You can subscribe to this list here.
2014 
_{Jan}

_{Feb}
(232) 
_{Mar}
(323) 
_{Apr}
(383) 
_{May}
(359) 
_{Jun}
(435) 
_{Jul}
(252) 
_{Aug}
(172) 
_{Sep}
(265) 
_{Oct}
(263) 
_{Nov}
(350) 
_{Dec}
(359) 

2015 
_{Jan}
(267) 
_{Feb}
(220) 
_{Mar}
(311) 
_{Apr}
(269) 
_{May}
(388) 
_{Jun}
(403) 
_{Jul}
(172) 
_{Aug}
(399) 
_{Sep}
(364) 
_{Oct}
(18) 
_{Nov}

_{Dec}

From: David Scherfgen <d.scherfgen@go...>  20150418 10:21:30

I wasn't suggesting that logabs should be turned on by default. My point was that if the option is provided, then it should behave consistently and give the same answer for sin(x)/cos(x) and tan(x), etc. Further, the Maxima manual entry for logabs says: > For definite integration, the logabs:true setting is used, because here > "evaluation" of the indefinite integral at the endpoints is often needed. Shouldn't it be possible then to find an example where, for some limits a, b, something like integrate(sin(x)/cos(x),x,a,b) would give a different answer than integrate(tan(x),x,a,b)? Best regards, David 20150418 8:29 GMT+02:00 Gunter Königsmann <gunter@...>: > Correctness is important. But sometimes I wonder if there is a way to make > maxima tell the user that it refuses to do something that *looks* obvious > since it is fundamentally incorrect. Let the computer help with at least > part of the teaching? I assume there is no possibility to do so, though. > And if there were any it would probably mean loads of bogus warnings. > > Kind regards, > > Gunter. > > Am 17. April 2015 23:49:59 MESZ, schrieb Raymond Toy < > toy.raymond@...>: >> >> "Barton" == Barton Willis <willisb@...> writes: >>>>>>> >>>>>> >> Barton> Telling calculus students that integrate(1/x,x) = >> Barton> log(abs(x)) is somewhat slothful. If we didn't conflate >> Barton> functions and their formula, maybe it would be a better >> Barton> world: >> >> >> Barton> integrate(x in (inf,0) > 1/x) = x in (inf 0) > log(x), (1) >> >> Barton> integrate(x in (0,inf) > 1/x) = x in (0, inf) > log(x), (2) >> >> Barton> integrate(x in [1,1]) > 1/x) = bogus (3) >> >> >> >> Barton> Surely a student should look at the limits of integration >> Barton> and choose between (1), (2), or (3). The logabs issue >> Barton> comes up every so often. Users want answers that match US >> Barton> calculus textbooksI can understand that point of >> Barton> view. But it does eliminate a teachable moment. >> >> I'm sure many people disagree with me, but I've always felt maxima >> should default to correctness, even if that means those high school >> and college students don't get the answer they expect. (Because >> they've been taught the wrong thing or because they don't remember.) >> >> I guess a corollary is not catering to beginners too much; you're only >> a beginner once, so let's educate them so they're not beginners >> anymore. >> >>  >> Ray >> >> >>  >> >> BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT >> Develop your own process in accordance with the BPMN 2 standard >> Learn Process modeling best practices with Bonita BPM through live exercises >> http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual event?utm_ >> source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF >>  >> >> Maximadiscuss mailing list >> Maximadiscuss@... >> https://lists.sourceforge.net/lists/listinfo/maximadiscuss >> >> >  > Diese Nachricht wurde von meinem Mobiltelefon mit Kaiten Mail gesendet. > > >  > BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT > Develop your own process in accordance with the BPMN 2 standard > Learn Process modeling best practices with Bonita BPM through live > exercises > http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual > event?utm_ > source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss > > 
From: David Scherfgen <d.scherfgen@go...>  20150418 10:15:32

Hi, in the current Git version, the following makes Maxima crash: integrate(cot(x),x,0,inf); The full output: Maxima branch_5_36_base_24_g18abe6c http://maxima.sourceforge.net using Lisp SBCL 1.0.55.0.debian Distributed under the GNU Public License. See the file COPYING. Dedicated to the memory of William Schelter. The function bug_report() provides bug reporting information. (%i1) integrate(cot(x),x,0,inf); INFO: Control stack guard page unprotected Control stack guard page temporarily disabled: proceed with caution Maxima encountered a Lisp error: Control stack exhausted (no more space for function call frames). This is probably due to heavily nested or infinitely recursive function calls, or a tail call that SBCL cannot or has not optimized away. PROCEED WITH CAUTION. Automatically continuing. To enable the Lisp debugger set *debuggerhook* to nil. Older Maxima versions give the correct output: (%i1) integrate(cot(x),x,0,inf); defint: integral is divergent.  an error. To debug this try: debugmode(true); Best regards, David 
From: Gunter Königsmann <gunter@pe...>  20150418 08:28:50

