1 Aug 2009 04:05

### underflow is overflow

M:makelist(block([fpprec:500,s,t],s:bfloat(5.5b0*i*10^-11),
t:f(5.5b0 * i * 10^-11,
2.749016986206715556222694148621332735422449427897640242607126584059308591366393146096574751526649826b-18,
2, 10),                         fpprec:10, bfloat([s,t])),i, -60*3, 60*3)\$

(%i89) draw2d(points_joined=true,color=blue,point_type=dot,points(M));

float: floating point overflow converting 0.0b0
-- an error.  To debug this try debugmode(true);

I had fpprec set to 500 while computing y = f(x)  and list M [[x1,y1],[x2,y2],[x3,y3],...] is using
fpprec=10.  Is this a bug?  This
is actually a case of underflow and I would be happy with 0b0 instead of the error.

Rich
1 Aug 2009 06:47

### Re: Repeated convolution of a continous uniform distribution

My program pw.mac can do this problem symbolically.

first get pw.mac from my site.

http://mysite.verizon.net/res11w2yb/id2.html

then try

(showtime:true);
Evaluation took 0.0000 seconds (0.0000 elapsed)
(out7) true
convolution(f, g, x) := block([s,p], f:subst(x = p, f), g:subst(x = x - p, g), pwdefint(f * g, p, minf, inf))\$
Evaluation took 0.0000 seconds (0.0000 elapsed)
define(boxcar(x), unit_step(x+a)-unit_step(x-a));
Evaluation took 0.0000 seconds (0.0000 elapsed)
(out9) boxcar(x):=unit_step(x+a)-unit_step(x-a)
f1:convolution(boxcar(x),boxcar(x),x);
Evaluation took 0.0800 seconds (0.0800 elapsed)
(out10) ((x+2*a)*signum(x+2*a)+(x-2*a)*signum(x-2*a)-2*x*signum(x))/2
f2:convolution(f1,f1,x)\$
Evaluation took 0.2000 seconds (0.2000 elapsed)
f3:convolution(f2,f2,x)\$
Evaluation took 4.0400 seconds (4.0400 elapsed)
f4:convolution(f3,f3,x)\$
Evaluation took 477.4300 seconds (477.4300 elapsed)
draw2d(
color=black,
explicit(f1,x,-5,5),
color=blue,
explicit(f2,x,-5,5),

1 Aug 2009 08:12

### [Off Topic] Suggest any Mailing Lists or Discussion Board for Mathematics?

I am reading the /ordinary differential equations /by V.I. Arnold. Now,
I encountered some difficulties to understand some paragraphs. After
examining what's the difficulties are, I consider the difficulties
should originate from the lacking of some background knowledge or
concepts. The problem is that I have no idea to choose any suitable
stuff to compensate the lack.

_______________________________________________
Maxima mailing list
Maxima <at> math.utexas.edu
http://www.math.utexas.edu/mailman/listinfo/maxima
1 Aug 2009 18:44

### Re: underflow is overflow

Looks like a bug to me. A shorter example:

(%i1) fpprec:500;
(%o1)                          500
(%i2) float(0b0);
float: floating point overflow converting 0.0b0

Aside from bigfloat 0, I think float should return 0.0 for bigfloats
which are smaller than the smallest float (instead of triggering
an underflow error).

Thanks for this report. It would be a good idea to make a
bug report in the tracker too so we don't lose it.

Robert Dodier
1 Aug 2009 19:50

### eeePC 1000 (xandros)

Hallo,

asus eeePC 1000 (Linux/xandos)?
Which repository?
problems? solutions?

mit freundlichen Grüßen
Friedrich Hattendorf

---
Lernen ist wie das Rudern gegen den Strom;
sobald man aufhört, treibt man zurück
Benjamin Britten
1 Aug 2009 19:52

### Maxima 5.19 release branch to occur Aug 2 or 3

Hi,

I'm planning to make a release branch for Maxima 5.19 in a day or two.

HTH

Robert Dodier
1 Aug 2009 20:29

### Re: eeePC 1000 (xandros)

I've got it running nicely on the EEEpc 701, but I'm also doing Debian
unstable at the same time.  No problems (assuming you know what you
want and how to do it with Debian).   I assume the 1000 would be
simpler, you don't have the 4Gb disk limitation :P).

2009/8/1 Friedrich Hattendorf <friedrich <at> hattendoerfer.de>:
> Hallo,
>
> asus eeePC 1000 (Linux/xandos)?
> Which repository?
> problems? solutions?
>
> mit freundlichen Grüßen
> Friedrich Hattendorf
>
> ---
> Lernen ist wie das Rudern gegen den Strom;
> sobald man aufhört, treibt man zurück
> Benjamin Britten
>
> _______________________________________________
> Maxima mailing list
> Maxima <at> math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>

--

--
best,
-tony

2 Aug 2009 00:01

### Re: eeePC 1000 (xandros)

On Sat, 1 Aug 2009, Friedrich Hattendorf wrote:

< Hallo,
<
< asus eeePC 1000 (Linux/xandos)?
< Which repository?
< problems? solutions?

I run debian lenny (stable) on an asus eee904hd. The debian stable
maxima package is very out-of-date (q: why is this the case?).

I installed clisp and cmucl from the repository and built maxima from
the latest release on sourceforge. (I've also built gcl 2.6.8pre from
source, but this is a bit more difficult.)

Leo

--

--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
2 Aug 2009 06:57

### Re: eeePC 1000 (xandros)

At Sat, 1 Aug 2009 23:01:37 +0100 (BST),
Leo Butler wrote:
>
> I run debian lenny (stable) on an asus eee904hd. The debian stable
> maxima package is very out-of-date (q: why is this the case?).
>
> I installed clisp and cmucl from the repository and built maxima from
> the latest release on sourceforge. (I've also built gcl 2.6.8pre from
> source, but this is a bit more difficult.)

I recommend Debian too.

http://wiki.debian.org/DebianEeePC

I think Stable is intended to be used on servers.
Testing is OK for desktops and laptops.

(Anyway, it is not hard to build a CL
implementation and Maxima.)

--

--
Yours sincerely,
2 Aug 2009 07:03

### Re: eeePC 1000 (xandros)

At Sat, 1 Aug 2009 23:01:37 +0100 (BST),
Leo Butler wrote:
>
> I run debian lenny (stable) on an asus eee904hd. The debian stable
> maxima package is very out-of-date (q: why is this the case?).

I checked http://packages.debian.net/.

There is 5.17 in Debian Testing (while Stable contains 5.13).

It is better to updrade to the Testing.

--

--
Yours sincerely,