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}
(254) 
_{Feb}

_{Mar}

_{Apr}

_{May}

_{Jun}

_{Jul}

_{Aug}

_{Sep}

_{Oct}

_{Nov}

_{Dec}

From: Robert Dodier <robert.dodier@gm...>  20140623 18:58:51

On 20140623, Robert Dodier <robert.dodier@...> wrote: > Ray, Mike, I've started working on that. Basic approach is to replace > the Fortran float format with alphanumeric format after converting the > bigfloat to a string (via FPFORMAT). I've run into a speedbump  not > sure what the problem is. I'll keep you posted. OK, I've figured out that I need to call BIGFLOAT::PRINTOBJECT from COBYLB since there are bigfloat objects involved (not just a list with caar=BIGFLOAT). The code runs now although the formatting is somewhat messy  I'll fiddle with it a little bit. best Robert Dodier 
From: Robert Dodier <robert.dodier@gm...>  20140623 17:33:55

On 20140623, Raymond Toy <toy.raymond@...> wrote: > The print statements will have to be replaced with something else to > print out bigfloats. Ray, Mike, I've started working on that. Basic approach is to replace the Fortran float format with alphanumeric format after converting the bigfloat to a string (via FPFORMAT). I've run into a speedbump  not sure what the problem is. I'll keep you posted. best Robert Dodier 
From: Raymond Toy <toy.raymond@gm...>  20140623 16:45:02

>>>>> "Mike" == Mike Valenzuela <mickle.mouse@...> writes: Mike> Okay, I managed to shrink the diff because I accidently Mike> added whitespace in my version. The attached version keeps Mike> whitespace the same (to the best of my knowledge). Mike> Also the only thing I did change was the print Mike> statements. Thats why it was crashing on me because I was Mike> using an iprint > 0. Thanks for the details. Yes, this is a problem with bigfloats where I didn't do anything about the print statements from the original Fortran that, of course, no nothing about bigfloats. The print statements will have to be replaced with something else to print out bigfloats. Oh, and, of course, the bigfloat version will perform slower (or very much slower) because the code can't take advantage of any hardware float support like there is for double floats.  Ray 
From: Robert Dodier <robert.dodier@gm...>  20140623 16:28:01

Take a look at share/algebra/solver/solver1.pdf which is a description of the share package 'solver'. The document is in German which I can't read, maybe you can. 'solver' can solve some equations which builtin 'solve' cannot, but I'm not sure how to characterize the class of problems it can handle. Solving equations is probably lowhanging fruit as my former project managers would say; Maxima is pretty weak in that respect, but there is a great demand for it, and strengthening Maxima's capabilities is just a "small matter of programming". best Robert Dodier 
From: Robert Dodier <robert.dodier@gm...>  20140623 16:17:23

On 20140623, Barton Willis <willisb@...> wrote: > (%i21) ?smono(p*q*x^k,x); > (%o21) true > > (%i22) :lisp(print *a) > ((MTIMES SIMP) $P $Q) > ((MTIMES SIMP) $P $Q) > > (%i22) :lisp(print *n) > $K > $K Oh look! Special variables as multiple return values! 8^) (Not volunteering to fix it ...) best, Robert Dodier 
From: Richard Fateman <fateman@be...>  20140623 14:44:45

Depends on what you mean by algebraic equations. Do you mean roots of polynomials, or algebra in the sense "computer algebra system"? In any case, nice formatting of solutions of cubic, quartic .. equations; hacks to reduce the degree e.g. if the polynomial is f(x^3); systems of linear equations, heuristics in ordering of elimination. And you can, of course, look at the source code for "solve". for more details, though not necessarily readable. You might also look at the source code for Axiom or FriCAS. RJF On 6/23/2014 3:47 AM, Evgeniy Maevskiy wrote: > http://en.wikipedia.org/wiki/Buchberger%27s_algorithm > http://en.wikipedia.org/wiki/Resultant > http://en.wikipedia.org/wiki/Factorization_of_polynomials > > Evgeniy > > 23.06.2014 9:32, Nijso Beishuizen пишет: >> Dear all, >> >> I am interested in knowing a bit more about algorithms to get explicit solutions of algebraic equations, i.e. the math behind 'solve'. >> Can somebody point me to an overview paper or historically important paper(s) or something similar that could be my starting point for learning more? >> >> >> Best regards, >> Nijso >> >> >> >> >>  >> HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions >> Find What Matters Most in Your Big Data with HPCC Systems >> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. >> Leverages Graph Analysis for Fast Processing & Easy Data Exploration >> http://p.sf.net/sfu/hpccsystems >> _______________________________________________ >> Maximadiscuss mailing list >> Maximadiscuss@... >> https://lists.sourceforge.net/lists/listinfo/maximadiscuss >> > >  > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems > Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. > Leverages Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss 
From: Richard Fateman <fateman@be...>  20140623 14:31:30

