matlab I have often used inline functions, they
# are very convenient for example when passing a function as
# an argument.
#=20
# For example, the following is a function returning the
# square of the argument:
#=20
# fun =3D inline('x^2','x');

  Try the attached m-file, called inline.m.

  Is that, or something alike, worthy of octave-forge?

  Cheers,

  Etienne
=20
# One can then pass this function as an argument to another
# function:
#=20
# x =3D foo(fun,...);
#=20
# So my question is, is there an equivalent construction in
# Octave?  If not, what is the best way to pass functions as
# arguments?
#=20
# Thanks,
# Steffan
#=20
#=20
#=20
#=20
# -------------------------------------------------------------
# Octave is freely available under the terms of the GNU GPL.
#=20
# Octave's home on the web:  http://www.octave.org
# How to fund new projects:  http://www.octave.org/funding.html
# Subscription information:  http://www.octave.org/archive.html
# -------------------------------------------------------------
#=20
#=20
#=20
#=20
# =00=00=00=00=00=00=00=00=00
--=20
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

Re: [OctDev] vrml / computer vision

[Etienne G. <et...@is...>]

On Mon, Jul 22, 2002 at 01:36:04AM -0400, Paul Kienzle wrote:
# Everything is welcome on octave forge, it is just a question of where to
# put it.  If it is very general, then put it in main.  If it is very domain
# specific, then put it in extra.   There are more criteria but I've written
# them all down in a README of some sort (sorry I don't remember where, but
# I'm away at the moment and can't find out).  I presume you are willing to do
# whatever maintenance is required for it.  I tried getting it to run on a
# couple of occasions just to play, but didn't get very far.

  The CV or VRML code? The vrml part should be quite straightforward
to use; but freewrl is needed to be able to browse a model from
octave's command line. I will document it better and make arguments
coherent. I will use vrml_xyz () names.

  The vision part can wait until I properly wrap and document it.

  Cheers,

  Etienne
 
# - Paul
# 
# On Fri, Jul 19, 2002 at 11:02:55AM +0100, Etienne Grossmann wrote:
# > 
# >   Hello,
# > 
# > would some octave functions for producing/viewing vrml and for doing
# > some computer vision be welcome on octave-forge?
# > 
# > vrml functions :
# > 
# >   low level : produce vrml nodes, with more or less vrml-wise params.
# >   mid level : vrml_points : code that displays 3D points or spheres
# >               vrml_cyl    : code for a cylinder linking points
# >               vrml_arrow  : an arrow
# >               vrml_surf   : a 2-parameter surface
# >               vrml_faces  : arbitrary 3D faces
# >   browsing  : (only if the browser freewrl is installed)
# > 
# >               vrml_browse (string)  : view the code in string.
# > 
# > computer vision :
# > 
# >    3D reconstruction functions :
# > 
# >       Multi-view reconstruction from matched points. Determines camera
# >       position, orientation, calibration from a matrix of 2D
# >       observations.
# > 
# >       Single- or multi-view reconstruction from 2D points and known
# >       geometric proberties (planarity, orthogonality, parallelism ...)
# >       of the reconstructed objects.
# > 
# > 
# >   Cheers,
# > 
# >   Etienne
# > 
# > -- 
# > Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne
# > 
# > 
# > -------------------------------------------------------
# > This sf.net email is sponsored by:ThinkGeek
# > Welcome to geek heaven.
# > http://thinkgeek.com/sf
# > _______________________________________________
# > Octave-dev mailing list
# > Oct...@li...
# > https://lists.sourceforge.net/lists/listinfo/octave-dev
# 
# 

-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] vrml / computer vision

[Etienne G. <et...@is...>]

  Hello,

would some octave functions for producing/viewing vrml and for doing
some computer vision be welcome on octave-forge?

vrml functions :

  low level : produce vrml nodes, with more or less vrml-wise params.
  mid level : vrml_points : code that displays 3D points or spheres
              vrml_cyl    : code for a cylinder linking points
              vrml_arrow  : an arrow
              vrml_surf   : a 2-parameter surface
              vrml_faces  : arbitrary 3D faces
  browsing  : (only if the browser freewrl is installed)

              vrml_browse (string)  : view the code in string.

computer vision :

   3D reconstruction functions :

      Multi-view reconstruction from matched points. Determines camera
      position, orientation, calibration from a matrix of 2D
      observations.

      Single- or multi-view reconstruction from 2D points and known
      geometric proberties (planarity, orthogonality, parallelism ...)
      of the reconstructed objects.


  Cheers,

  Etienne

-- 
Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne

[OctDev] corrected kron.cc

[Obed S. <Obe...@gr...>]

FYI:

Attached is a corrected version of kron.cc.

oss

O. Scott Sands, Ph.D.
Communications System Analyst
NASA Glenn Research Center
Space Communications Office
Project Development Branch
MS 54-6
21000 Brookpark Rd.
Cleveland, OH  44135
Voice: (216) 433-2607
Fax:   (216) 433-6371
mailto:Obe...@gr...

[OctDev] nonlinear constrained optimization work

[Joseph P. S. <js...@js...>]

Hi. In Octave 2.0.14, I've prototyped multmethod, a robust general
purpose nonlinear constrained optimization routine; it is an
implementation of the method of multipliers, based on my Mathematica
work; cf

  http://www.mathsource.com/Content/Applications/Mathematics/0207-289

multmethod incorporates a selection of unconstrained solvers: a very
robust but somewhat expensive [as expected] unconstrained non-gradient
optimizer based on Nick Higham's work (mdsmax.m), a much faster but less
robust [as expected] bfgs-based (and may include a trust region)
optimizer based on Tim Kelley's work (bfgswopt.m, cgtrust.m, and
related), and a numerical gradient routine based on Peter Spellucci's
work (part of donlp2), all written as Octave M-files.  I have secured
permission to use the work of Tim Kelley and Nick Higham as a starting
point.  The tiny reimplemented numgrad routine will not be an issue.

This should meet the needs of the "free NLP solver" project, which can
also be used as the "free QP solver".

As requested by http://www.octave.org/projects.html:

	Also, if you start working steadily on a project,
	please let bug...@be... know.

I'm sending this email to let you know.

I'm aware of the optim work at sourceforge, and will be looking into
that in an attempt to reconcile differences and unify the offering.  I
still need to refine, document, validate, format, rename, reconcile, and
package this work; I expect to work on this again in a couple of months,
and it will take a couple of months of work to make it happen.

FYI.  /Jskud

Re: [OctDev] imrotate

[Andy A. <ad...@fr...>]

On Fri, 28 Jun 2002, Jeffery Orchard wrote:
> Your concernes about a .m implementation of imrotate are wise. However, 
> my implementation does not have any loops. It applies the rotation by 
> decomposing the rotaion matrix into 3 shears, and then applies the 
> shears using Fourier interpolation. As a result, it uses the 1D fft 
> function heavily.

Wow. That's cool. I didn't realize that you could do a 
rotate with shearing. Is the explanation simple?
Do you know of any (online) references that explain it?

andy

BTW. Dynamically loaded functions for octave are _very_
well designed and are quite easy to write. They're
much easier to learn than (for example) MATLAB mex syntax
or perl xs syntax
_______________________________________
Andy Adler,                ad...@nc...

Re: [OctDev] imrotate

[Jeffery O. <jj...@cs...>]

Your concernes about a .m implementation of imrotate are wise. However, 
my implementation does not have any loops. It applies the rotation by 
decomposing the rotaion matrix into 3 shears, and then applies the 
shears using Fourier interpolation. As a result, it uses the 1D fft 
function heavily. But since the bulk of those operations are implemented 
efficiently, the speed is OK. I haven't had the chance to run any 
benchmarks yet, but I will soon.

The function uses another function I could submit, called "imshear.m". 
Guess what it does! That's right... applies a shear using Fourier 
interpolation.

Certainly, a C implementation of these functions would be at LEAST 
slightly faster, but I must admit that I'm hesitant because I've never 
written a .oct routine, and I'm not familiar enough with the API to do 
it in a timely manner. Not to mention, I hope I wouldn't have to 
re-implement the fft, etc., myself. Since the vast majority of the data 
manipulation is offloaded to other routines, it might not be worth it.

Since it seems that there are no imrotates lurking around, I guess I 
wouldn't be re-inventing the wheel. I'll spend some time tying up some 
loose ends, and you can check out the "alpha release" once I'm done, if 
you wish.

Jeff

Andy Adler wrote:

>On Fri, 28 Jun 2002, Jeff Orchard wrote:
>
>>I have developed "imrotate.m". I noticed that it is not yet implemented 
>>in the octave collections. Is there one out there that I've simply not 
>>seen? Just want to check before going through the effort of making it 
>>compatible with RGB images, colormaps, etc.
>>
>
>I don't think we have imrotate. It would be a great thing
>to have.
>
>One question: don't you think an *.m file implementation
>would be slow?
>
>I have some C code that does image scaling and
>rotation. (greyscale only) I could contribute that if
>you'd be interested in doing an *oct file implementation.
>
>andy
>_______________________________________
>Andy Adler,
>ad...@nc...
>
>
>
>-------------------------------------------------------
>This sf.net email is sponsored by:ThinkGeek
>Caffeinated soap. No kidding.
>http://thinkgeek.com/sf
>_______________________________________________
>Octave-dev mailing list
>Oct...@li...
>https://lists.sourceforge.net/lists/listinfo/octave-dev
>

Re: [OctDev] imrotate

[Andy A. <ad...@fr...>]

On Fri, 28 Jun 2002, Jeff Orchard wrote:
> I have developed "imrotate.m". I noticed that it is not yet implemented 
> in the octave collections. Is there one out there that I've simply not 
> seen? Just want to check before going through the effort of making it 
> compatible with RGB images, colormaps, etc.

I don't think we have imrotate. It would be a great thing
to have.

One question: don't you think an *.m file implementation
would be slow?

I have some C code that does image scaling and
rotation. (greyscale only) I could contribute that if
you'd be interested in doing an *oct file implementation.

andy
_______________________________________
Andy Adler,
ad...@nc...

[OctDev] imrotate

[Jeff O. <jj...@cs...>]

Hello all,

I have developed "imrotate.m". I noticed that it is not yet implemented 
in the octave collections. Is there one out there that I've simply not 
seen? Just want to check before going through the effort of making it 
compatible with RGB images, colormaps, etc.

Jeff

[OctDev] Problems using vtk with tk_octave

[Obed S. <os...@sp...>]

I've got tk_octave to build with the blt library. However, I am unable
to link against vtk.

Here's the rub: The vtk distribution I use (4.0) does not actully
produce a vtk.h file. It makes (and installs) lots of include files but,
alas, no vtk.h file. When you configure octave_forge, it looks for vtk.h
and fails.

Is there a new version of tk_octave to work with vtk 4.0? Is there some
other way to make a vtk.h file?

Any hints are appreciated.

Scott Sands

Re: [OctDev] m-file error?

[Paul K. <pki...@ja...>]

On Wed, Jun 19, 2002 at 11:13:02AM -0500, Hammett, Rich (Contractor - SimTech) wrote:
> To whom should I direct questions about "toolkit" m-files?  My specific
> question is about the signal processing fir2.m file.  There is a line
> toward the bottom where the output is being centered and truncated,
> as so:
> 
> b = real ([ b([2*grid_n-floor(mid)+1:2*grid_n]) ; b(1:ceil(mid)) ]);
> 
> However, the left half of that vector only goes to 2*grid_n, while
> the input data goes to 2*grid_n + 1.  And since we're wrapping the
> output data, the point being left out is from the middle of the
> data (just left of DC), not an outlier.

	grid = interp1(f,m,linspace(0,1,grid_n+1)');
				 	^^^^^^^^
grid will have grid_n+1 points.

  	b = ifft([grid ; grid(grid_n:-1:2)]);
			      ^^^^^^^^^^^
length(grid_n:-1:2) is grid_n-1, so b will have 
grid_n+1 + grid_n-1 = 2*grid_n points, not 2*grid_n + 1.  The
reflection contains neither the 0 nor the pi frequency components.

Paul Kienzle
pki...@us...

[OctDev] m-file error?

[Hammett, R. (C. - SimTech) <ric...@rd...>]

To whom should I direct questions about "toolkit" m-files?  My specific
question is about the signal processing fir2.m file.  There is a line
toward the bottom where the output is being centered and truncated,
as so:

b = real ([ b([2*grid_n-floor(mid)+1:2*grid_n]) ; b(1:ceil(mid)) ]);

However, the left half of that vector only goes to 2*grid_n, while
the input data goes to 2*grid_n + 1.  And since we're wrapping the
output data, the point being left out is from the middle of the
data (just left of DC), not an outlier.

Is there some way I can get in touch with the developer to 
make sure I'm not missing something important?  The developer is
Paul Kienzle according to the file.

Thanks!

rich hammett

[OctDev] re: optim -- please have gradient return COLUMN vector

[Joseph P. S. <js...@js...>]

Hi.  I was looking in the most recent packaged download
(octave-forge-2002.05.09.tar.gz) at the main/optim code, and noticed
that bfgs.m states that derivative must return a row vector.  This is
pretty non-standard, and is likely to annoy or confuse folks.

Bertsekas's Nonlinear Programming (2nd ed, 1999, p. 664) clearly states
that Df is a column vector.  Nocedal and Wright's Numerical Optimization
(1999, p. 582, eq. A.12) clearly states that Df is a column vector.  Tim
Kelley's Iterative Methods for Optimization (1999, p. 4) clearly states
that vectors (including gradients) are assumed to be column vectors.

So please, let's use standard conventions, and have the gradient of a
scalar function returned as a column vector (and the derivative of a
vector value function returned as [ Df1 Df2 ... ], the transpose of the
Jacobian).

Thanks.

/Jskud

[OctDev] Manipulating the CVS tree through CVSROOT/loginfo

[Rafael L. <lab...@mp...>]

Hi,

Some days ago, on Paul Kienzle's request, I removed the annoying tsa/
directory at the top level of the CVS repository for octave @ SourceForge.
As promised, I will tell you how it has been done.

First, I checked out the CVSROOT directory, added the line:

    ^CVSROOT       rm -rf $CVSROOT/tsa    

to the loginfo file and CVS committed it.

Finally, I removed that same line and committed again.  The magic was done.

[Actually, if you look at the CVS history of the loginfo file, you will see
that I did more things than told above, mainly preliminary experimentation,
but the net result was that the tsa directory has been removed.]

Since arbritary shell commands can be put in loginfo, please be very careful
when using the loginfo technique to manipulate the CVS tree.

-- 
Rafael Laboissiere

Re: [OctDev] stripping binaries

[Andy A. <ad...@nc...>]

On Wed, 22 May 2002, Paul Kienzle wrote:
> In the latest release of octave-forge, I moved the binary stripping code
> from sparse/Makefile to octave-forge/Makeconf so now all oct-files are
> stripped after compiling.  This happens to break the Mac OSX version of
> octave-forge, so something has to change.  This is arguably an octave bug,
> so it may be octave which changes.

I also suspect this is an octave bug. The various strip options with
octave don't work well.

1. make install-strip in 2.1.36 doesn't work for me.

2. mkoctfile with the strip options sometimes doesn't work -
   but I forget when.

> Anybody have any opinions on whether oct-files should be stripped?
> I see smaller binaries.  Is that mostly due to the -g option?
> Are there speed considerations, or are those due to -g as well?
> Should we selectively strip some of the binaries?

In my opinion, standard installs of executables should be stripped.
This can be a 10 times saving in space. Developers who can
use gdb know how to leave in symbols anyway.

andy
_______________________________________
Andy Adler,                ad...@nc...

[OctDev] stripping binaries

[Paul K. <pki...@ja...>]

Hi,

In the latest release of octave-forge, I moved the binary stripping code
from sparse/Makefile to octave-forge/Makeconf so now all oct-files are
stripped after compiling.  This happens to break the Mac OSX version of
octave-forge, so something has to change.  This is arguably an octave bug,
so it may be octave which changes.

Anybody have any opinions on whether oct-files should be stripped?
I see smaller binaries.  Is that mostly due to the -g option?
Are there speed considerations, or are those due to -g as well?
Should we selectively strip some of the binaries?

Thanks for your input.

Paul Kienzle
pki...@us...

Re: [OctDev] more cleaning

[Paul K. <pki...@ja...>]

I didn't include any of the optim/test* functions in make check this time
around because they weren't working consistently.  Has that been
addressed?

Specifically, for make check I need to know if the functions are working as
expected, or if there has been some change in octave or octave-forge which
is causing them to break.  It would be nice if the optimization routines
converge to the correct value in a reasonable time no matter what initial
conditions they are presented with, but until they do make check needs to
test from known good initial conditions.

Regarding tests with random input, I'm open to suggestions.  Maybe put them
in make icheck since the user is expected to watch the output of that to
verify that it is reasonable.  In that case, use a set of random initial
conditions and report the probability of them converging correctly like
Andy does in sparse:

octave -qf sp_test.m 2>/dev/null
178 operations tested sucessfully for 100 iterations
operation #80 in sp_test.m exceeds error tolerance with probability  1.00%
operation #125 in sp_test.m exceeds error tolerance with probability  1.00%
operation #173 in sp_test.m exceeds error tolerance with probability  2.00%
operation #174 in sp_test.m exceeds error tolerance with probability  7.00%
operation #178 in sp_test.m exceeds error tolerance with probability  1.00%

BTW, I consider batch_test to be a temporary fix.  make check really needs
to be distributed throughout the tree.

Paul Kienzle
pki...@us...

On Fri, May 17, 2002 at 07:30:43PM +0100, Etienne Grossmann wrote:
> 
>   Also,
> 
> cg_min.m, bracket_min.m, semi_bracket.m, brent_line_min.m and the
> associated test functions can also be removed. I will rename
> test_cg_min_N.m test_min_N.m.
> 
>   Cheers,
> 
>   Etienne
> 
> -- 
> Etienne Grossmann ------ http://www.isr.ist.utl.pt/~etienne
> 
> _______________________________________________________________
> 
> Hundreds of nodes, one monster rendering program.
> Now that’s a super model! Visit http://clustering.foundries.sf.net/
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev