Nabble - Octave - General

Re: keyboard does not provide local context

2009-11-23T23:11:59Z

On Fri, Nov 20, 2009 at 5:01 PM, John W. Eaton <jwe@...> wrote:

On 20-Nov-2009, Eric Chassande-Mottin wrote:

| using foo.m below,
| if I say foo(1) from prompt I get access to local vars
|
| octave3.2:1> foo(1)
| keyboard: stopped in /home/ecm/foo.m
| a = 1
| debug> whos
| Variables in the current scope:
|
| Attr Name Size Bytes Class
| ==== ==== ==== ===== =====
| a 1x1 8 double
| argn 1x1 1 char
| f x 1x1 8 double
|
| Total is 3 elements using 17 bytes
|
| but when calling foo() from a script I don't
|
| octave3.2:1> test
| keyboard: stopped in /home/ecm/foo.m
| b = 2
| a = 1
| debug> whos
| Variables in the current scope:
|
| Attr Name Size Bytes Class
| ==== ==== ==== ===== =====
| ans 1x30 30 char
| b 1x1 8 double
|
| Total is 31 elements using 38 bytes
|
| why is that? is there a way to access local vars when calling from a script?
|
| I am using
|
| GNU Octave, version 3.2.3

It's best to report bugs to the bug@... list.

This problem has been reported several times now and it is fixed in
the current development version of Octave.

I think the relevant patches are

http://hg.savannah.gnu.org/hgweb/octave/rev/fb22dd5d6242
http://hg.savannah.gnu.org/hgweb/octave/rev/25c2e92ee03c
http://hg.savannah.gnu.org/hgweb/octave/rev/bbe033dcfe13

Jaroslav, I think it is safe to apply the first two to the 3.2.x
branch. I'm not so sure about the third one though as it removes some
variables from a class and changes at least one function signature,
which might cause some binary compatibility problems.

These don't apply due to unwind_protect changes. I think the second one reverts the first one, so maybe it would suffice to adapt the second one for 3.2.x. Does it make sense without also taking the third one?

--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: playaduio in octave 3.2.3

2009-11-23T18:31:02Z

on 11/23/09 5:19 AM, Jaroslav Hajek at highegg@... wrote:

> On Sat, Nov 21, 2009 at 6:39 AM, Adrian pamin <relleradrian@...>wrote:
>
>> Hi,
>> Im a little new to using octave and GNU systems but there seems to be a
>> problem with the playaudio() function in the Octave 3.2.3 patch. When trying
>> to use the playaudio(x) the error "The system cannot find the path
>> specified." I have tried this in a Vista and Win XP system yielding the same
>> results. Below is the code used:
>>
>> t = [0:1/8000:2]
>> sig3 = 0.8*sin(2*pi*3*t)
>> plot(t,sig3)
>> playaudio(sig3)
>>
>> My friends and professor are stumped with this error as they can use the
>> function without problems on a Linux and Mac OS. Is this possibly a Windows
>> problem? would switching to a VM or dual booting help? Thanks in advance...
>>
>>
> I'm afraid the current playaudio is only designed to run on Unix systems,
> because all it does is send data to /dev/dsp or /dev/audio. There could
> probably be an alternative path for Windows, using a temporary wav file or
> something similar.
>
> hth

I cannot find /dev/dsp or /dev/audio on my mac:
[/dev] -bash-2.05b 504$ locate /dev/dsp

[/dev] -bash-2.05b 505$ locate /dev/audio

[/dev] -bash-2.05b 506$

Accordingly? :
octave-3.2.2:1> t = [0:1/8000:2];
octave-3.2.2:2> sig3 = 0.8*sin(2*pi*3*t);
octave-3.2.2:3> plot(t,sig3)
octave-3.2.2:4> playaudio(sig3)
sh: line 1: /dev/dsp: Permission denied
octave-3.2.2:5> help playaudio
`playaudio' is a function from the file
/Applications/Octave.app/Contents/Resources/share/octave/3.2.2/m/audio/playa
udio.m

-- Function File: playaudio (NAME, EXT)
-- Function File: playaudio (X)
Plays the audio file `NAME.EXT' or the audio data stored in the
vector X.

See also: lin2mu, mu2lin, loadaudio, saveaudio, setaudio, record

Henry

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: Need help with nelder_mead_min.m

2009-11-23T18:07:12Z

Hello,

something like

octave:408> function val = foo (xyz), val = sum (xyz .^ 2); endfunction
octave:409> help nelder_mead_min
[snip]
### Start nelder_mead_min() from a random point
octave:410> [xyz0, val0, nev] = nelder_mead_min ("foo", randn(1,3))
xyz0 =

2.1237e-08 1.0243e-08 -4.4610e-09

val0 = 5.7582e-16
nev = 201
### Start nelder_mead_min() from the minimum
octave:411> [xyz0, val0, nev] = nelder_mead_min ("foo", [0 0 0])
xyz0 =

0 0 0

val0 = 0
nev = 236

Hth,

Etienne

Quoting ???????? ???????
<chainreactionerator@...> (Mon, 23 Nov 2009 21:43:57 +0300):

> Hi all!
>
> I can't figure out how to work with this function. I will be very
> glad if one could explain the subject.
>
>
> For an example lets see the problem.
>
>
> function y = foo (x, y, z)
> y = x**2 + y**2 + z**2;
> endfunction
>
>
> I need to find a minima of the foo in [-1;1]*[-1;1]*[-1;1] region.
> I know the answer, but how I must set up an options and arguments
> for nelder_mead_min.m?
>
>
> Best regards,
> Grigory

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

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: Sound File Images?

