octave-dev — GNU Octave developers' list

[OctDev] Fem PLotting: fpl 1.3.3 released

[Carnë D. <car...@gm...>]

Hi everyone

a new release of fpl package is out, version 1.3.2, by Carlo de Falco.
Enjoy Octave responsibly.

Carnë

Re: [OctDev] [pkg.m] Code sprint

[Sergei S. <ser...@ya...>]

----- Original Message -----
> From: Juan Pablo Carbajal <aju...@gm...>
> To: Octave Forge <oct...@li...>; hel...@oc...
> Cc: 
> Sent: Thursday, November 1, 2012 2:08 PM
> Subject: [pkg.m] Code sprint
> 
> Hi all,
> 
> The code sprint to improve pkg.m will take place on the 17th and 18th
> of November.
> Choose your task here
> http://wiki.octave.org/Code_sprint:_pkg.m
> 
> If you still want to join us you can add yourself here or just pop in
> in the IRC channel those days.
> http://doodle.com/x3bh7ity5knekda8
> 
> See you space coder ...
> _______________________________________________
> Help-octave mailing list
> Hel...@oc...
> https://mailman.cae.wisc.edu/listinfo/help-octave
>

To write code _not_ having a spec is kindergarten nonsense.

As I wrote many times, intents (if I understand them correctly) of 'pkg.m' can _not_ be implemented unless the issue of namespaces is resolved.

I even offered my help: I architect the "thing" (packaging) and you guys implement it per my spec (which we first discuss).

There _must_ be an architect - somebody, anybody, not at all necessarily me.

Regards,
  Sergei.

[OctDev] [pkg.m] Code sprint

[Juan P. C. <aju...@gm...>]

Hi all,

The code sprint to improve pkg.m will take place on the 17th and 18th
of November.
Choose your task here
http://wiki.octave.org/Code_sprint:_pkg.m

If you still want to join us you can add yourself here or just pop in
in the IRC channel those days.
http://doodle.com/x3bh7ity5knekda8

See you space coder ...

Re: [OctDev] density calculation by Octave

[Juan P. C. <aju...@gm...>]

On Tue, Oct 30, 2012 at 3:12 PM, Surandokht Nikzad <sur...@gm...> wrote:
> I am afraid, I am not sure how to code/ compute it...
>
> On 12-10-30 5:08 PM, "Juan Pablo Carbajal" <aju...@gm...> wrote:
>
>>On Tue, Oct 30, 2012 at 3:03 PM, Surandokht Nikzad <sur...@gm...>
>>wrote:
>>> Hi there,
>>>
>>> I am beginner in Octave.
>>> I am using this for calculation kernel density for some concentrations.
>>>
>>> This is one example data:
>>>
>>> ssw ef
>>> 1.85 1.497297297
>>> 1.21 1.719008264
>>> 1.44 1.451388889
>>> 1.71 1.514619883
>>> 1.4         1.192857143
>>> 0.396 1.795454545
>>>
>>> Here is the codes:
>>>
>>> doc_ef=a(:,2);
>>> doc_ssw=a(:,1);
>>> run_doc=0:d:20;
>>> dens_est_doc_ef=kernel_dens(ones(size(run_doc)),doc_ef,run_doc);
>>> plot(run_doc,dens_est_doc_ef,'b')
>>>
>>> The integral of each distribution should be equal to one!My question is
>>>how
>>> can I avoid having plots over one in my graph. The example for graph is
>>> attached by this email.
>>>
>>> Any idea and suggestion would be appreciated.
>>>
>>> Thanks,
>>> Sue
>>>
>>>
>>>-------------------------------------------------------------------------
>>>-----
>>> Everyone hates slow websites. So do we.
>>> Make your web apps faster with AppDynamics
>>> Download AppDynamics Lite for free today:
>>> http://p.sf.net/sfu/appdyn_sfd2d_oct
>>> _______________________________________________
>>> Octave-dev mailing list
>>> Oct...@li...
>>> https://lists.sourceforge.net/lists/listinfo/octave-dev
>>>
>>
>>As you said the integral of probability density function is one, but
>>the density function itself could be bigger then one. Think of the
>>delta distribution.
>>http://en.wikipedia.org/wiki/Delta_distribution
>>
>>If you want, you can plot the normalized frequency plot using hist
>>with the normalization argument equal to 1.
>
>

do
help hist
in your Octave prompt

Re: [OctDev] density calculation by Octave

[Surandokht N. <sur...@gm...>]

I am afraid, I am not sure how to code/ compute it...