On 6/22/2014 3:43 PM, Robert Dodier wrote: > On 20140622, Richard Fateman <fateman@...> wrote: > >> This seems like extra burden, converting bignums into fullfledged >> symbols. > I don't see what's the problem. MAKNAM is always called to format > bigfloats for output. Maybe maknam shouldn't be used in formatting for output, either. an uninterned symbol still has a structure associated with it (like value  function  property  package ... stuff) which has to be garbagecollected. vs. a string. > It returns an uninterned symbol so it doesn't > clutter up the symbol table. I'm not sure about that... > The time/memory involved is negligible. > > best > > Robert Dodier > > >  > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems > Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. > Leverages Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss 
From: Barton Willis <willisb@un...>  20140623 11:33:22

The function smono (defined in asum.lisp) matches an expression to a*x^n (I think). While working on series.lisp, I noticed that smono fails to match for subvarp variables: OK: (%i21) ?smono(p*q*x^k,x); (%o21) true (%i22) :lisp(print *a) ((MTIMES SIMP) $P $Q) ((MTIMES SIMP) $P $Q) (%i22) :lisp(print *n) $K $K Not OK: (%i22) ?smono(p*q*x[1]^k,x[1]); (%o22) false If somebody fixes this, it might eliminate a one line workaround in a patch I have for series.lisp. Barton 
From: Evgeniy Maevskiy <emaevskiy@e...>  20140623 09:34:23

http://en.wikipedia.org/wiki/Buchberger%27s_algorithm http://en.wikipedia.org/wiki/Resultant http://en.wikipedia.org/wiki/Factorization_of_polynomials Evgeniy 23.06.2014 9:32, Nijso Beishuizen пишет: > Dear all, > > I am interested in knowing a bit more about algorithms to get explicit solutions of algebraic equations, i.e. the math behind 'solve'. > Can somebody point me to an overview paper or historically important paper(s) or something similar that could be my starting point for learning more? > > > Best regards, > Nijso > > > > >  > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems > Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. > Leverages Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss > 
From: Jan Hendrik Mueller <jan.mueller@ma...>  20140623 07:54:54

Are there really more than 3 solutions? I plotted abs((x+iy)^22^(x+iy)) It don't seems so... Jan > Am 21.06.2014 um 23:49 schrieb Barton Willis <willisb@...>: > > In the complex field, the solution set to 2^x = x^2 is countably infinite. Expressing all > the solutions in terms of the branches of the Lambert function is amusing, but it's not > all that useful, I think. The to_poly_solver isn't able to find even one solve 2^x = x^2, > but (no warning that there are other solutions...) > > (%i114) %solve(map('log, 2^x = x^2),x); > (%o114) %union([x=2]) > > > Barton > > > > >  > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems > Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. > Leverages Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss 
From: Nijso Beishuizen <nijso@ho...>  20140623 06:32:51

Dear all, I am interested in knowing a bit more about algorithms to get explicit solutions of algebraic equations, i.e. the math behind 'solve'. Can somebody point me to an overview paper or historically important paper(s) or something similar that could be my starting point for learning more? Best regards, Nijso 
From: Robert Dodier <robert.dodier@gm...>  20140623 05:40:07