2009-11-23T17:25:33Z

2009/11/20 rs-232 <rs@...>:
> Pardon the Noob question, but does anyone know the best method for
> transferring the data from a range of samples of a sound file into a
> printable image file?

Have you tried using the functions from the audio package in Octave Forge?

http://octave.sourceforge.net/doc/funref_audio.html
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: imshow is slow on Mac OSX 10.5

2009-11-23T17:22:58Z

After more examination I realize that the display range being double
won't have much effect on performance.

On Mon, Nov 23, 2009 at 12:04 PM, Todd Rovito <rovitotv@...> wrote:

> Carlo,
> Thanks for the information! For sure the FLTK backend is much
> faster. I was looking at the imshow command to GNUPlot and was
> surprised to see that the code seems to be converting the data to
> double. Here is the code snippet from imshow.m:
>
> ========================================================
> ## Set default display range if display_range not set yet.
> if (isempty (display_range))
> display_range = [min(im(:)), max(im(:))];
> elseif (isna (display_range))
> t = class (im);
> switch (t)
> case {"double", "single", "logical"}
> display_range = [0, 1];
> case {"int8", "int16", "int32", "uint8", "uint16", "uint32"}
> ## For compatibility, uint8 data should not be handled as
> ## double. Doing so is a quick fix to allow the images to be
> ## displayed correctly.
> display_range = double ([intmin(t), intmax(t)]);
> otherwise
> error ("imshow: invalid data type for image");
> endswitch
> endif
> ======================================================
>
> Notice that the uint8 data is actually converted to double. If I read
> the code correctly that would make the transmission to GNU Plot over
> the pipe much slower. I will have to look into the FLTK backend too.
> Again thanks for the help.
>
> On Mon, Nov 23, 2009 at 2:16 AM, Carlo de Falco <carlo.defalco@...> wrote:
>>
>> On 23 Nov 2009, at 12:30, Todd Rovito wrote:
>>
>>> I just started to use Octave for some image processing work and have
>>> noticed how slow Octave is to display images in gnu-plot.
>>> On my 1
>>> year old Mac Book Pro it can take 5 seconds to display a 4 mega-pixel
>>> 8 bit image. Attached is the function I use to read such image. I am
>>> wondering if I have something installed wrong or this is a natural
>>> behavior or Octave? Please understand I am not complaining about
>>> Octave's performance, it seems speedy on everything except for image
>>> display.
>>>
>>>
>>> Thanks for the help.
>>
>>
>> I am afraid this is due to the way communication between Octave and gnuplot
>> is implemented.
>> As communication is implemented via pipes, transferring the image data to
>> gnuplot takes about
>> the same time as writing the whole uncompressed image data to a file and
>> then reading it back.
>> This is one of the reasons why a new graphics backend based on OpenGL is
>> currently under
>> developement.
>> To get an idea of the speed improvement given by the new backend you can
>> try:
>>
>> sombrero (500)
>> close all
>> backend ('fltk')
>> sombrero (500)
>>
>> The difference is really impressive on my own white plastic Core 2 Duo
>> MacBook.
>> unfortunately the new backend does not yet support imshow in the current
>> stable release.
>> If this feature is important to you might want to consider joining the
>> developers mailing list
>> and helping get Octave 3.4 out soon.
>>
>> HTH,
>> c.
>>
>>
>>
>>
>

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: imshow is slow on Mac OSX 10.5

2009-11-23T14:04:28Z

Carlo,
Thanks for the information! For sure the FLTK backend is much
faster. I was looking at the imshow command to GNUPlot and was
surprised to see that the code seems to be converting the data to
double. Here is the code snippet from imshow.m:

========================================================
## Set default display range if display_range not set yet.
if (isempty (display_range))
display_range = [min(im(:)), max(im(:))];
elseif (isna (display_range))
t = class (im);
switch (t)
case {"double", "single", "logical"}
display_range = [0, 1];
case {"int8", "int16", "int32", "uint8", "uint16", "uint32"}
## For compatibility, uint8 data should not be handled as
## double. Doing so is a quick fix to allow the images to be
## displayed correctly.
display_range = double ([intmin(t), intmax(t)]);
otherwise
error ("imshow: invalid data type for image");
endswitch
endif
======================================================

Notice that the uint8 data is actually converted to double. If I read
the code correctly that would make the transmission to GNU Plot over
the pipe much slower. I will have to look into the FLTK backend too.
Again thanks for the help.

On Mon, Nov 23, 2009 at 2:16 AM, Carlo de Falco <carlo.defalco@...> wrote:

>
> On 23 Nov 2009, at 12:30, Todd Rovito wrote:
>
>> I just started to use Octave for some image processing work and have
>> noticed how slow Octave is to display images in gnu-plot.
>> On my 1
>> year old Mac Book Pro it can take 5 seconds to display a 4 mega-pixel
>> 8 bit image. Attached is the function I use to read such image. I am
>> wondering if I have something installed wrong or this is a natural
>> behavior or Octave? Please understand I am not complaining about
>> Octave's performance, it seems speedy on everything except for image
>> display.
>>
>>
>> Thanks for the help.
>
>
> I am afraid this is due to the way communication between Octave and gnuplot
> is implemented.
> As communication is implemented via pipes, transferring the image data to
> gnuplot takes about
> the same time as writing the whole uncompressed image data to a file and
> then reading it back.
> This is one of the reasons why a new graphics backend based on OpenGL is
> currently under
> developement.
> To get an idea of the speed improvement given by the new backend you can
> try:
>
> sombrero (500)
> close all
> backend ('fltk')
> sombrero (500)
>
> The difference is really impressive on my own white plastic Core 2 Duo
> MacBook.
> unfortunately the new backend does not yet support imshow in the current
> stable release.
> If this feature is important to you might want to consider joining the
> developers mailing list
> and helping get Octave 3.4 out soon.
>
> HTH,
> c.
>
>
>
>