On 12-10-30 5:08 PM, "Juan Pablo Carbajal" <aju...@gm...> wrote:

>On Tue, Oct 30, 2012 at 3:03 PM, Surandokht Nikzad <sur...@gm...>
>wrote:
>> Hi there,
>>
>> I am beginner in Octave.
>> I am using this for calculation kernel density for some concentrations.
>>
>> This is one example data:
>>
>> ssw ef
>> 1.85 1.497297297
>> 1.21 1.719008264
>> 1.44 1.451388889
>> 1.71 1.514619883
>> 1.4         1.192857143
>> 0.396 1.795454545
>>
>> Here is the codes:
>>
>> doc_ef=a(:,2);
>> doc_ssw=a(:,1);
>> run_doc=0:d:20;
>> dens_est_doc_ef=kernel_dens(ones(size(run_doc)),doc_ef,run_doc);
>> plot(run_doc,dens_est_doc_ef,'b')
>>
>> The integral of each distribution should be equal to one!My question is
>>how
>> can I avoid having plots over one in my graph. The example for graph is
>> attached by this email.
>>
>> Any idea and suggestion would be appreciated.
>>
>> Thanks,
>> Sue
>>
>> 
>>-------------------------------------------------------------------------
>>-----
>> Everyone hates slow websites. So do we.
>> Make your web apps faster with AppDynamics
>> Download AppDynamics Lite for free today:
>> http://p.sf.net/sfu/appdyn_sfd2d_oct
>> _______________________________________________
>> Octave-dev mailing list
>> Oct...@li...
>> https://lists.sourceforge.net/lists/listinfo/octave-dev
>>
>
>As you said the integral of probability density function is one, but
>the density function itself could be bigger then one. Think of the
>delta distribution.
>http://en.wikipedia.org/wiki/Delta_distribution
>
>If you want, you can plot the normalized frequency plot using hist
>with the normalization argument equal to 1.

Re: [OctDev] density calculation by Octave

[Juan P. C. <aju...@gm...>]

On Tue, Oct 30, 2012 at 3:03 PM, Surandokht Nikzad <sur...@gm...> wrote:
> Hi there,
>
> I am beginner in Octave.
> I am using this for calculation kernel density for some concentrations.
>
> This is one example data:
>
> ssw ef
> 1.85 1.497297297
> 1.21 1.719008264
> 1.44 1.451388889
> 1.71 1.514619883
> 1.4         1.192857143
> 0.396 1.795454545
>
> Here is the codes:
>
> doc_ef=a(:,2);
> doc_ssw=a(:,1);
> run_doc=0:d:20;
> dens_est_doc_ef=kernel_dens(ones(size(run_doc)),doc_ef,run_doc);
> plot(run_doc,dens_est_doc_ef,'b')
>
> The integral of each distribution should be equal to one!My question is how
> can I avoid having plots over one in my graph. The example for graph is
> attached by this email.
>
> Any idea and suggestion would be appreciated.
>
> Thanks,
> Sue
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev
>

As you said the integral of probability density function is one, but
the density function itself could be bigger then one. Think of the
delta distribution.
http://en.wikipedia.org/wiki/Delta_distribution

If you want, you can plot the normalized frequency plot using hist
with the normalization argument equal to 1.

[OctDev] FreeBSD vs simplified BSD

[Carnë D. <car...@gm...>]

Hi

I'm the maintainer of the Octave-Forge project
http://octave.sourceforge.net/ which is a loosely organized collection
of functions for GNU Octave. Because of this, each file has its own
license. I'm trying to organize this properly and facing a problem
with the multiple BSD licenses.

The FreeBSD license seems to exist in 2 forms, with and without a
disclaimer about views and opinions expressed in the software. However
the FSF list of licenses has both of them under the same name:

http://www.gnu.org/licenses/license-list.html#FreeBSD links to
http://directory.fsf.org/wiki?title=License:FreeBSD which displays the
disclaimer
http://www.gnu.org/licenses/license-list.html#FreeBSDDL links to
http://www.freebsd.org/copyright/freebsd-doc-license.html which does
not displays the disclaimer

I was reading about the subject and apparently, the OSI recognized
both of them, calling the version without the disclaimer, simplified
BSD license. In Octave Forge we do have some code under such license
but are planning to adopt
http://www.gnu.org/licenses/license-list.html#GPLCompatibleLicenses as
the list of allowed licenses. I believe this license also belongs to
list of GPL compatible, there's simply no FSF-given name for it. Could
you please update the list and advise a name to avoid confusions?

Thanks in advance,
Carnë Draug

Re: [OctDev] removal of non-standard licenses in Octave Forge

[Juan P. C. <aju...@gm...>]

On Sun, Oct 28, 2012 at 8:13 PM, Carnë Draug <car...@gm...> wrote:
> Hi everyone
>
> we have a couple of files in Octave Forge with non-standard licenses.
> This is bad. Being able to give it a recognisable name not only eases
> our organisation but also its acceptance by downstream package
> maintainers such as Debian. From Debian's upstream guide "Please do
> not write your own license text if you can at all avoid it. Depending
> on your wishes, the GPL, LGPL or a BSD-style license will most likely
> be appropriate, and it is far easier to tell whether something is
> allowed if we can look at past discussions of the same text. "
>
> I believe that in most cases such user-made licenses are not made
> because there's no appropriate license out there, but out of
> indifference for the subject and belief on others better part. For
> example, some functions in the image package have the following
> license "This code has no warrany whatsoever. Do what you like with
> this code as long as you leave this copyright in place" which could
> easily be replaced by something such as the simplified BSD, FreeBSD or
> ICS license which I have already suggested to the original author.
>
> I'd like to propose that we no longer accept such non-standard
> licenses and propose this list
> http://www.gnu.org/licenses/license-list.html#GPLCompatibleLicenses as
> the ones that are acceptable. Does anyone oppose to such change?
>
> Carnë
>
> ------------------------------------------------------------------------------
> WINDOWS 8 is here.
> Millions of people.  Your app in 30 days.
> Visit The Windows 8 Center at Sourceforge for all your go to resources.
> http://windows8center.sourceforge.net/
> join-generation-app-and-make-money-coding-fast/
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev

I subscribe to the initiative.

We could also provide this table (which is far more easy to read)
http://en.wikipedia.org/wiki/List_of_FSF-approved_software_licenses

[OctDev] removal of non-standard licenses in Octave Forge

[Carnë D. <car...@gm...>]

Hi everyone

we have a couple of files in Octave Forge with non-standard licenses.
This is bad. Being able to give it a recognisable name not only eases
our organisation but also its acceptance by downstream package
maintainers such as Debian. From Debian's upstream guide "Please do
not write your own license text if you can at all avoid it. Depending
on your wishes, the GPL, LGPL or a BSD-style license will most likely
be appropriate, and it is far easier to tell whether something is
allowed if we can look at past discussions of the same text. "

I believe that in most cases such user-made licenses are not made
because there's no appropriate license out there, but out of
indifference for the subject and belief on others better part. For
example, some functions in the image package have the following
license "This code has no warrany whatsoever. Do what you like with
this code as long as you leave this copyright in place" which could
easily be replaced by something such as the simplified BSD, FreeBSD or
ICS license which I have already suggested to the original author.

I'd like to propose that we no longer accept such non-standard
licenses and propose this list
http://www.gnu.org/licenses/license-list.html#GPLCompatibleLicenses as
the ones that are acceptable. Does anyone oppose to such change?

Carnë

Re: [OctDev] Bundles in Agora

[Juan P. C. <aju...@gm...>]

On Sun, Oct 28, 2012 at 8:15 AM, c. <car...@gm...> wrote:
>
> On 28 Oct 2012, at 08:01, Juan Pablo Carbajal wrote:
>
>> Carlo,
>> If you were commenting "3) Browsing bundle contents", her eis my opinion.
>>
>> I do not think being able to see the contents of the file means we are
>> replicating a VCS.
>>
>> 1- The functionality is already there.
>> 2- We will not allow edition of the bundle once it is uploaded,
>> nevertheless you can see the content of the files.
>> 3- Form a user (not developer perspective): I want to look at some
>> code in this package that seems to be what I need, before I download
>> it.
>>
>> Cheers
>
> Yes if editing is ruled out, being able to view the package contents can do no harm.
>
> My main concern is that making the site bloated by adding more features than actually
> needed can make the thing overcomplicated and difficult to maintain.
>
> c.
>
>
> P.S. Was this message private on purpose? I replied privatly in case your idea was to not create confusion in the public discussion.
>

Carlo,

Not it was meant to be public. :D

Thanks for your discretion.