On 20140623, Mike Valenzuela <mickle.mouse@...> wrote: > I know little about lisp & its best practices. I never heard of MAKNAM nor > the other suggested way. Considering this is only for print statements that > can mostly be turned off, I think the most correct & easiest to understand > way should be implemented. I just don't know what that way is. FPFORMAT returns a list of symbols which are the digits of the bigfloat. MAKNAM squashes the list into a single symbol whose name is the formatted bigfloat. This is entirely specific to Maxima. I don't know why FPFORMAT doesn't simply return a string but it has been like that for many decades ... I suppose it's not too late to change it but not today. (symbolname (maknam (fpformat <foo>))) is indeed the simplest way I can think of. best Robert Dodier 
From: Mike Valenzuela <mickle.mouse@gm...>  20140623 04:46:14

I know little about lisp & its best practices. I never heard of MAKNAM nor the other suggested way. Considering this is only for print statements that can mostly be turned off, I think the most correct & easiest to understand way should be implemented. I just don't know what that way is. On Jun 22, 2014 3:45 PM, "Robert Dodier" <robert.dodier@...> wrote: > On 20140622, Richard Fateman <fateman@...> wrote: > > > This seems like extra burden, converting bignums into fullfledged > > symbols. > > I don't see what's the problem. MAKNAM is always called to format > bigfloats for output. It returns an uninterned symbol so it doesn't > clutter up the symbol table. The time/memory involved is negligible. > > best > > Robert Dodier > > > >  > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems > Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. > Leverages Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss > 
From: Barton Willis <willisb@un...>  20140623 02:39:04

Got it done. > I'd say %o5 is OK because it is a sum of powers? Agreed. OK by me  please go ahead. 
From: Robert Dodier <robert.dodier@gm...>  20140622 22:44:19

On 20140622, Richard Fateman <fateman@...> wrote: > This seems like extra burden, converting bignums into fullfledged > symbols. I don't see what's the problem. MAKNAM is always called to format bigfloats for output. It returns an uninterned symbol so it doesn't clutter up the symbol table. The time/memory involved is negligible. best Robert Dodier 
From: Richard Fateman <fateman@be...>  20140622 22:26:08

On 6/22/2014 3:14 PM, Robert Dodier wrote: > On 20140622, Mike Valenzuela <mickle.mouse@...> wrote: > > > (symbolname (maknam (fpformat <somebigfloat>))) > > We have to do that, right? because bigfloats have a range larger than > any float type. This seems like extra burden, converting bignums into fullfledged symbols. Maybe you can just make these guys into strings, doing something like this... (concatenate 'string (mapcar #'(lambda(x) (aref (symbolname x)0)) (fpformat <somebigfloat>))) RJF 
From: Robert Dodier <robert.dodier@gm...>  20140622 22:14:39

On 20140622, Mike Valenzuela <mickle.mouse@...> wrote: > Also the only thing I did change was the print statements. Thats why it was > crashing on me because I was using an iprint > 0. OK, I've tried the patch, seems to mostly work as expected, but for some problems I get stuff like: FP2SINGLE: floating point overflow converting 3.43405964187283b35 to float. Can you avoid coercion to ordinary floats and arrange for bigfloats to be formatted as such? E.g. to get a string representation for a bigfloat, try: (symbolname (maknam (fpformat <somebigfloat>))) We have to do that, right? because bigfloats have a range larger than any float type. best Robert Dodier 
From: Richard Fateman <fateman@be...>  20140622 22:02:14