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: Need help with nelder_mead_min.m

2009-11-23T11:23:35Z

I got this to work:

>> function w=foo(x); w=x.*x+2;endfunction;
>> x=[-1:.01:1];
>> nelder_mead_min("foo","x")
Answer 2

> Hi all!
>
> I can't figure out how to work with this function. I will be very glad if
> one could explain the subject.
>
> For an example lets see the problem.
>
> function y = foo (x, y, z)
> y = x**2 + y**2 + z**2;
> endfunction
>
> I need to find a minima of the foo in [-1;1]*[-1;1]*[-1;1] region. I know
> the answer, but how I must set up an options and arguments for
> nelder_mead_min.m?
>
> Best regards,
> Grigory

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: playaduio in octave 3.2.3

2009-11-23T11:09:47Z

--- On Mon, 11/23/09, Jaroslav Hajek <highegg@...> wrote:

> From: Jaroslav Hajek <highegg@...>
> Subject: Re: playaduio in octave 3.2.3
> To: "Adrian pamin" <relleradrian@...>
> Cc: help@...
> Date: Monday, November 23, 2009, 5:19 AM
>
>
> On Sat, Nov 21, 2009 at 6:39 AM,
> Adrian pamin <relleradrian@...>
> wrote:
>
> Hi,
> Im a little new to using octave and GNU systems but there
> seems to be a problem with the playaudio() function in the
> Octave 3.2.3 patch. When trying to use the playaudio(x) the
> error "The system cannot find the path specified."
> I have tried this in a Vista and Win XP system yielding the
> same results. Below is the code used:
>
>
> t = [0:1/8000:2]
> sig3 = 0.8*sin(2*pi*3*t)
> plot(t,sig3)
> playaudio(sig3)
>
> My friends and professor are stumped with this error as
> they can use the function without problems on a Linux and
> Mac OS. Is this possibly a Windows problem? would switching
> to a VM or dual booting help? Thanks in advance...
>
>
>
>
>
>
>
> I'm afraid the current playaudio is only designed to
> run on Unix systems, because all it does is send data to
> /dev/dsp or /dev/audio. There could probably be an
> alternative path for Windows, using a temporary wav file or
> something similar.
>
>
> hth
>
> --
> RNDr. Jaroslav Hajek
> computing expert & GNU Octave developer
> Aeronautical Research and Test Institute (VZLU)
> Prague, Czech Republic
> url: www.highegg.matfyz.cz

'sox' works both on Linux and Windows and can accept data from stdin.

Under Windows sox' needs CYGWIN dll, but, I think, 'sox' is distributed in
binary for with the dll. I.e. in practical terms it behaves like regular
Windows console application - it doesn't need full fledged CYGWIN.

It might make sense to implement octave -> sox interface.

Regards,
Sergei.

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Need help with nelder_mead_min.m

2009-11-23T10:43:57Z

Hi all!

I can't figure out how to work with this function. I will be very glad if one could explain the subject.

For an example lets see the problem.

function y = foo (x, y, z)

y = x**2 + y**2 + z**2;

endfunction

I need to find a minima of the foo in [-1;1]*[-1;1]*[-1;1] region. I know the answer, but how I must set up an options and arguments for nelder_mead_min.m?

Best regards,

Grigory

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: Printing a figure as a png... This should be basic!

2009-11-23T10:04:58Z

2009.11.23.13:04:51 EST

Hi Kristen (and octave folks),

I've been using only the 'print' command
in octave-3.2.3 to do the kind of thing that
you need:

print(handle,"graph.png","-dpng","-S429,348");

I've trained myself to stay off the gnuplot
calls. The octave developers keep saying
'gnuplot calls deprecated'.

It is likely that the 'handle' argument is not
needed when the current graph is the only one.

I've been using double-quotes for the
arguments. Again, I'm not certain that this is
absolutely required, but it works for me under
the CygWin/bash/octave environment with the
Xwin.exe version of x11.

The last argument is how I force the
image exporter (I believe that this is
ImageMagick ?) to give my output image the
correct proportions.

Try the 'print' call as above. If it
does not work, then perhaps you have found
a bug.

George gbarrick@...

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: Passing an array from java to octave

2009-11-23T09:42:32Z

Hi Kim,all

thanks, I have been using this, however the problem is when you want to pass values from an existing java array to an octave array. The fact that the octave array type is an OctaveMatrix type (extension of OctaveType), complicates a bit things. So I have written this somehow crude but effective script for copying one by one the elements of the java array to the octave workspace. I attach the code, it creates a 10x5 array (easily made as parameters) , initializes the values to 5 and then passes it to octave. Just for future reference if someone else needs it.