Re: [OctDev] no more combinatorics and physical-constants package

[Carnë D. <car...@gm...>]

On 17 October 2012 13:07, Rafael Laboissiere <ra...@la...> wrote:
> * c. <car...@gm...> [2012-10-17 09:23]:
>>
>> On 16 Oct 2012, at 02:21, Carnë Draug wrote:
>>
>>> I have removed the combinatorics and physical-constants packages.
>>>
>>> They had been moved into the miscellaneous package some months ago and
>>> since a new release of it was made, I have also released these two as
>>> well. They are empty packages, only have a dummy file and are
>>> dependent on the new version of miscellaneous just to ease transition.
>>
>> I do agree with the move of physical_constant into the miscellaneous package
>> as it reduces the total number of packages.
>
> Another obvious candidate for removal is the special-matrix package,
> which contains currently only two files (arecibo.m and lauchli.m). They
> could also be moved into miscellaneous.

You're right. But I find them too miscellaneous even for the
miscellaneous package. I'd rather prepare a special-functions-legacy
package for them, now that agora is up and running.

Carnë

Re: [OctDev] Disable warnings

[Carnë D. <car...@gm...>]

Hi

you have come to the octave forge mailing list. Your query belongs to
octave core mailing list. CC'ing it.

On 28 October 2012 01:00,  <fe...@st...> wrote:
> Hello,
>
> There are a couple of warnings I want to desable without having to issue
> the general command
>
> warning('off','all');
>
> Is there a list where I can see what the string identifications of the
> following warnings?
>
>
> warning: load: file found in load path
>
> warning: quadgk: maximum interval count (650) met
> warning: quadgk: Error tolerance not met. Estimated error 2.7601e-12
>
>
> Thank you very much in advance!
>
> Fabián

I don't think that is possible. These warnings are in quadgk.m and
have no ID. I'm guessing that if you really want to, you can do it in
your system only by editing the source and add one yourself. Then you
could disable them.

Of you can turn all warnings locally by using a unwind protect block:

warn_state = warning ("query", "all");
unwind_protect
  warning ("off", "all");
  ## your code goes here
unwind_protect_cleanup
  warning (warn_state);
end_unwind_protect


Carnë

[OctDev] Disable warnings

[<fe...@st...>]

Hello,

There are a couple of warnings I want to desable without having to issue
the general command

warning('off','all');

Is there a list where I can see what the string identifications of the
following warnings?


warning: load: file found in load path

warning: quadgk: maximum interval count (650) met
warning: quadgk: Error tolerance not met. Estimated error 2.7601e-12


Thank you very much in advance!

Fabián

Re: [OctDev] complex error function

[Steven G. J. <st...@al...>]

Juan Pablo Carbajal wrote:
> You can upload your function to the feature request forum
> https://sourceforge.net/p/octave/feature-requests/
> 
> Then we can look at your code and suggest improvements, etc..., if needed.

I've created a ticket:

	https://sourceforge.net/p/octave/feature-requests/42/

I didn't attach the code but instead gave the link, mainly because I'll 
probably be posting an updated version of the code shortly with some 
accuracy improvements.

--SGJ

Re: [OctDev] finishing up new release of image package

[Carnë D. <car...@gm...>]

On 26 October 2012 20:14, Lukas Reichlin <luk...@gm...> wrote:
> On 26.10.2012, at 20:37, Jordi Gutiérrez Hermoso <jo...@oc...> wrote:
>
>> On 26 October 2012 14:16, Lukas Reichlin <luk...@gm...> wrote:
>>> On 26.10.2012, at 20:01, Carnë Draug <car...@gm...> wrote:
>>
>>>> A new release candidate
>>>> http://carandraug.no-ip.org/octave/image-1.9.95.tar.gz
>>>>
>>>> This time should really really install with clang. Please let me
>>>> know of any problems.
>>>
>>> Is your system up and running? I'm unable to connect.
>>
>> Try this:
>>
>>    http://jordi.platinum.linux.pl/octave/image-1.9.95.tar.gz
>>
>> - Jordi G. H.
>
>
> Your link worked:
>
> octave:1> pkg install image*
> In file included from bwlabeln.cc:20:
> In file included from /sw/include/octave-3.6.3/octave/oct.h:36:
> In file included from /sw/include/octave-3.6.3/octave/defun-dld.h:30:
> In file included from /sw/include/octave-3.6.3/octave/defun-int.h:28:
> In file included from /sw/include/octave-3.6.3/octave/ov-builtin.h:28:
> In file included from /sw/include/octave-3.6.3/octave/ov-fcn.h:35:
> In file included from /sw/include/octave-3.6.3/octave/symtab.h:34:
> /sw/include/octave-3.6.3/octave/regexp.h:90:3: warning: 'opts' defined as a struct here but previously declared as a
>       class [-Wmismatched-tags]
>   struct opts
>   ^
> /sw/include/octave-3.6.3/octave/regexp.h:40:3: note: did you mean struct here?
>   class opts;
>   ^~~~~
>   struct
> 1 warning generated.
> For information about changes from previous versions of the image package, run 'news ("image")'.
> octave:2>
>
> Regards,
> Lukas

Thanks for testing Lukas. Good to see it working. That warning comes
from Octave core, there's nothing I can do about it and should already
been fixed in dev.

If no one reports any problem with this. I'll make the release tomorrow.

Carnë

Re: [OctDev] [bug #37616] holidays result incorrect when New Year's Day falls on Saturday

[Carnë D. <car...@gm...>]

On 27 October 2012 16:26,  <lit...@ya...> wrote:
> On 27 October 2012 14:46, Carnë Draug <car...@gm...> wrote:
>> On 26 October 2012 15:24,  <lit...@ya...> wrote:
>>> Hi Carnë,
>>>
>>> After more careful inspection, it appears that when New Years Day falls
>>> on a Saturday, the holiday is not observed at all. The holidays function
>>> does not check that, so modifying line 41 will fix that (and also allow
>>> holidays to pass its test). So please disregard the previous code change
>>> suggestion, and consider the following instead.
>>>
>>> from holidays.m:
>>> ------------------------------------------
>>>  ## New Year's Day
>>>  tmphol = datenum (yrs, 1, 1);
>>>  hol = [hol; tmphol(weekday(tmphol(:)) ~= 7)(:)];
>>> ------------------------------------------
>>> I modified the last line from
>>>  hol = [hol; tmphol(:)];
>>>
>>> Thanks for the link. I will submit octave-forge bugs there in the future.
>>> Should I open up a new thread there for this bug, or is that unnecessary?
>>>
>>> -Randy Chamberlin
>>
>> Hi Randy
>>
>> not necessary this time. I have commited your changes see
>> https://sourceforge.net/p/octave/code/11368
>
> I was taking a closer look at the function and noticed the block near
> the end that is meant to adjust for sundays and saturdays.
>
>   wd = weekday (hol);
>   if any (wd == 1)
>     hol(wd == 1) = hol(wd == 1) + 1;
>   endif
>   if any (wd == 7)
>     hol(wd == 7) = hol(wd == 7) - 1;
>   endif
>
> I believe that a better fix should go in here.
>
> Carnë
> ----------------------
>
> Hi Carne,
>
> I think the problem is that according to NYSE rules, holidays are not
> allowed to span across a calendar year.
>
> From the NYSE rules at
> http://nyserules.nyse.com/nysetools/PlatformViewer.asp?SelectedNode=chp_1_3&manual=/nyse/rules/nyse-rules/
>
> "The Exchange Board has also determined that, when any holiday observed by
> the Exchange falls on a Saturday, the Exchange will not be open for business
> on the preceding Friday and when any holiday observed by the Exchange falls
> on a Sunday, the Exchange will not be open for business on the succeeding
> Monday, unless unusual business conditions exist, such as the ending of a
> monthly or the yearly accounting period."
>
> So in general, Saturday holidays should be shifted to the previous Friday,
> and Sunday holidays to the next Monday, except when it crosses a calendar
> year. So I think the original logic (pre my patch) is correct, with the
> exception that it does not attempt to account for the case where New Years
> Day falls on a Saturday and crosses a yearly accounting period. In that
> case, the NYSE rules say it should be ignored, but the original logic alone
> doesn't do that. If you check the link posted in the help file for holidays
> (http://www.chronos-st.org/NYSE_Observed_Holidays-1885-Present.html), you
> can see that Saturday, Jan 1, 2005 was not observed. Since New Years Day is
> the only holiday that can cross "yearly account periods", the check is only
> required for that one holiday, and so I would argue the previous fix (line
> 41) is best for minimizing processor time. If you think it would be easier
> to read by putting a change in the weekday pruning block, I believe the
> following, added at line 77 would also do it:
>
> ## Delete any New Years Day holidays that cross calendar years
>   hol(day(hol) == 31 & month(hol) == 12) = [];
>
> This must be added after the adjustment for Saturday..
>
> -Randy Chamberlin

Hi Randy

I think I prefer your previous fix. Anyway, for matlab compatibility,
I have added a bunch of extra "holidays" due to closing for special
occasions. This means hardcoding all dates between 1885 and 2050. The
previous code (with your fix), is used for dates outside that range. I
have also simplified the original code by removing temporary variables
(Octave can directly index output of functions).

See https://sourceforge.net/p/octave/code/11371/tree/trunk/octave-forge/main/financial/inst/holidays.m
and please give it a try.

Carnë

Re: [OctDev] complex error function

[Steven G. J. <st...@al...>]

Juan Pablo Carbajal wrote:
> Then we can look at your code and suggest improvements, etc..., if needed.
> 
> In which field is your function used the most?

The complex error function (generalizing erf, erfc, and erfx) is used in 
lots of fields.  We needed it to solve electromagnetic scattering 
problems via integral-equation methods, where it appears in Ewald 
summations of Green's functions in periodic domains.  It is also used to 
compute the Voigt functions in spectroscopy 
(http://en.wikipedia.org/wiki/Voigt_profile), the dispersion relation in 
plasmas 
(http://farside.ph.utexas.edu/teaching/plasma/lectures/node87.html), 
diffusion problems, and probably in other areas I don't know about.

Steven

Re: [OctDev] complex error function

[Steven G. J. <st...@al...>]

Jordi Gutiérrez Hermoso wrote:
> On 24 October 2012 23:34, Steven G. Johnson <st...@al...> wrote:
>> Hi all, in case there is interest I wanted to let you know that I've
>> written a free complex-argument error function in C++ with an Octave
>> wrapper at:
>>
>>         http://ab-initio.mit.edu/wiki/index.php/Faddeeva_w
> 
> Thanks. This belongs in core Octave since that's where the current
> implementation of erf and friends is. I would need to modify your
> source to fit our style. Besides stylistic issues, there's also the
> issue that it should not be a dynamically loaded function, but
> built-in, so your wrapper needs to be rewritten. I am happy to do this
> myself. Do you have a DVCS somewhere, preferrably Mercurial, where we
> could keep your style on one branch and ours in another? Hopefully you
> plan to keep working on this function, so it's worthwhile to version
> your code. If you don't already have a repository set up, I would be
> most pleased to get you started with Mercurial.

I have a private git repository, but I can't make it public because it 
contains a couple of semi-free programs for performance and accuracy 
comparisons.  It wouldn't be hard for me to check in my changes to a 
public Mercurial repository as well, however, if that is much easier for 
you than just grabbing the file from my web site periodically.

I do plan to keep updating my code as needed (e.g. right now I am 
tweaking it to compute Re[w] more accurately in regions close to the 
real-z axis where Re[w] is small), so it would be nice if you could 
track my changes.

> Do you recommend replacing the current implementation of erf for real
> arguments with your implementation? Currently for real arguments we
> rely on the C library's math.h implementation, and for complex
> arguments we error out.

Probably it is better to stick with the libc erf and erfc for real 
arguments.  What my code computes is the scaled error function w(z), 
equivalent to erfcx(x) = exp(x^2) erfc(x) = w(ix) on the real axis. 
This means that my code has to work over a vastly greater domain, even 
on the real axis (where both erf and erfc are trivial except for small 
|x|), with a corresponding sacrifice of performance (a factor of 2-10 on 
my machine).

You might use my code to provide an erfcx function even for real 
arguments, however.  There are also a few other important special 
functions that can be computed from w(z), like the Voigt functions and 
Dawson's integral.

--SGJ

Re: [OctDev] [bug #37616] holidays result incorrect when New Year's Day falls on Saturday

[<lit...@ya...>]

________________________________
 From: Carnë Draug <car...@gm...>
To: lit...@ya... 
Cc: Octave Forge <oct...@li...> 
Sent: Saturday, October 27, 2012 8:50 AM
Subject: Re: [OctDev] [bug #37616] holidays result incorrect when New Year's Day falls on Saturday
 
On 27 October 2012 14:46, Carnë Draug <car...@gm...> wrote:
> On 26 October 2012 15:24,  <lit...@ya...> wrote:
>> Hi Carnë,
>>
>> After more careful inspection, it appears that when New Years Day falls on a Saturday, the holiday is not observed at all. The holidays function does not check that, so modifying line 41 will fix that (and also allow holidays to pass its test). So please disregard the previous code change suggestion, and consider the following instead.
>>
>> from holidays.m:
>> ------------------------------------------
>>   ## New Year's Day
>>   tmphol = datenum (yrs, 1, 1);
>>   hol = [hol; tmphol(weekday(tmphol(:)) ~= 7)(:)];
>> ------------------------------------------
>> I modified the last line from
>>   hol = [hol; tmphol(:)];
>>
>> Thanks for the link. I will submit octave-forge bugs there in the future. Should I open up a new thread there for this bug, or is that unnecessary?
>>
>> -Randy Chamberlin
>
> Hi Randy
>
> not necessary this time. I have commited your changes see
> https://sourceforge.net/p/octave/code/11368

I was taking a closer look at the function and noticed the block near
the end that is meant to adjust for sundays and saturdays.

  wd = weekday (hol);
  if any (wd == 1)
    hol(wd == 1) = hol(wd == 1) + 1;
  endif
  if any (wd == 7)
    hol(wd == 7) = hol(wd == 7) - 1;
  endif

I believe that a better fix should go in here.

Carnë
----------------------

Hi Carne,

I think the problem is that according to NYSE rules, holidays are not allowed to span across a calendar year.

From the NYSE rules at http://nyserules.nyse.com/nysetools/PlatformViewer.asp?SelectedNode=chp_1_3&manual=/nyse/rules/nyse-rules/
 
"The Exchange Board has also determined that, when any holiday observed 
by the Exchange falls on a Saturday, the Exchange will not be open for 
business on the preceding Friday and when any holiday observed by the 
Exchange falls on a Sunday, the Exchange will not be open for business 
on the succeeding Monday, unless unusual business conditions exist, such
 as the ending of a monthly or the yearly accounting period."

So in general, Saturday holidays should be shifted to the previous Friday, and Sunday holidays to the next Monday, except when it crosses a calendar year. So I think the original logic (pre my patch) is correct, with the exception that it does not attempt to account for the case where New Years Day falls on a Saturday and crosses a yearly accounting period. In that case, the NYSE rules say it should be ignored, but the original logic alone doesn't do that. If you check the link posted in the help file for holidays (http://www.chronos-st.org/NYSE_Observed_Holidays-1885-Present.html), you can see that Saturday, Jan 1, 2005 was not observed. Since New Years Day is the only holiday that can cross "yearly account periods", the check is only required for that one holiday, and so I would argue the previous fix (line 41) is best for minimizing processor time. If you think it would be easier to read by putting a change in the weekday pruning block, I believe the
 following, added at line 77 would also do it:
 
## Delete any New Years Day holidays that cross calendar years
  hol(day(hol) == 31 & month(hol) == 12) = []; 

This must be added after the adjustment for Saturday..


-Randy Chamberlin

Re: [OctDev] [bug #37616] holidays result incorrect when New Year's Day falls on Saturday

[Carnë D. <car...@gm...>]

On 27 October 2012 14:46, Carnë Draug <car...@gm...> wrote:
> On 26 October 2012 15:24,  <lit...@ya...> wrote:
>> Hi Carnë,
>>
>> After more careful inspection, it appears that when New Years Day falls on a Saturday, the holiday is not observed at all. The holidays function does not check that, so modifying line 41 will fix that (and also allow holidays to pass its test). So please disregard the previous code change suggestion, and consider the following instead.
>>
>> from holidays.m:
>> ------------------------------------------
>>   ## New Year's Day
>>   tmphol = datenum (yrs, 1, 1);
>>   hol = [hol; tmphol(weekday(tmphol(:)) ~= 7)(:)];
>> ------------------------------------------
>> I modified the last line from
>>   hol = [hol; tmphol(:)];
>>
>> Thanks for the link. I will submit octave-forge bugs there in the future. Should I open up a new thread there for this bug, or is that unnecessary?
>>
>> -Randy Chamberlin
>
> Hi Randy
>
> not necessary this time. I have commited your changes see
> https://sourceforge.net/p/octave/code/11368

I was taking a closer look at the function and noticed the block near
the end that is meant to adjust for sundays and saturdays.

  wd = weekday (hol);
  if any (wd == 1)
    hol(wd == 1) = hol(wd == 1) + 1;
  endif
  if any (wd == 7)
    hol(wd == 7) = hol(wd == 7) - 1;
  endif

I believe that a better fix should go in here.

Carnë

Re: [OctDev] [bug #37616] holidays result incorrect when New Year's Day falls on Saturday

[Carnë D. <car...@gm...>]

On 26 October 2012 15:24,  <lit...@ya...> wrote:
> Hi Carnë,
>
> After more careful inspection, it appears that when New Years Day falls on a Saturday, the holiday is not observed at all. The holidays function does not check that, so modifying line 41 will fix that (and also allow holidays to pass its test). So please disregard the previous code change suggestion, and consider the following instead.
>
> from holidays.m:
> ------------------------------------------
>   ## New Year's Day
>   tmphol = datenum (yrs, 1, 1);
>   hol = [hol; tmphol(weekday(tmphol(:)) ~= 7)(:)];
> ------------------------------------------
> I modified the last line from
>   hol = [hol; tmphol(:)];
>
> Thanks for the link. I will submit octave-forge bugs there in the future. Should I open up a new thread there for this bug, or is that unnecessary?
>
> -Randy Chamberlin

Hi Randy

not necessary this time. I have commited your changes see
https://sourceforge.net/p/octave/code/11368

Thank you,
Carnë

Re: [OctDev] new package releases: bim 1.1.0 and msh 1.0.5

[c. <car...@gm...>]

On 27 Oct 2012, at 10:36, Juan Pablo Carbajal wrote:

> The warning is just telling the user that broadcast was applied.
> Whether Carlo wants that or not I do not know.
It is expected and actually gives a great improvement in performance in the specific case.

> @Carlo: In the latest Octave you will be able to deactivate the
> warnings locally to each function.
I know, I am working with the development version, but I was not sure it would work with the 3.6 release

c.

Re: [OctDev] new package releases: bim 1.1.0 and msh 1.0.5

[c. <car...@gm...>]

On 27 Oct 2012, at 10:31, Rafael Laboissiere wrote:

> Well, when testing functions bim3a_osc_advection_diffusion and 
> bim3a_osc_laplacian, I get lots of this message:
> 
>     warning: product: automatic broadcasting operation applied
> 
> I guess they are harmless.
> 
> Rafael

Yes, they are.
c.

Re: [OctDev] new package releases: bim 1.1.0 and msh 1.0.5

[Juan P. C. <aju...@gm...>]

On Sat, Oct 27, 2012 at 10:31 AM, Rafael Laboissiere
<ra...@la...> wrote:
> * Rafael Laboissiere <ra...@la...> [2012-10-26 13:43]:
>
>> * c. <car...@gm...> [2012-10-26 01:02]:
>>>
>>> I posted a new version of bim here:
>>>
>>> https://sourceforge.net/p/octave/package-releases/9/#5b7d
>>
>> This tarball works fine, thanks.
>
> Well, when testing functions bim3a_osc_advection_diffusion and
> bim3a_osc_laplacian, I get lots of this message:
>
>      warning: product: automatic broadcasting operation applied
>
> I guess they are harmless.
>
> Rafael
>
>
> ------------------------------------------------------------------------------
> WINDOWS 8 is here.
> Millions of people.  Your app in 30 days.
> Visit The Windows 8 Center at Sourceforge for all your go to resources.
> http://windows8center.sourceforge.net/
> join-generation-app-and-make-money-coding-fast/
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev

The warning is just telling the user that broadcast was applied.
Whether Carlo wants that or not I do not know.
@Carlo: In the latest Octave you will be able to deactivate the
warnings locally to each function.

Re: [OctDev] new package releases: bim 1.1.0 and msh 1.0.5

[Rafael L. <ra...@la...>]

* Rafael Laboissiere <ra...@la...> [2012-10-26 13:43]:

> * c. <car...@gm...> [2012-10-26 01:02]:
>>
>> I posted a new version of bim here:
>>
>> https://sourceforge.net/p/octave/package-releases/9/#5b7d
>
> This tarball works fine, thanks.

Well, when testing functions bim3a_osc_advection_diffusion and 
bim3a_osc_laplacian, I get lots of this message:

     warning: product: automatic broadcasting operation applied

I guess they are harmless.

Rafael