Henry.. Just to be sure that you have thought about this ... there are really two different languages under consideration. 1. The "object" language of mathematical expressions in which + has the meaning : x+y constructs the expression x+y and 2. the "programming" language of computer transformations on expressions in which 2+3 computes 5. Yes, they can overlap and they can even be comfortable coexisting in some contexts. Sometimes evaluation and simplification work together. In other contexts they cause confusion, inefficiency, and difficulttoexplain apparent nonsense. Sometimes it is better to write programs in Lisp, where the semantics of programs are simpler and clearer. RJF On 6/22/2014 2:42 PM, Robert Dodier wrote: > On 20140622, Henry Baker <hbaker1@...> wrote: > >> What needs to be done in this instance? > OK, some of this is already possible. > >> I'd like to define some "scalar" functions S(), N(), dot(,), triple(,,), >> I(), J(), K(), and some vector nonscalar functions V(), cross(,), and a >> quaternion product function "." whose result is nonscalar. >> >> I'd like to define some vector nonscalar constants I,J,K. > We talked about allowing functions to be declared separately from > variables, but I'm having slightly cold feet about that today. > It's going to invite confusion to give the function I and the variable I > incompatible declarations. Are you sure you want that? Maybe one or > the other needs a notation to show it's different? (That invites a > different problem with 'declare', namely that the argument to declare > can only be a symbol ... if it's not one thing it's another.) > >> I'd like the variables S, N, V to be allowed to be either scalar or >> nonscalar. >> >> I'd like it even better if I could overload "conjugate" to do something >> different for nonscalar arguments and return a nonscalar value. > 'conjugate' is a simplifying function and it already refuses to simplify > when its argument is nonscalar, so you can implement that via > tellsimp/tellsimpafter. > >> I'd like it even better if I could overload abs(A) and abs(Q) for >> nonscalar arguments, such that abs(A)^2 did NOT simplify to A*A; I'd >> like to tellsimp the simplification myself (one of the following 3: >> abs(A)^2=dot(A,A)=N(A)=A.A; this last is associative quaternion >> muliplication, not vector dot product). > Well, abs seems to treat noncommutative exponent "^^" differently than > "^" already; does that help? > > Hope this helps  thanks for your interest. > > best > > Robert Dodier > > >  > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems > Open Source. Fast. Scalable. Simple. Ideal for Dirty Data. > Leverages Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > _______________________________________________ > Maximadiscuss mailing list > Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss 
From: Robert Dodier <robert.dodier@gm...>  20140622 21:50:12

On 20140617, Bruno Fischer <br.fischer@...> wrote: > Got the same problem. Someone as posted a solution at > http://d.hatena.ne.jp/ooharak/20140228/1393594156 > > solution: > open: C:\Program Files (x86)\TeXmacs\plugins\maxima\lisp\texmacsmaxima.lisp > > Replace (defconstant texport *standardoutput*) > with (setq texport *standardoutput*) Hey, that's terrific, thanks for the info. Good to know. I hope a patch has been submitted to the TeXmacs project. best, Robert Dodier 
From: Robert Dodier <robert.dodier@gm...>  20140622 21:49:12

On 20140622, Mike Valenzuela <mickle.mouse@...> wrote: > Okay, I managed to shrink the diff because I accidently added whitespace in > my version. The attached version keeps whitespace the same (to the best of > my knowledge). > > Also the only thing I did change was the print statements. Thats why it was > crashing on me because I was using an iprint > 0. OK, sorry, I missed the point before. I'll take a look at the revised patch. best Robert Dodier 
From: Robert Dodier <robert.dodier@gm...>  20140622 21:43:06

On 20140622, Henry Baker <hbaker1@...> wrote: > What needs to be done in this instance? OK, some of this is already possible. > I'd like to define some "scalar" functions S(), N(), dot(,), triple(,,), > I(), J(), K(), and some vector nonscalar functions V(), cross(,), and a > quaternion product function "." whose result is nonscalar. > > I'd like to define some vector nonscalar constants I,J,K. We talked about allowing functions to be declared separately from variables, but I'm having slightly cold feet about that today. It's going to invite confusion to give the function I and the variable I incompatible declarations. Are you sure you want that? Maybe one or the other needs a notation to show it's different? (That invites a different problem with 'declare', namely that the argument to declare can only be a symbol ... if it's not one thing it's another.) > I'd like the variables S, N, V to be allowed to be either scalar or > nonscalar. > > I'd like it even better if I could overload "conjugate" to do something > different for nonscalar arguments and return a nonscalar value. 'conjugate' is a simplifying function and it already refuses to simplify when its argument is nonscalar, so you can implement that via tellsimp/tellsimpafter. > I'd like it even better if I could overload abs(A) and abs(Q) for > nonscalar arguments, such that abs(A)^2 did NOT simplify to A*A; I'd > like to tellsimp the simplification myself (one of the following 3: > abs(A)^2=dot(A,A)=N(A)=A.A; this last is associative quaternion > muliplication, not vector dot product). Well, abs seems to treat noncommutative exponent "^^" differently than "^" already; does that help? Hope this helps  thanks for your interest. best Robert Dodier 
From: Robert Dodier <robert.dodier@gm...>  20140622 21:21:31

On 20140622, Pankaj Sejwal <pankajsejwal@...> wrote: > (defun $concat (lis) > (cons '(mlist) (apply #'concatenate 'list (rest lis)))) > it just hangs for no apparent reason. I tried with "to_lisp()" to but it > also hangs and never comes to maxima mode after it. Well, if you :lisp (setq *debuggerhook* nil) before that and then hit ctrlC when it is apparently hanging, you should get a Lisp debugger prompt. Some varieties of Lisp will give you a stack trace  e.g. in Clisp, the command is just "backtrace". You can look at the stack trace to see where it is stuck. I seem to recall GCL gives much less useful info in a stack trace than other Lisps. If you are interesting in Lisp hacking, maybe switch to another variety (e.g. Clisp, SBCL, Clozure CL, etc). But maybe you can skip all that  isn't the existing function 'append' what you are trying to implement here? Btw there is already a function 'concat' which probably has a different effect ... best Robert Dodier 
From: Robert Dodier <robert.dodier@gm...>  20140622 21:15:45

On 20140622, Barton Willis <willisb@...> wrote: > (%i5) powerseries(1+x^n,x,0); > (%o5) x^n+1 > > (%i6) powerseries(1 + (1x)^(1/3),x,0); > (%o6) (3*sum(((1)^i5*x^i5)/beta(4/3i5,i5+1),i5,0,inf))/4+1 > > I'd say %o5 is OK because it is a sum of powers? Agreed. OK by me  please go ahead. best Robert Dodier 
From: Gunter Königsmann <gunter@pe...>  20140622 20:42:33

BEGIN PGP SIGNED MESSAGE Hash: SHA1 Did try to load the package on my linux box without wxmaxima. Since it loaded fine and I seem to remember you told there are problems loading it I assume that this is caused by the compilation mechanism not recompiling a file it thinks it doesn't need to recompile after a lisp upgrade or similar. Kind regards, Gunter. On 20.06.2014 12:08, Mike Valenzuela wrote: > Nevermind there are more bugs I think > > > On Thu, Jun 19, 2014 at 11:50 PM, Mike Valenzuela > <mickle.mouse@... <mailto:mickle.mouse@...>> wrote: > > I apologize as I forgot to mention my test environment. > > This was tested under Windows 7 Maxima 5.31.2 using wxMaxima > (because for some odd reason wxMaxima /can/ load the > bf_fmin_cobyla package). > > > On Thu, Jun 19, 2014 at 11:45 PM, Mike Valenzuela > <mickle.mouse@... <mailto:mickle.mouse@...>> wrote: > > So here is an update. > > I managed to get it working albeit with a few more tweaks than I > feel comfortable submitting to the community. > > Replace lines 11411170 and 146175 (approximately, as I think I > may have removed or added some whitespace in my editing) which > should look like: > > (f2cllib:fformat t ("~%" "~3@..." "NFVALS =" 1 (("~5D")) "~3@..." "F > =" 1 (("~13,6,2,1,'*,,'EE")) "~4@..." "MAXCV =" 1 > (("~13,6,2,1,'*,,'EE")) "~%" "~3@..." "X =" 1 (("~13,6,2,1,'*,,'EE")) > 4 (("~15,6,2,1,'*,,'EE")) "~%") nfvals f resmax (do ((i 1 > (f2cllib:intadd i 1)) (%ret nil)) ((> i iptem) (nreverse %ret)) > (declare (type f2cllib:integer4 i)) (push (f2cllib:fref x%data% > (i) ((1 *)) x%offset%) %ret))) (if (< iptem n) (f2cllib:fformat > t (1 (("~19,6,2,1,'*,,'EE")) 4 (("~15,6,2,1,'*,,'EE")) "~%") (do > ((i iptemp (f2cllib:intadd i 1)) (%ret nil)) ((> i n) (nreverse > %ret)) (declare (type f2cllib:integer4 i)) (push (f2cllib:fref > x%data% (i) ((1 *)) x%offset%) %ret)))))) > > with this modified version. (f2cllib:fformat t ("~%" "~3@..." > "NFVALS =" 1 (("~5D")) "~3@..." "F =" 1 (("~13,6,2,1,'*,,'EE")) > "~4@..." "MAXCV =" 1 (("~13,6,2,1,'*,,'EE")) "~%" "~3@..." "X =" 1 > (("~13,6,2,1,'*,,'EE")) 4 (("~15,6,2,1,'*,,'EE")) "~%") nfvals > (coerce f 'lisp:float) (coerce resmax 'lisp:float) (do ((i 1 > (f2cllib:intadd i 1)) (%ret nil)) ((> i iptem) (nreverse %ret)) > (declare (type f2cllib:integer4 i)) (push (coerce (f2cllib:fref > x%data% (i) ((1 *)) x%offset%) 'lisp:float) %ret))) (if (< iptem > n) (f2cllib:fformat t (4 (("~15,6,2,1,'*,,'EE")) "~%") (do ((i > iptemp (f2cllib:intadd i 1)) (%ret nil)) ((> i n) (nreverse > %ret)) (declare (type f2cllib:integer4 i)) (push (coerce > (f2cllib:fref x%data% (i) ((1 *)) x%offset%) 'lisp:float) > %ret)))))) > > > I am unsure how to submit this formally to the community. I > apologize. I have a 7zipped version of my modified file from > (\Maxima5.31.2\share\maxima\5.31.2\share\cobyla\lisp\bfcobylb.lisp). > > I have uploaded this file to my public directory of my dropbox > (https://dl.dropboxusercontent.com/u/63812930/Maxima/bfcobylb.7z). > > Alternatively, if for some reason the link is not working, > please email me and I should be able to give it to you. > > > > On Thu, Jun 19, 2014 at 9:11 PM, Mike Valenzuela > <mickle.mouse@... <mailto:mickle.mouse@...>> wrote: > > Hello again, > > I think there may be errors in the bf_fmin_cobyla package. Lines > (146161) and (11411156) > > (f2cllib:fformat t ("~%" "~3@..." "NFVALS =" 1 (("~5D")) "~3@..." "F > =" 1 (("~13,6,2,1,'*,,'EE")) "~4@..." "MAXCV =" 1 > (("~13,6,2,1,'*,,'EE")) "~%" "~3@..." "X =" 1 (("~13,6,2,1,'*,,'EE")) > 4 (("~15,6,2,1,'*,,'EE")) "~%") nfvals f resmax (do ((i 1 > (f2cllib:intadd i 1)) (%ret nil)) ((> i iptem) (nreverse %ret)) > (declare (type f2cllib:integer4 i)) (push (f2cllib:fref x%data% > (i) ((1 *)) x%offset%) %ret))) > > When I run the code I get an error from f saying (some bigfloat, > e.g., 1.123581321b34) is not a SEQUENCE. I do not know enough lisp > to fix this problem, but I found a bandaid solution > > (f2cllib:fformat t ("~%" "~3@..." "NFVALS =" 1 (("~5D")) "~3@..." "F > =" 1 (("~19,6,2,1,'*,,'EE")) "~4@..." "MAXCV =" 1 > (("~13,6,2,1,'*,,'EE")) "~%" "~3@..." "X =" 1 (("~13,6,2,1,'*,,'EE")) > 4 (("~15,6,2,1,'*,,'EE")) "~%") nfvals (coerce f 'lisp:float) > (coerce resmax 'lisp:float) (do ((i 1 (f2cllib:intadd i 1)) (%ret > nil)) ((> i iptem) (nreverse %ret)) (declare (type > f2cllib:integer4 i)) (push (coerce (f2cllib:fref x%data% (i) ((1 > *)) x%offset%) 'lisp:float) %ret))) > > I'm not sure if there are further problems; I'm testing that now. I > doubt this workaround is the correct way to do things in lisp, but > before today I never wrote a line of lisp. I think all bfloats > might need to be coerced before being displayed. If this is the > case then the following lines (161175) will also have a problem. > I'll keep everyone posted if I manage to get bf_fmin_cobyla to > run. > > > > > > > >  > > HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions > Find What Matters Most in Your Big Data with HPCC Systems Open > Source. Fast. Scalable. Simple. Ideal for Dirty Data. Leverages > Graph Analysis for Fast Processing & Easy Data Exploration > http://p.sf.net/sfu/hpccsystems > > > > _______________________________________________ Maximadiscuss > mailing list Maximadiscuss@... > https://lists.sourceforge.net/lists/listinfo/maximadiscuss > BEGIN PGP SIGNATURE Version: GnuPG v1 Comment: Using GnuPG with Thunderbird  http://www.enigmail.net/ iQIcBAEBAgAGBQJTpA4eAAoJEFyGwOQhHVuOzBsP/0lBmZfLufsOVIKGfH3njWj/ 4+FTN4tbAiTRSfDmX9zhXCnCZL5211R2ASW69XLUxYhYrnYzrzJBjnRKZXP+Ka7g LMZoT6c5mguGX9WyCMIfUPKylXuVRHFXkj2dL+X/6/UyzSG+4/Xgye2jGOnmNbi4 3sjK353y1NkEKdMXWT3QnbSc+zVaPIivO/q2+NOZAmtmnUrUMlc79JsNSB+g9fjT efJfnY2dE5k36E95euFvu2yLhtHVukjrstl6GHxl07zrUkksl7WMhL36+3hQXV31 FozDWDr7YuFHQq7mQ2A0ZzwZzVhhRJARXeQ/HbkBxVBNT5gpnhaR93ykKT2fiKP3 kCe+HAjondN8MZ3ut+p78cxm21ZuZu16UAjgo02AXJLQiYDpE6Wy5qpnFKOjtDTw FLsGP4C3bUEJW2q25+335uJOKrQpj1+3g36f6D7YHT6y0Ml65KO+tBKzo3gGApqq lmdzi6oGF8+626gcwZ5whuFrYHc4QQjxJ49ermwWBlPvx8aBv7qU5Nnk6FhYRdim kYOdlnGfHHqrvuIiX4JkE4Y9tUsvagfmAEGcjE6W+7UqBb+86DIHd5lxOJ6BK6a/ R6GsuOkaNowChPlIIpmJdzVaKiSt3PgGjR43HmdApcefgh5s+ub/YFdtMeks7ODf NoHb8vGIVng6IJNnHemW =V2T1 END PGP SIGNATURE 
From: Barton Willis <willisb@un...>  20140622 20:25:11

See also: Commit:0bb019cfba6c26844818cece60b955a4f20f233d * sf bug 1730044  powerseries(1+x^n,x,0) wrong code before series.lisp v 1.5 only worked when srbinexpnd called with noninteger exponent in numerator of expr to expand, or with integer exponent in denominator. v 1.5 fixed noninteger in denominateor, this changes fixes intger in numerator: powerseries(1+x^n,x,0). Always negate exponent in denominator (not just for nonints), negate again when calling srintegexpd from srbinexpnd. ________________________________ From: Barton Willis <willisb@...> Sent: Sunday, June 22, 2014 15:07 To: maximadiscuss@... Subject: [Maximadiscuss] powerseries(1+x^n,x,0) bug We have (both reported bugs): Oopssumand isn't defined for i2 = 2: (%i2) powerseries(1+x^n,x,0); (%o2) sum(x^(i2*n)/beta(2i2,i2+1),i2,0,inf)/2 Could be better: (%i3) powerseries(1 + (1x)^(1/3),x,0); (%o3) "Failed to expand" How terrible would it be if I changed series.lisp (move a test for mapping over a sum up) so that (%i5) powerseries(1+x^n,x,0); (%o5) x^n+1 (%i6) powerseries(1 + (1x)^(1/3),x,0); (%o6) (3*sum(((1)^i5*x^i5)/beta(4/3i5,i5+1),i5,0,inf))/4+1 I'd say %o5 is OK because it is a sum of powers? With the exception of the rtest16 test powerseries(1+x^n,x,0), the testsuite is OK with this change. 