We should put this installer into the download section on sourceforge.net, too. Kind regards, Gunter. Am 18. April 2015 10:19:42 MESZ, schrieb Wolfgang Dautermann <wolfgang@...>: >Am 16.04.2015 um 23:07 schrieb Robert Dodier: >> What is the result of: >> >> (encodeuniversaltime 0 0 0 1 1 1900 0) >> >> What about: >> >> (decodeuniversaltime 0 0) > >Maxima 5.36.0 http://maxima.sourceforge.net >using Lisp CLISP 2.49 (20100707) >Distributed under the GNU Public License. See the file COPYING. >Dedicated to the memory of William Schelter. >The function bug_report() provides bug reporting information. >(%i1) :lisp (encodeuniversaltime 0 0 0 1 1 1900 0) >0 >(%i1) :lisp (decodeuniversaltime 0 0) >0 >0 >0 >1 >1 >1900 >0 >NIL >0 >(%i1) > >I assume, that is the expected result? > >Anyway, if someone is interested in the Clisp based crosscompiled >installer (or if it helps solving the issue)  I have put it on my >website: >http://wolfgang.dautermann.at/maxima/ > >(of course with the build scripts which I used for generating it). > >The previous version (5.35.1) is still available. I do not want to >keep a complete history of maxima windows installers on my website (as >on sourceforge) but will provide the last two or three versions. > >Sorry for the delay. I am rather busy now organizing a local Linux >event ("Grazer Linuxtage"), which will be on 24th and 25th of April. > >Best regards, Wolfgang > > > >BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT >Develop your own process in accordance with the BPMN 2 standard >Learn Process modeling best practices with Bonita BPM through live >exercises >http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual >event?utm_ >source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF >_______________________________________________ >Maximadiscuss mailing list >Maximadiscuss@... >https://lists.sourceforge.net/lists/listinfo/maximadiscuss  Diese Nachricht wurde von meinem Mobiltelefon mit Kaiten Mail gesendet. 
From: Wolfgang Dautermann <wolfgang@da...>  20150418 08:19:53

Am 16.04.2015 um 23:07 schrieb Robert Dodier: > What is the result of: > > (encodeuniversaltime 0 0 0 1 1 1900 0) > > What about: > > (decodeuniversaltime 0 0) Maxima 5.36.0 http://maxima.sourceforge.net using Lisp CLISP 2.49 (20100707) Distributed under the GNU Public License. See the file COPYING. Dedicated to the memory of William Schelter. The function bug_report() provides bug reporting information. (%i1) :lisp (encodeuniversaltime 0 0 0 1 1 1900 0) 0 (%i1) :lisp (decodeuniversaltime 0 0) 0 0 0 1 1 1900 0 NIL 0 (%i1) I assume, that is the expected result? Anyway, if someone is interested in the Clisp based crosscompiled installer (or if it helps solving the issue)  I have put it on my website: http://wolfgang.dautermann.at/maxima/ (of course with the build scripts which I used for generating it). The previous version (5.35.1) is still available. I do not want to keep a complete history of maxima windows installers on my website (as on sourceforge) but will provide the last two or three versions. Sorry for the delay. I am rather busy now organizing a local Linux event ("Grazer Linuxtage"), which will be on 24th and 25th of April. Best regards, Wolfgang 
From: Gunter Königsmann <gunter@pe...>  20150418 06:29:26

Correctness is important. But sometimes I wonder if there is a way to make maxima tell the user that it refuses to do something that *looks* obvious since it is fundamentally incorrect. Let the computer help with at least part of the teaching? I assume there is no possibility to do so, though. And if there were any it would probably mean loads of bogus warnings. Kind regards, Gunter. Am 17. April 2015 23:49:59 MESZ, schrieb Raymond Toy <toy.raymond@...>: >>>>>> "Barton" == Barton Willis <willisb@...> writes: > > Barton> Telling calculus students that integrate(1/x,x) = > Barton> log(abs(x)) is somewhat slothful. If we didn't conflate > Barton> functions and their formula, maybe it would be a better > Barton> world: > > >Barton> integrate(x in (inf,0) > 1/x) = x in (inf 0) > >log(x), (1) > >Barton> integrate(x in (0,inf) > 1/x) = x in (0, inf) > >log(x), (2) > >Barton> integrate(x in [1,1]) > 1/x) = bogus > (3) > > > > Barton> Surely a student should look at the limits of integration > Barton> and choose between (1), (2), or (3). The logabs issue > Barton> comes up every so often. Users want answers that match US > Barton> calculus textbooksI can understand that point of > Barton> view. But it does eliminate a teachable moment. > >I'm sure many people disagree with me, but I've always felt maxima >should default to correctness, even if that means those high school >and college students don't get the answer they expect. (Because >they've been taught the wrong thing or because they don't remember.) > >I guess a corollary is not catering to beginners too much; you're only >a beginner once, so let's educate them so they're not beginners >anymore. > > >Ray > > > >BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT >Develop your own process in accordance with the BPMN 2 standard >Learn Process modeling best practices with Bonita BPM through live >exercises >http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual >event?utm_ >source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF >_______________________________________________ >Maximadiscuss mailing list >Maximadiscuss@... >https://lists.sourceforge.net/lists/listinfo/maximadiscuss  Diese Nachricht wurde von meinem Mobiltelefon mit Kaiten Mail gesendet. 
From: Raymond Toy <toy.raymond@gm...>  20150417 21:50:25

>>>>> "Barton" == Barton Willis <willisb@...> writes: Barton> Telling calculus students that integrate(1/x,x) = Barton> log(abs(x)) is somewhat slothful. If we didn't conflate Barton> functions and their formula, maybe it would be a better Barton> world: Barton> integrate(x in (inf,0) > 1/x) = x in (inf 0) > log(x), (1) Barton> integrate(x in (0,inf) > 1/x) = x in (0, inf) > log(x), (2) Barton> integrate(x in [1,1]) > 1/x) = bogus (3) Barton> Surely a student should look at the limits of integration Barton> and choose between (1), (2), or (3). The logabs issue Barton> comes up every so often. Users want answers that match US Barton> calculus textbooksI can understand that point of Barton> view. But it does eliminate a teachable moment. I'm sure many people disagree with me, but I've always felt maxima should default to correctness, even if that means those high school and college students don't get the answer they expect. (Because they've been taught the wrong thing or because they don't remember.) I guess a corollary is not catering to beginners too much; you're only a beginner once, so let's educate them so they're not beginners anymore.  Ray 
From: Gunter Königsmann <gunter@pe...>  20150417 19:46:26

I assume I found an error in the CCL version that is used in the current windows installer: %i1 load("engineeringformat")$ %i2 0.5; %o2 5.0*10^7 SBCL and GCL did always show 500.0E3 in this case: engineeringformat.lisp tries to make all floats have exponents dividable by 3 when printed. Kind regards, Gunter. 
From: Barton Willis <willisb@un...>  20150417 19:33:20

Telling calculus students that integrate(1/x,x) = log(abs(x)) is somewhat slothful. If we didn't conflate functions and their formula, maybe it would be a better world: integrate(x in (inf,0) > 1/x) = x in (inf 0) > log(x), (1) integrate(x in (0,inf) > 1/x) = x in (0, inf) > log(x), (2) integrate(x in [1,1]) > 1/x) = bogus (3) Surely a student should look at the limits of integration and choose between (1), (2), or (3). The logabs issue comes up every so often. Users want answers that match US calculus textbooksI can understand that point of view. But it does eliminate a teachable moment. Barton ________________________________ From: Richard Fateman <fateman@...> Sent: Friday, April 17, 2015 12:38 To: maximadiscuss@... Subject: Re: [Maximadiscuss] logabs: inconsistent behavior cos(x)/sin(x) vs. cot(x) Here's another fix. Remove logabs flag from Maxima. It has some questionable properties. For example w: 1/(1x)+1/(x1) integrate( 1/w,x), logabs:true; results in an unevaluated integral. (!) leaving off the logabs:true, integrate gets log(1x)+log(x1) which is... a constant. Given that some people do not know about complex numbers and would prefer log to be defined only for nonnegative argument, why restrict this to the result of integration? If you want log to be log(abs()), then define something like this... logabsify(z):=subst(lambda([%r],log(abs(%r))),'log,z); try logabsify(log(3)+log(q); logabsify(log(3)+log(x1)+log(1x)); oh, here's a bug... since w, above is zero, maybe integrate(w,x,0,10) should not say "integral is divergent". Though integrate(w,x,10,10) comes out as zero. RJF On 4/17/2015 8:25 AM, Barton Willis wrote: > I imagine that Maxima uses a lookup table for simple functions, and that in the case of cot(x), tan(x), coth(x) and tanh(x) the lookup entry > just doesn't take into account the logabs setting. There might be special cases, but this is more true than false: the operators %cot %tan, %tanh, and %coth all have the integral symbol property. At the top level of integration, Maxima tries the derivative divides rule for these operators. You can check for this symbol property using (%i3) :lisp(mapcar #'(lambda (x) (get x 'integral"")) (list '%cot '%tan '%tanh '%coth)) (((X) ((%LOG) ((%SIN) X))) ((X) ((%LOG) ((%SEC) X))) ((X) ((%LOG) ((%COSH) X))) ((X) ((%LOG) ((%SINH) X)))) The symbol properties can be changed: (%i3) :lisp(putprop '%coth `((x) ,#'cothintegral) 'integral)); ((X) (LAMBDABLOCK COTHINTEGRAL (X) (TAKE '(%LOG) (IF $LOGABS (TAKE '(MABS) (TAKE '(%SINH) X)) (TAKE '(%SINH) X))))) (%i3) integrate(coth(x),x), logabs : false; (%o3) log(sinh(x)) (%i4) integrate(coth(x),x), logabs : true; (%o4) log(abs(sinh(x))) I understand the desire to make results match calculus textbooksit's a worthy goal. But the absolute value also makes things wrong everywhere in the complex planethus the user option logabs. Maybe somebody would like to fix up the symbol properties to take in account logabs. Barton  BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT Develop your own process in accordance with the BPMN 2 standard Learn Process modeling best practices with Bonita BPM through live exercises http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual event?utm_ source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF _______________________________________________ Maximadiscuss mailing list Maximadiscuss@...<mailto:Maximadiscuss@...> https://lists.sourceforge.net/lists/listinfo/maximadiscuss 
From: Richard Fateman <fateman@be...>  20150417 17:37:38

Here's another fix. Remove logabs flag from Maxima. It has some questionable properties. For example w: 1/(1x)+1/(x1) integrate( 1/w,x), logabs:true; results in an unevaluated integral. (!) leaving off the logabs:true, integrate gets log(1x)+log(x1) which is... a constant. Given that some people do not know about complex numbers and would prefer log to be defined only for nonnegative argument, why restrict this to the result of integration? If you want log to be log(abs()), then define something like this... logabsify(z):=subst(lambda([%r],log(abs(%r))),'log,z); try logabsify(log(3)+log(q); logabsify(log(3)+log(x1)+log(1x)); oh, here's a bug... since w, above is zero, maybe integrate(w,x,0,10) should not say "integral is divergent". Though integrate(w,x,10,10) comes out as zero. RJF On 4/17/2015 8:25 AM, Barton Willis wrote: > > > I imagine that Maxima uses a lookup table for simple functions, and > that in the case of cot(x), tan(x), coth(x) and tanh(x) the lookup entry > > > just doesn't take into account the logabs setting. > > > > There might be special cases, but this is more true than false: the > operators %cot %tan, %tanh, and %coth all have the integral symbol > property. At the top level of integration, Maxima tries the derivative > divides rule for these operators. You can check for this symbol > property using > > > (%i3) :lisp(mapcar #'(lambda (x) (get x 'integral"")) (list '%cot > '%tan '%tanh '%coth)) > (((X) ((%LOG) ((%SIN) X))) ((X) ((%LOG) ((%SEC) X))) > ((X) ((%LOG) ((%COSH) X))) ((X) ((%LOG) ((%SINH) X)))) > > > The symbol properties can be changed: > > > (%i3) :lisp(putprop '%coth `((x) ,#'cothintegral) 'integral)); > ((X) > (LAMBDABLOCK COTHINTEGRAL (X) > (TAKE '(%LOG) > (IF $LOGABS (TAKE '(MABS) (TAKE '(%SINH) X)) > (TAKE '(%SINH) X))))) > > > (%i3) integrate(coth(x),x), logabs : false; > (%o3) log(sinh(x)) > > > (%i4) integrate(coth(x),x), logabs : true; > (%o4) log(abs(sinh(x))) > > > > I understand the desire to make results match calculus textbooksit's > a worthy goal. But the absolute value also makes things wrong > everywhere in the complex planethus the user option logabs. Maybe > somebody would like to fix up the symbol properties to take in account > logabs. > > > Barton > > > >  > BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT > Develop your own process in accordance with the BPMN 2 standard > Learn Process modeling best practices with Bonita BPM through live exercises > http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual event?utm_ > source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF > > > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss 
From: Barton Willis <willisb@un...>  20150417 15:25:32

> I imagine that Maxima uses a lookup table for simple functions, and that in the case of cot(x), tan(x), coth(x) and tanh(x) the lookup entry > just doesn't take into account the logabs setting. There might be special cases, but this is more true than false: the operators %cot %tan, %tanh, and %coth all have the integral symbol property. At the top level of integration, Maxima tries the derivative divides rule for these operators. You can check for this symbol property using (%i3) :lisp(mapcar #'(lambda (x) (get x 'integral"")) (list '%cot '%tan '%tanh '%coth)) (((X) ((%LOG) ((%SIN) X))) ((X) ((%LOG) ((%SEC) X))) ((X) ((%LOG) ((%COSH) X))) ((X) ((%LOG) ((%SINH) X)))) The symbol properties can be changed: (%i3) :lisp(putprop '%coth `((x) ,#'cothintegral) 'integral)); ((X) (LAMBDABLOCK COTHINTEGRAL (X) (TAKE '(%LOG) (IF $LOGABS (TAKE '(MABS) (TAKE '(%SINH) X)) (TAKE '(%SINH) X))))) (%i3) integrate(coth(x),x), logabs : false; (%o3) log(sinh(x)) (%i4) integrate(coth(x),x), logabs : true; (%o4) log(abs(sinh(x))) I understand the desire to make results match calculus textbooksit's a worthy goal. But the absolute value also makes things wrong everywhere in the complex planethus the user option logabs. Maybe somebody would like to fix up the symbol properties to take in account logabs. Barton 
From: Richard Fateman <fateman@be...>  20150417 14:55:10

http://arxiv.org/ftp/arxiv/papers/1404/1404.1905.pdf This was prepared for the National Research Council of the National Academies (US) by a collection of wellknown senior mathematical academics (mostly.) And some database/library types. It contains some proposed relationships with computer algebra systems, but a quick scan gives me the impression that the authors' emphasis is on supporting math research per se, and less so math libraries for people who use math. There is no mention of Maxima, though there is some mention of Mathematica / Mathworld /WolframAlpha, also Maple and Sage is mentioned. The envisioned relationship seems to be (p. 52) that one could cut out a formula and paste it in to Matlab or Mathematica. This seems unimaginative; it would seem to me that a static view of a library as something to be searched, is very old fashioned. Even applying "machine learning" to find the likely mostrelevant material is still old fashion. A 21st century view would consider the appropriateness of storing and retrieving algorithms and EXECUTING THEM. Not just printing them out as PDF or transferring data as MathML. 
From: David Scherfgen <d.scherfgen@go...>  20150417 13:59:20

Oh, and logabs also doesn't change anything for sec(x) and csc(x): (%i1) integrate(csc(x),x); (%o1) log(csc(x)+cot(x)) (%i2) integrate(csc(x),x), logabs; (%o2) log(csc(x)+cot(x)) (%i3) integrate(sec(x),x); (%o3) log(tan(x)+sec(x)) (%i4) integrate(sec(x),x), logabs; (%o4) log(tan(x)+sec(x)) 20150417 15:52 GMT+02:00 David Scherfgen <d.scherfgen@...>: > Hi, > > please have a look at this transcript: > > (%i1) integrate(cos(x)/sin(x),x); > (%o1) log(sin(x)) > > (%i2) integrate(cos(x)/sin(x),x), logabs; > (%o2) log(abs(sin(x))) > > (%i3) integrate(cot(x),x); > (%o3) log(sin(x)) > > (%i4) integrate(cot(x),x), logabs; > (%o4) log(sin(x)) > > (%i5) integrate(1/tan(x),x); > (%o5) log(sin(x)) > > (%i6) integrate(1/tan(x),x), logabs; > (%o6) log(abs(sin(x))) > > cos(x)/sin(x), cot(x) and 1/tan(x) should mathematically all be the same, > right? > Still, Maxima gives a different antiderivative for cot(x) when logabs is > used. > The same problem appears with sin(x)/cos(x) and tan(x) and the hyperbolic > variants of all of them. > > I imagine that Maxima uses a lookup table for simple functions, and that > in the case of cot(x), tan(x), coth(x) and tanh(x) the lookup entry just > doesn't take into account the logabs setting. > > I think this should be fixed, what do you think? > > Best regards, > David > 
From: David Scherfgen <d.scherfgen@go...>  20150417 13:52:26

Hi, please have a look at this transcript: (%i1) integrate(cos(x)/sin(x),x); (%o1) log(sin(x)) (%i2) integrate(cos(x)/sin(x),x), logabs; (%o2) log(abs(sin(x))) (%i3) integrate(cot(x),x); (%o3) log(sin(x)) (%i4) integrate(cot(x),x), logabs; (%o4) log(sin(x)) (%i5) integrate(1/tan(x),x); (%o5) log(sin(x)) (%i6) integrate(1/tan(x),x), logabs; (%o6) log(abs(sin(x))) cos(x)/sin(x), cot(x) and 1/tan(x) should mathematically all be the same, right? Still, Maxima gives a different antiderivative for cot(x) when logabs is used. The same problem appears with sin(x)/cos(x) and tan(x) and the hyperbolic variants of all of them. I imagine that Maxima uses a lookup table for simple functions, and that in the case of cot(x), tan(x), coth(x) and tanh(x) the lookup entry just doesn't take into account the logabs setting. I think this should be fixed, what do you think? Best regards, David 
From: Andrej Vodopivec <andrej.vodopivec@gm...>  20150417 07:01:25

This is a known issue with sbcl. The heap size must be increased so that lapack can be compiled. Unfortunately I have not been able to find a correct setting for this on Windows (if I increase the heap size over a certain limit, then sbcl fails to start). You can try the openmcl binary. I can compile lapack with it. Andrej On Fri, Apr 17, 2015 at 12:02 AM, Mike Valenzuela <mickle.mouse@...> wrote: > Sorry to keep spamming this list, I am just trying to be helpful. > > I tried rebuilding lapack and saw this error during the build procedure: > " > Total bytes allocated = 359019624 > Dynamicspacesize bytes = 536870912 > GC control variables: > *GCINHIBIT* = true > *GCPENDING* = true > *STOPFORGCPENDING* = false > fatal error encountered in SBCL pid 4552(tid 34396216): > Heap exhausted, game over. > " > > On Thu, Apr 16, 2015 at 2:39 PM, Mike Valenzuela <mickle.mouse@...> > wrote: >> >> Oh I forgot one potentially useful detail. I am using the >> maximasbcl5.36.0. I have not tried this with the openmcl. >> >> >> >> On Thu, Apr 16, 2015 at 2:38 PM, Mike Valenzuela <mickle.mouse@...> >> wrote: >>> >>> Hello all, >>> >>> I just thought I would test out a few things from the Maxima 5.36.0 in >>> windows 7. From the command line version (because the wxmaxima gui has a >>> history of problems with lapack), I get >>> >>> " >>> ; compilation unit aborted >>> ; caught 1 fatal ERROR condition >>> ; >>> ; compilation unit aborted >>> ; caught 1 fatal ERROR condition >>> ; >>> ; compilation unit aborted >>> ; caught 1 fatal ERROR condition >>> loadfile: failed to load C:/Program Files >>> (x86)/Maximasbcl.36.0/share/maxima/5.36.0/share/lapack/loadlapack.lisp >>> " >>> >>> Can anyone else on Windows confirm? >> >> > > >  > BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT > Develop your own process in accordance with the BPMN 2 standard > Learn Process modeling best practices with Bonita BPM through live exercises > http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual event?utm_ > source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss > 
From: Robert Dodier <robert.dodier@gm...>  20150417 05:13:51

On 20150416, Dimiter Prodanov <dimiterpp@...> wrote: > Is it possible then to put a new program in the pipeline to check an > input for a number of rules and do substitutions before passing the > output to simlifier? I'm not sure I understand what you want to do. Maybe this helps: there is a function named TOPLEVELMACSYMAEVAL which is called to compute the output for every input which you type into Maxima. If you redefine that function, it will change the way every input is processed. Here's an example. I've saved the builtin definition which you can see in src/macsys.lisp; it doesn't matter what it is, we're just going to call it from the new definition. FOO is the saved definition. (%i1) to_lisp(); Type (tomaxima) to restart, ($quit) to quit Maxima. MAXIMA> (let ((foo #'toplevelmacsymaeval)) (defun toplevelmacsymaeval (x) (funcall foo ($substitute '1234 '$%pi x)))) TOPLEVELMACSYMAEVAL MAXIMA> (tomaxima) Returning to Maxima (%o1) true (%i2) x + f(%pi) + g(%pi^2); (%o2) x + g(1522756) + f(1234) If you say more about exactly what rules and substitutions you want to apply, we can give some advice about how to write the new TOPLEVELMACSYMAEVAL. best Robert Dodier 
From: Mike Valenzuela <mickle.mouse@gm...>  20150416 22:02:38

Sorry to keep spamming this list, I am just trying to be helpful. I tried rebuilding lapack and saw this error during the build procedure: " Total bytes allocated = 359019624 Dynamicspacesize bytes = 536870912 GC control variables: *GCINHIBIT* = true *GCPENDING* = true *STOPFORGCPENDING* = false fatal error encountered in SBCL pid 4552(tid 34396216): Heap exhausted, game over. " On Thu, Apr 16, 2015 at 2:39 PM, Mike Valenzuela <mickle.mouse@...> wrote: > Oh I forgot one potentially useful detail. I am using > the maximasbcl5.36.0. I have not tried this with the openmcl. > > > > On Thu, Apr 16, 2015 at 2:38 PM, Mike Valenzuela <mickle.mouse@...> > wrote: > >> Hello all, >> >> I just thought I would test out a few things from the Maxima 5.36.0 in >> windows 7. From the command line version (because the wxmaxima gui has a >> history of problems with lapack), I get >> >> " >> ; compilation unit aborted >> ; caught 1 fatal ERROR condition >> ; >> ; compilation unit aborted >> ; caught 1 fatal ERROR condition >> ; >> ; compilation unit aborted >> ; caught 1 fatal ERROR condition >> loadfile: failed to load C:/Program Files >> (x86)/Maximasbcl.36.0/share/maxima/5.36.0/share/lapack/loadlapack.lisp >> " >> >> Can anyone else on Windows confirm? >> > > 
From: Mike Valenzuela <mickle.mouse@gm...>  20150416 21:39:51

Oh I forgot one potentially useful detail. I am using the maximasbcl5.36.0. I have not tried this with the openmcl. On Thu, Apr 16, 2015 at 2:38 PM, Mike Valenzuela <mickle.mouse@...> wrote: > Hello all, > > I just thought I would test out a few things from the Maxima 5.36.0 in > windows 7. From the command line version (because the wxmaxima gui has a > history of problems with lapack), I get > > " > ; compilation unit aborted > ; caught 1 fatal ERROR condition > ; > ; compilation unit aborted > ; caught 1 fatal ERROR condition > ; > ; compilation unit aborted > ; caught 1 fatal ERROR condition > loadfile: failed to load C:/Program Files > (x86)/Maximasbcl.36.0/share/maxima/5.36.0/share/lapack/loadlapack.lisp > " > > Can anyone else on Windows confirm? > 
From: Mike Valenzuela <mickle.mouse@gm...>  20150416 21:38:41

Hello all, I just thought I would test out a few things from the Maxima 5.36.0 in windows 7. From the command line version (because the wxmaxima gui has a history of problems with lapack), I get " ; compilation unit aborted ; caught 1 fatal ERROR condition ; ; compilation unit aborted ; caught 1 fatal ERROR condition ; ; compilation unit aborted ; caught 1 fatal ERROR condition loadfile: failed to load C:/Program Files (x86)/Maximasbcl.36.0/share/maxima/5.36.0/share/lapack/loadlapack.lisp " Can anyone else on Windows confirm? 
From: Robert Dodier <robert.dodier@gm...>  20150416 21:08:09

On 20150416, Wolfgang Dautermann <wolfgang@...> wrote: > (%i44) parse_timedate("19700101 00:59:59+00:00"); > Maxima encountered a Lisp error: > > Win32 error 5 (ERROR_ACCESS_DENIED): Access is denied. > (%i45) parse_timedate("19700101 01:00:00+00:00"); > (%o45) 2208988800 > > (this was tested both using Wine (and Linux) and also Windows 7) Hmm, I can't confirm this using Wine + Linux. I copied the parse_timedate code from src/commac.lisp and loaded it (and just enough other stuff to get it to run) into clisp.exe. I find that it works OK for dates before 19700101, all the way back to: ($parse_timedate "18991231 16:00:00") ;; I'm in timezone UTC8:00 => 0 But aside from that, if dates before 19700101 cause trouble, that would seem to be a bug in Clisp, right? because Common Lisp universal time starts at 19000101. What is the result of: (encodeuniversaltime 0 0 0 1 1 1900 0) What about: (decodeuniversaltime 0 0) > But I did also compile a (clisp based) Version on Linux, that works > without problems. Yes, Clisp (native build) on Linux passes these tests for me. best, Robert Dodier 
From: Richard Fateman <fateman@be...>  20150416 20:42:58

Yes, this can happen because radcan is not a zeroequivalence algorithm for _nested _radicals. So what happens here, I think, is that each call reduces the depth of nesting somewhat, and ultimately the expression simplifies to zero. On 4/16/2015 9:16 AM, Stavros Macrakis (Σταῦρος Μακράκης) wrote: > Consider > > ex: > 256^(log(sqrt(3^(1/4)*sqrt(3^(1/4)*(2*sqrt(16859)+50*3^(3/2))^(2/3)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)4*3^(1/4)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3).... <snip> there is a program sqrtdenest() that might be of some use, except it seems that it doesn't solve this problem and still requires 3 radcans. RJF 
From: Robert Dodier <robert.dodier@gm...>  20150416 20:09:49

On 20150416, Mario Rodriguez <biomates@...> wrote: >> I suspect I made commit 856a1c... from an earlier version of >> gnuplot.lisp, and inadvertently removed your changes. >> >> I'll revert your changes as soon as possible. >> > > Done. OK, great. I will merge eb96c1f into branch5_36 and tag 5.36.1 in a few days. best, Robert Dodier 
From: Mike Valenzuela <mickle.mouse@gm...>  20150416 17:24:25

Hello all, Fernando Pacheco made an excellent suggestion: the piecewise Maxima package from Sourceforge (pw.mac). I found it here: http://sourceforge.net/projects/piecewisefunc/?source=typ_redirect. If Richard Hennessy is still on this mailing list, thank you for developing this package. Anyways, since Fernando messaged me directly rather than through the mailing list, I am forwarding his response to the whole mailing list to help future people whom may have encountered what I've encountered. I would also like to ask if anyone has used this package? The instructions seem easy enough and the results seem correct. Maybe the Maxima development team might want to include it as an unsupported contrib file? It is released under the GPL2 license. On Thu, Apr 16, 2015 at 4:57 AM, Fernando Pacheco <fbpwalter@...> wrote: > Hi, > > do You know about pw.mac in SourceForge ? > > Best regards > > F. Pacheco > >  > Date: Wed, 15 Apr 2015 16:08:40 0700 > From: mickle.mouse@... > To: maximadiscuss@... > Subject: [Maximadiscuss] Abs Integrate missing an identity? Slightly > wrong behavior > > > Hello all, > > I'm using Maxima 5.32.1.2 for Windows 7 x64. I believe Maxima is missing > an identity in the abs integrate package. > > Consider the following mwe: > (%i1) display2d: false$ > (%i2) load(abs_integrate)$ > (%i3) integrate( signum(x), x); > (%o3) abs(x) > (%i4) diff( abs(x), x); > (%o4) x/abs(x) > (%i5) integrate( signum(xa), x); > (%o5) abs(xa); > (%i6) integrate( signum(a*x), x); > (%o6) signum(a)*abs(x); > > The above appear to work correctly, but now for where it fails: > (%i7) integrate( signum(a*x  b), x); > (%o7) 'integrate(signum(a*xb),x); > (%i8) diff( signum(a) * abs(xb), x ); > (%o8) signum(a)*(xb)/abs(xb) > > I expected %i7 to produce: signum(a) * abs(xb) ). This is not much of an > issue for simple integrals, but by the time you have something like: > integrate( signum(a*xb)*signum(c*xd)*signum(e*xf),x), it would be > useful to have the correct behavior. > > Also I think Maxima's diff on abs(x) should return signum(x) instead of > x/abs(x). I argue this for two reasons. First, so diff(integrate( > signum(x),x ) ) returns signum(x). Secondly, signum(0)=0, but ev(x/abs(x), > x=0) doesn't behave well. I also realize many may argue against such a > change. > > diff(abs(x),x)=x/abs(x) is technically more correct in a sense: > diff(abs(x),x) at x=0 is undefined. However using a tiny bit of theory > from nonlinear programming (set theory), the subgradient of abs(x) at x=0 > includes all values in [1,1]. Maybe Maxima should have a subgradient? The > average or center subgradient makes a lot of sense to use in practice. If > one is following abs(x) to its minimum, the subgradient including the > value 0 at x=0 means that point is a candidate minimizer (and in fact > actually is in this trivial example). Furthermore, since I come from > engineering more than formal mathematics, we engineers typically ignore the > behavior about any set of points with zero measure. Not everyone is allowed > this luxury however, so maybe some global flag could be set? This would > allow some people to have their pedantically correct interpretation and > others to gloss over tiny details when they happen with probability zero. > Regardless, it should would be nice to see a closed form expression for integrate( > signum(a*xb)*signum(c*xd)*signum(e*xf)*...,x) from Maxima. > > Anyways, enough of my ramblings. Any thoughts from the community? > >  > BPM Camp  Free Virtual Workshop May 6th at 10am PDT/1PM EDT Develop your > own process in accordance with the BPMN 2 standard Learn Process modeling > best practices with Bonita BPM through live exercises > http://www.bonitasoft.com/bepartofit/events/bpmcampvirtual > event?utm_ > source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF > _______________________________________________ Maximadiscuss mailing > list Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss > 
From: Andrey G. Grozin <A.G.G<rozin@in...>  20150416 17:01:09

I've committed maxima5.36.0 to Gentoo. Testsuite run times (in seconds) on my 64bit box are gcl2.6.12 152 sbcl1.2.10 155 ccl1.10 313 * ecl15.3.7 559 * clisp2.49 1060 * as usual, rtest16 test 50 fails due to a bug in floats formatting in these lisps. This time, I don't list times on my 32bit box: currently, I'm having problems building sbcl on it. Maybe, later, when I'll fix sbcl building... Andrey Grozin 
From: Stavros Macrakis (Σταῦρος Μακράκης) <macrakis@al...>  20150416 16:16:30

Consider ex: 256^(log(sqrt(3^(1/4)*sqrt(3^(1/4)*(2*sqrt(16859)+50*3^(3/2))^(2/3)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)4*3^(1/4)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)+60*sqrt(2*sqrt(16859)+50*3^(3/2)))+3^(3/8)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(3/4)+2*3^(5/8)*(2*sqrt(16859)+50*3^(3/2))^(1/6)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/4))/(sqrt(2)*3^(5/16)*(2*sqrt(16859)+50*3^(3/2))^(1/12)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/8)))/log(2))4*64^(log(sqrt(3^(1/4)*sqrt(3^(1/4)*(2*sqrt(16859)+50*3^(3/2))^(2/3)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)4*3^(1/4)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)+60*sqrt(2*sqrt(16859)+50*3^(3/2)))+3^(3/8)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(3/4)+2*3^(5/8)*(2*sqrt(16859)+50*3^(3/2))^(1/6)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/4))/(sqrt(2)*3^(5/16)*(2*sqrt(16859)+50*3^(3/2))^(1/12)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/8)))/log(2))+6*16^(log(sqrt(3^(1/4)*sqrt(3^(1/4)*(2*sqrt(16859)+50*3^(3/2))^(2/3)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)4*3^(1/4)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)+60*sqrt(2*sqrt(16859)+50*3^(3/2)))+3^(3/8)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(3/4)+2*3^(5/8)*(2*sqrt(16859)+50*3^(3/2))^(1/6)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/4))/(sqrt(2)*3^(5/16)*(2*sqrt(16859)+50*3^(3/2))^(1/12)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/8)))/log(2))14*4^(log(sqrt(3^(1/4)*sqrt(3^(1/4)*(2*sqrt(16859)+50*3^(3/2))^(2/3)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)4*3^(1/4)*sqrt((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)+60*sqrt(2*sqrt(16859)+50*3^(3/2)))+3^(3/8)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(3/4)+2*3^(5/8)*(2*sqrt(16859)+50*3^(3/2))^(1/6)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/4))/(sqrt(2)*3^(5/16)*(2*sqrt(16859)+50*3^(3/2))^(1/12)*((2*sqrt(16859)+50*3^(3/2))^(2/3)+4)^(1/8)))/log(2))+12 $ radcan(ex) simplifies this somewhat (using 25 secs of CPU). And radcan(radcan(ex)) simplifies it some more. But radcan(radcan(radcan(ex))) simplifies to 0. Is that expected behavior? s 
From: Mario Rodriguez <biomates@te...>  20150416 16:09:05

El 16/04/15 a las 17:43, Mario Rodriguez escribió: > > El 16/04/15 a las 12:24, Andrej Vodopivec escribió: >> On Thu, Apr 16, 2015 at 11:57 AM, Gunter Königsmann <gunter@...> wrote: >>> If it was me who did this it was done by pure lack of git knowledge and I >>> have to apologise deeply. >>> >>> Kind regards, >>> >>> Gunter. >> No, the changes were to the draw package in maxima. The system command >> in maxima behaves differently on windows than on linux, so it must be >> called differently. This change was reverted in the draw package. >> >> Andrej >> > It's my fault, sorry. > > I suspect I made commit 856a1c... from an earlier version of > gnuplot.lisp, and inadvertently removed your changes. > > I'll revert your changes as soon as possible. > Done.  Mario 