          OctaveEngine octave = new OctaveEngineFactory().getScriptEngine();
         int rows=10;
         int columns=5;
         OctaveMatrix arr=new OctaveMatrix(rows,columns);

        int octave_array_index_offset_r,octave_array_index_offset_c;

        //Creation and initialization of test array
        int[][] a2 = new int[10][5];
         // print array in rectangular form
         for (int r=0; r<a2.length; r++) {
             for (int c=0; c<a2[r].length; c++) {

                 a2[r][c]=5;
                 System.out.print(" " + a2[r][c]);
                 octave_array_index_offset_r=r+1; //octave measures array indexes from 1, java from 0
                 octave_array_index_offset_c=c+1;
                //Passing of array to octave workspace
                 octave.eval("arr("+octave_array_index_offset_r+","+octave_array_index_offset_c+")="+a2[r][c]);
            }
             System.out.println("");
         }

        //Saving of variables to octave file
        octave.eval("save params.m arr");
        octave.close();

Kim Hansen wrote:

2009/11/2 George Kousiouris gkousiou@...:

Hi all,

I want to pass a Java array to Octave, so that I can process the data
inside. Does anyone have an idea how this can be done? I have tried a
java to octave bridge (javaoctave-0.4.0.jar) however this seems to be
only for executing Octave commands inside a Java program and not
actually passing data from Java to Octave...


JavaOctave can transfer many different types of data between Java and Octave.

In the following simple example "a" is a 2x2 matrix that is passed
from Java to Octave and "b" is pulled out of Octave.

The example is from:
http://kenai.com/projects/javaoctave/pages/SimpleExampleOfJavaOctaveUsage
==============
 final OctaveEngine octave = new OctaveEngineFactory().getScriptEngine();
 final OctaveMatrix a = new OctaveMatrix(new double[] { 1, 2, 3, 4 }, 2, 2);
 octave.put("a", a);
 final String func = "" //
         + "function res = my_func(a)\n" //
         + " res = 2 * a;\n" //
         + "endfunction\n" //
         + "";
 octave.eval(func);
 octave.eval("b = my_func(a);");
 final OctaveMatrix b = octave.get("b");
 octave.close();
==============

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

RE: Printing a figure as a png... This should be basic!

2009-11-23T08:56:30Z

Here is some working example code I use in Octave:

figure(4);

plot (freq_dev_array/1e9,ild,";insertion loss deviation;", freq_dev_array/1e9, -il_max, ";bottom limit;",freq_dev_array/1e9,il_max,";top limit;");

title('Insertion loss deviation');

xlabel('Freq (Ghz)');

ylabel('ILD (db)');

grid on

print -dpng x.png

in gnuplot I use:

set terminal png font "arial, 30" size 3200, 2400 transparent

set output ‘foo.png’

I hope this helps.

-Eddie Frie

(Not speaking for Intel)

From: Kristen Richter [mailto:christy1865@...]
Sent: Friday, November 20, 2009 11:06 AM
To: help@...
Subject: Printing a figure as a png... This should be basic!

Hi Octave family,

I feel rather embarrassed, but I have been unable to print a figure as a png despite several searches on help forums.

The commands I am using are as follows:
print('graph.png','-dpng')

And I get the following error:

octave:13> print('graph1','-dpng')

gnuplot> set terminal png enhanced size 1200,900
                      ^
         line 0: unknown or ambiguous terminal type; type just 'set terminal' for a list

         line 0: No terminal defined

gnuplot> plot "-" binary format='%float64' record=15 using ($1):($2) axes x1y1 title "" with lines linestyle 1 , "-" binary format='%float64' record=15 using ($1):($2) axes x1y1 title "" with lines linestyle 2 , "-" binary format='%float64' record=15 using ($1):($2) axes x1y1 title "" with lines linestyle 3 , "-" binary format='%float64' record=15 using ($1):($2) axes x1y1 title "" with lines linestyle 4 , "-" binary format='%float64' record=15 using ($1):($2) axes x1y1 title "" with lines linestyle 5 , "-" binary format='%float64' record=15 using ($1):($2) axes x1y1 title "" with lines linestyle 6 ;
              ^
         line 0: use 'set term' to set terminal type first

gnuplot> @��g@��~��(@
         ^
         line 0: invalid character @

gnuplot> ��#)@
         ^
         line 0: invalid character �

This error also applies to attempting to print as a jpeg, but not as an eps... Perhaps because it doesn't have to convert the file?

Searching google for the first two lines of the error message gives results reporting problems related to a multitude of plotting functionalities in multiple versions of Octave on different operating systems. Is there something simple I can correct to print a png (running Octave 3.2.2 and Ubuntu)?

Thank you in advance for your help. I'd appreciate any advice.

-Kristen

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: bwlabeln or 3d version of bwlabel

2009-11-23T07:10:23Z

> Is there an opensource implementation for multidimensional connected
> component labeling?

Not that I know of. If you find one or decides to implement yourself,
please let me know, as that would be a valuable contribution to the
'image' package.

The scipy.ndimage has a label function.

http://www.google.com/codesearch/p?hl=en&sa=N&cd=1&ct=rc#sL5YdHMOlIg/scipy-0.4.9/Lib/ndimage/src/ni_measure.c&q=ndimage%20numpy

http://docs.scipy.org/doc/scipy/reference/generated/scipy.ndimage.measurements.label.html#scipy.ndimage.measurements.label

the ndimage is advertised for multidimensional arrays. The label function does not specifically list any limitations for dim > 2. I will see if I can translate it to octave, but i haven't been able to compile octave on mingw yet... Had trouble finding the glob library that is supposed to be on the octave website : http://hg.savannah.gnu.org/hgweb/octave/file/8d9e4752441a/README.MSVC

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: playaduio in octave 3.2.3

2009-11-23T05:19:10Z

On Sat, Nov 21, 2009 at 6:39 AM, Adrian pamin <relleradrian@...> wrote:

Hi,
Im a little new to using octave and GNU systems but there seems to be a problem with the playaudio() function in the Octave 3.2.3 patch. When trying to use the playaudio(x) the error "The system cannot find the path specified." I have tried this in a Vista and Win XP system yielding the same results. Below is the code used:

t = [0:1/8000:2]
sig3 = 0.8*sin(2*pi*3*t)
plot(t,sig3)
playaudio(sig3)

My friends and professor are stumped with this error as they can use the function without problems on a Linux and Mac OS. Is this possibly a Windows problem? would switching to a VM or dual booting help? Thanks in advance...

I'm afraid the current playaudio is only designed to run on Unix systems, because all it does is send data to /dev/dsp or /dev/audio. There could probably be an alternative path for Windows, using a temporary wav file or something similar.

hth

--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: imshow is slow on Mac OSX 10.5

2009-11-23T04:16:37Z

On 23 Nov 2009, at 12:30, Todd Rovito wrote:

> I just started to use Octave for some image processing work and have
> noticed how slow Octave is to display images in gnu-plot.
> On my 1
> year old Mac Book Pro it can take 5 seconds to display a 4 mega-pixel
> 8 bit image. Attached is the function I use to read such image. I am
> wondering if I have something installed wrong or this is a natural
> behavior or Octave? Please understand I am not complaining about
> Octave's performance, it seems speedy on everything except for image
> display.
>
>
> Thanks for the help.

I am afraid this is due to the way communication between Octave and
gnuplot is implemented.
As communication is implemented via pipes, transferring the image data
to gnuplot takes about
the same time as writing the whole uncompressed image data to a file
and then reading it back.
This is one of the reasons why a new graphics backend based on OpenGL
is currently under
developement.
To get an idea of the speed improvement given by the new backend you
can try:

sombrero (500)
close all
backend ('fltk')
sombrero (500)

The difference is really impressive on my own white plastic Core 2 Duo
MacBook.
unfortunately the new backend does not yet support imshow in the
current stable release.
If this feature is important to you might want to consider joining the
developers mailing list
and helping get Octave 3.4 out soon.

HTH,
c.

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

imshow is slow on Mac OSX 10.5

2009-11-23T03:30:17Z

I just started to use Octave for some image processing work and have
noticed how slow Octave is to display images in gnu-plot. On my 1
year old Mac Book Pro it can take 5 seconds to display a 4 mega-pixel
8 bit image. Attached is the function I use to read such image. I am
wondering if I have something installed wrong or this is a natural
behavior or Octave? Please understand I am not complaining about
Octave's performance, it seems speedy on everything except for image
display.

Thanks for the help.

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

ReadRawImage.m (446 bytes) Download Attachment

Re: bwlabeln or 3d version of bwlabel

2009-11-22T23:43:40Z

søn, 22 11 2009 kl. 22:37 -0500, skrev Naveen Garg:
> Is there an opensource implementation for multidimensional connected
> component labeling?

Not that I know of. If you find one or decides to implement yourself,
please let me know, as that would be a valuable contribution to the
'image' package.

Søren

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: fltk backend and undefined gluOrtho2D

2009-11-22T20:00:44Z

I had the same problem. I could compile with no problem after installing libftgl-dev.
Cheers...
Jorge

Jeff Keacher wrote:

John W. Eaton-3 wrote:

On 21-Jul-2009, teuobk wrote:

| John W. Eaton-3 wrote:
| >
| > On 16-Jul-2009, Alexander Barth wrote:
| >
| > | Hi all,
| > | I try to compile a recent checkout of octave (from
| > | http://www.octave.org/hg/octave). The compilation fails at:
| > |
| > | ./DLD-FUNCTIONS/fltk_backend.cc: In member function �3u=1rxvirtual void
| > | OpenGL_fltk::draw_overlay()�3u=1ry:
| > | ./DLD-FUNCTIONS/fltk_backend.cc:145: error: �3u=1rxgluOrtho2D�3u=1ry was not
| > | declared in this scope
| > |
| > | I have fltk version 1.1.7. Is this version recent enough? On my system
| > | gluOrtho2D is declared in /usr/include/GL/glu.h but this file is not
| > | included in fltk_backend.cc.
| >
| > On my system, /usr/include/GL/glu.h is part of the libglu1-mesa-dev
| > package, and the version of that package is 7.0.3. I also have
| > version 1.1.9 of libfltk1.1-dev.
| >
|
| Ran into this problem, too. I have fltk 1.1.9 on an Intrepid-release Ubuntu
| box, and I was trying to compile Octave 3.2.0 from source. My solution was
| to edit fltk_backend.cc to #include <FL/glu.h> . Now everything builds
| fine, and all of the integrated tests pass.

Strange that you need to include this file and I don't. Does anyone
know why that might be?

At least on my system, the file FL/glu.h doesn't appear to declare
gluOrtho2D directly, but gets the declaration when it includes
GL/glu.h.

The file src/DLD-FUNCTIONS/fltk_backend.cc includes gl-render.h which
includes GL/glu.h if it exists. So if everything is configured
correctly, I think the symbol gluOrtho2D should be defined correctly,
and no changes should be required.

Does your system have GL/glu.h? Does it provide a declaration for
the gluOrtho2D function? Does Octave's configure script find GL/glu.h
and define HAVE_GL_GLU_H on your system?

Indeed, gluOrtho2D is declared in GL/glu.h. I included it (indirectly) via FL/glu.h mainly for consistency with the other includes in that file, which tend to pull from the FL directory.

Curiously, Octave's configuration script seems to find GL/glu.h, as it defines HAVE_GL_GLU_H. Since that is defined, GL/glu.h should be included via gl-render.h, which of course is included conditionally if HAVE_GL_GLU_H is defined -- but that didn't seem to be happening.

To make matters worse, I backed out the change to fltk_backend.cc, ran make clean, and tried the build again in an attempt to reproduce the problem. This time the build completed without errors. Odd. I assume that something is sticking around from the previous builds, but I'm not sure what it could be.

Jeff

bwlabeln or 3d version of bwlabel

2009-11-22T19:37:26Z

Is there an opensource implementation for multidimensional connected component labeling?

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: calling octave from C

2009-11-22T18:04:25Z

On Sun, Nov 22, 2009 at 4:04 PM, Jaroslav Hajek <highegg@...> wrote:
> As a matter of fact, I do *not* really understand how this conclusion
> follows from the text of GPL 3, I just trust the FAQ. I respect the opinion

I can't say that I do either. I think the FSF avoids specific language
within the FAQ statements purposefully to avoid helping scumbags find
holes in the licenses.

Based on the FAQ entry it may be that if someone were to create a new
program that combines octave, pytave and python into a new
GPL-licensed single work then it may be possible to monkey with what
has been said in the FAQ. That said, the FAQ is not the GPL so who
knows if that would be successful at all anyway.

In any case octave+pytave+python is not currently a single work and
the FAQ entry applies immediately to pytave's relationship to python.
Again, sorry for the noise.

--judd
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: exporting sparse matrix or inverse of spconvert

2009-11-22T16:19:47Z

tinku99 wrote:
> I want access to the 3 vectors representing rows, columns, and values of my
> sparse matrix.
> Is there a way to get at these easier than having to iterate through the
> nonsparse matrix?
>
Is there anything wrong with

[r, c, v] = find (S);

D.

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: exporting sparse matrix or inverse of spconvert

2009-11-22T16:17:01Z

ok, find does it...
[i, j, v] = find (a);
somehow i skipped over that in the octave manual
luckily searching the matlab google group found it easily
http://www.google.com/groups/search?ie=UTF-8&q=spconvert&btnG=Search&sitesearch=groups.google.com

tinku99 wrote:

I want access to the 3 vectors representing rows, columns, and values of my sparse matrix.
Is there a way to get at these easier than having to iterate through the nonsparse matrix?

exporting sparse matrix or inverse of spconvert

2009-11-22T16:08:37Z

I want access to the 3 vectors representing rows, columns, and values of my sparse matrix.
Is there a way to get at these easier than having to iterate through the nonsparse matrix?

I am trying to find the orientation of the largest connected region in a binary image as follows:

I need a column and row vectors of the sparse matrix to feed fitellipse.m by fitzgibbon here:
http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/FITZGIBBON/ELLIPSE/

I am able to get the largest connected region with the following:

% Copyright 2009 Naveen Garg (naveen.garg@gmail.com)
% GPL V2 http://www.gnu.org/licenses/gpl-2.0.html

function largest_blob = maxregion(binary_image)
% $$$ return region with largest area for given binary image
[labeled_image, num] = bwlabel(binary_image);
[N, X]=imhist(labeled_image, num + 1);
N(1) = 0;
[max,index_max] = max(N);
largest_blob = (labeled_image == (index_max - 1));

Re: calling octave from C

2009-11-22T13:04:12Z

On Sun, Nov 22, 2009 at 5:54 PM, Judd Storrs <storrsjm@...> wrote:

Sorry, I think it's pretty clear I was being dense. It's pretty
obvious that using pytave is a binary link into octave and should
invoke the full GPL on python scripts that use it. Using pytave is
equivalent to using the .oct interface not m-files. Which should have
been pretty obvious to me. How embarrassing. Sorry for the noise.

--judd

Now I don't know if you're serious :)
Maybe there was some confusion that is to be blamed. You thought (or at least said) that we're adding new restrictions to the GPL in Pytave; we're not. We're just using the GPL, and also respect the FSF's authority regarding specific licensing implications.
And the FSF's FAQ states that if a script uses Pytave, it is "effectively linked to" Octave (hence must comply with GPL).
As a matter of fact, I do *not* really understand how this conclusion follows from the text of GPL 3, I just trust the FAQ. I respect the opinion of FSF because they probably studied GPL-related issues more than most GPL developers, and as creators they should think in accordance with the "spirit" of GPL. But of course even FSF can be wrong.

If the licensing issues pose a problem for your work; I'd suggest you still contact the FSF; maybe they can advise something.

best regards

--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: calling octave from C

2009-11-22T08:54:43Z

Sorry, I think it's pretty clear I was being dense. It's pretty
obvious that using pytave is a binary link into octave and should
invoke the full GPL on python scripts that use it. Using pytave is
equivalent to using the .oct interface not m-files. Which should have
been pretty obvious to me. How embarrassing. Sorry for the noise.

--judd
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: CUDA acceleration in GNU Octave

2009-11-22T08:21:27Z

Well, I've underestimated Nv's plot.
--
Best Regards

Sincerely Yours
Chengqi Chang
-------------------------------
China Center for Economic Research
Peking University, P.R. China
Gtalk/MSN: XaeroChina@...
Personal Website: http://Macro2.cn
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: CUDA acceleration in GNU Octave

2009-11-22T08:02:22Z

Chengqi Chang skrev:

> On Sun, 22 Nov 2009 21:39:44 +0800, John W. Eaton <jwe@...> wrote:
>
>
>> On 22-Nov-2009, Chengqi Chang wrote:
>>
>> | I think it would be great if someone can provide a "bridge" -- like MPI
>> | toolbox(http://atc.ugr.es/javier-bin/mpitb) -- connecting Octave and
>> CUDA
>> | computing libraries. Besides, in my opinion, it matters little whether
>> | this "bridge" is universal or tied to a single vendor (Nv/ATI). Both of
>> | the two choices have merits and drawbacks. If it is tied to a single
>> | vendor, it can be easier to take advantage of some special features of
>> the
>> | GPU and it also has a larger possibility to get donation from that
>> vendor.
>>
>> Are the libraries that take advantage of this single vendor
>> distributed under a license that is compatible with the GPL?
>>
>> If so, then I think this would be acceptable, but not great as it is
>> still ties you to a single vendor.
>>
>> If not, then you would not be able to distribute .oct files that
>> require these libraries, and such an extension would never become a
>> part of the core Octave itself. The Octave project is about software
>> freedom, not about promoting links to proprietary software, or
>> encouraging users to lock themselves into single-vendor solutions.
>>
>> jwe
>>
> GPL compatibility depends on how the libraries use CUDA -- if the libs
> only call them, GPL can be retained.
> Since there are limited numbers of popular vendors, it is possible to
> initialize different projects. Once you buy your graphic card, you are
> actually tied to a vendor.
> BTW, I like the freedom philosophy as well, while I think whether the
> freedom is pure or conditional on some exogenous constraints can be
> equally acceptable.
>
>

So freedom has a price, and it is parallel computing? Come on!

Besides, I think you are underestimating the license clash between
Nvidia's CUDA license and the GNU GPL.

I think OpenCL will be a more vendor-agnostic way to go.

David

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: CUDA acceleration in GNU Octave

2009-11-22T07:50:59Z

On 22-Nov-2009, Chengqi Chang wrote:

| GPL compatibility depends on how the libraries use CUDA -- if the libs
| only call them, GPL can be retained.

I don't think that is correct, nor is it the interpretation of the GPL
that the FSF has.

| Once you buy your graphic card, you are
| actually tied to a vendor.

That does not have to be true. If there were one high-level free
software library that supported multiple types of hardware, you would
not be tied to a single vendor for software support or future hardware
purchases.

jwe
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: CUDA acceleration in GNU Octave

2009-11-22T07:25:04Z

On Sun, 22 Nov 2009 21:39:44 +0800, John W. Eaton <jwe@...> wrote:

> On 22-Nov-2009, Chengqi Chang wrote:
>
> | I think it would be great if someone can provide a "bridge" -- like MPI
> | toolbox(http://atc.ugr.es/javier-bin/mpitb) -- connecting Octave and
> CUDA
> | computing libraries. Besides, in my opinion, it matters little whether
> | this "bridge" is universal or tied to a single vendor (Nv/ATI). Both of
> | the two choices have merits and drawbacks. If it is tied to a single
> | vendor, it can be easier to take advantage of some special features of
> the
> | GPU and it also has a larger possibility to get donation from that
> vendor.
>
> Are the libraries that take advantage of this single vendor
> distributed under a license that is compatible with the GPL?
>
> If so, then I think this would be acceptable, but not great as it is
> still ties you to a single vendor.
>
> If not, then you would not be able to distribute .oct files that
> require these libraries, and such an extension would never become a
> part of the core Octave itself. The Octave project is about software
> freedom, not about promoting links to proprietary software, or
> encouraging users to lock themselves into single-vendor solutions.
>
> jwe

GPL compatibility depends on how the libraries use CUDA -- if the libs
only call them, GPL can be retained.
Since there are limited numbers of popular vendors, it is possible to
initialize different projects. Once you buy your graphic card, you are
actually tied to a vendor.
BTW, I like the freedom philosophy as well, while I think whether the
freedom is pure or conditional on some exogenous constraints can be
equally acceptable.

--
Best Regards!
---------------------------
Sincerely Yours Chengqi Chang
China Center for Economic Research
Mail: XaeroChina@...
Personal Website: http://macro2.cn
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: CUDA acceleration in GNU Octave

2009-11-22T05:39:44Z

On 22-Nov-2009, Chengqi Chang wrote:

| I think it would be great if someone can provide a "bridge" -- like MPI
| toolbox(http://atc.ugr.es/javier-bin/mpitb) -- connecting Octave and CUDA
| computing libraries. Besides, in my opinion, it matters little whether
| this "bridge" is universal or tied to a single vendor (Nv/ATI). Both of
| the two choices have merits and drawbacks. If it is tied to a single
| vendor, it can be easier to take advantage of some special features of the
| GPU and it also has a larger possibility to get donation from that vendor.

Are the libraries that take advantage of this single vendor
distributed under a license that is compatible with the GPL?

If so, then I think this would be acceptable, but not great as it is
still ties you to a single vendor.

If not, then you would not be able to distribute .oct files that
require these libraries, and such an extension would never become a
part of the core Octave itself. The Octave project is about software
freedom, not about promoting links to proprietary software, or
encouraging users to lock themselves into single-vendor solutions.

jwe
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: Passing an array from java to octave

2009-11-22T04:20:20Z

2009/11/2 George Kousiouris <gkousiou@...>:
> Hi all,
>
> I want to pass a Java array to Octave, so that I can process the data
> inside. Does anyone have an idea how this can be done? I have tried a
> java to octave bridge (javaoctave-0.4.0.jar) however this seems to be
> only for executing Octave commands inside a Java program and not
> actually passing data from Java to Octave...

JavaOctave can transfer many different types of data between Java and Octave.

In the following simple example "a" is a 2x2 matrix that is passed
from Java to Octave and "b" is pulled out of Octave.

The example is from:
http://kenai.com/projects/javaoctave/pages/SimpleExampleOfJavaOctaveUsage
==============
final OctaveEngine octave = new OctaveEngineFactory().getScriptEngine();
final OctaveMatrix a = new OctaveMatrix(new double[] { 1, 2, 3, 4 }, 2, 2);
octave.put("a", a);
final String func = "" //
+ "function res = my_func(a)\n" //
+ " res = 2 * a;\n" //
+ "endfunction\n" //
+ "";
octave.eval(func);
octave.eval("b = my_func(a);");
final OctaveMatrix b = octave.get("b");
octave.close();
==============

--
Kim Hansen
Vadgårdsvej 3, 2.tv
2860 Søborg
Fastnet: 3956 2437 -- Mobil: 3091 2437

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: CUDA acceleration in GNU Octave

2009-11-21T23:38:35Z

I think it would be great if someone can provide a "bridge" -- like MPI
toolbox(http://atc.ugr.es/javier-bin/mpitb) -- connecting Octave and CUDA
computing libraries. Besides, in my opinion, it matters little whether
this "bridge" is universal or tied to a single vendor (Nv/ATI). Both of
the two choices have merits and drawbacks. If it is tied to a single
vendor, it can be easier to take advantage of some special features of the
GPU and it also has a larger possibility to get donation from that vendor.

--
Best Regards!
---------------------------
Sincerely Yours Chengqi Chang
China Center for Economic Research
Mail: XaeroChina@...
Personal Website: http://macro2.cn
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: calling octave from C

2009-11-21T21:46:41Z

On Sat, Nov 21, 2009 at 9:53 PM, Judd Storrs <storrsjm@...> wrote:

On Sat, Nov 21, 2009 at 3:26 PM, Jaroslav Hajek <highegg@...> wrote:
> Why?

Because non-GPL scripts are either allowed in both octave and python
or they are forbidden in both octave and python. They are covered by
exactly the same license. The GPL must apply everywhere equally.

Octave and Python are not covered by the same license. Besides, you're getting of the point, because (according to FSF's FAQ) it's not Octave or Python, but Pytave (the "bindings" to external software) that makes the difference. And Pytave is not a part of Python.

Proposing that an interpreter that embeds octave is free to impose new
restrictions on octave is the kind of doublethink you would expect
from a corporate entity trying to circumvent the GPL.

Oh my. So please tell that to FSF, and how they are screwing the license they invented.

Unless I missed a change in octave's licensing, pytave does not have
the authority to alter the license of octave.

--judd

In fact Pytave could add additional terms under section 7, but you have been told that Pytave just uses GPL 3, so I don't understand why you keep saying that.

Unlike its predecessor, GPL 3 does not talk about scripts, interpreters and libraries, not even derivative works. It simply says that a covered work (modified, or based on the source GPL work in question) is anything that would normally require a copyright permission under applicable law.

So ultimately the question is whether a script using Pytave would require copyright permission if Pytave and Octave were proprietary. Apparently, the FSF thinks that the answer is yes.

--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: calling octave from C

2009-11-21T13:10:40Z

Judd Storrs skrev:

> I understand your position. But fundamentally, pytave is clearly a
> derivative work of octave and is bound to octave's license.
> Restrictions on octave by derivative works are incompatible with the
> GPL. That is both the beauty and the curse of the GPL. I guess the
> clearest way to express my consern is that one of these two cases must
> be true:
>
> 1) Pytave's restriction limiting use of octave to only GPL-compatible
> interpreter scripts is a novel GPL-incompatible restriction on
> octave's code
>
>

I think your word "GPL-incompatible" is misleading. Calling it a "novel
GPL-incompatible restriction on Octave's code" is plain wrong. It's not
novel. It's not "GPL-incompatible". Pytave is a derivate of Octave.
That's why it is licensed under GPLv3. And it's not a restriction of
Octave's code. You can do the same thing with Pytave that you can do
with Octave to begin with. In fact, the GPL license is there to
guarantee it.

The GNU GPL is not a random set of rules regarding distribution. I think
this is essential to truly understanding the license. It is there with
the explicit purpose of preserving the four freedoms of the users.
Hypothetically, if Pytave somehow weren't GPL, the hole it with leave
open would be big enough to drive a truck through.

> OR
>
> 2) Use of GPL-incompatible interpreter code (m-files) was already
> forbidden within the octave interpreter.
>
> My opinion is that the FSF FAQ makes it clear that #2 is false.
>
>

The Octave interpreter can be used to interpret any m-file. You can
interpret any m-file through Pytave too. You can run Pytave or Octave
however you wish, it is freedom 0. (Freedom to run the software for any
purpose)

David

_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave

Re: calling octave from C

2009-11-21T12:53:08Z

On Sat, Nov 21, 2009 at 3:26 PM, Jaroslav Hajek <highegg@...> wrote:
> Why?

Because non-GPL scripts are either allowed in both octave and python
or they are forbidden in both octave and python. They are covered by
exactly the same license. The GPL must apply everywhere equally.
Proposing that an interpreter that embeds octave is free to impose new
restrictions on octave is the kind of doublethink you would expect
from a corporate entity trying to circumvent the GPL.

Unless I missed a change in octave's licensing, pytave does not have
the authority to alter the license of octave.

--judd
_______________________________________________
Help-octave mailing list
Help-octave@...
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave