[Maxima-bugs] [ maxima-Bugs-842863 ] qq.lisp is broken (numerical integration)

[SourceForge.net <no...@so...>]

Bugs item #842863, was opened at 2003-11-15 21:10
Message generated for change (Comment added) made by wjenkner
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=104933&aid=842863&group_id=4933

Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Barton Willis (willisbl)
Assigned to: Nobody/Anonymous (nobody)
Summary: qq.lisp is broken (numerical integration)

Initial Comment:
(C1) load("qq.lisp");
Load failed for 
C:/maxima/Maxima/share/maxima/5.9.0/share/numeric/qq
.lisp
 -- an error.  Quitting.  To debug this try DEBUGMODE
(TRUE);)
(C2) build_info();

Maxima version: 5.9.0
Maxima build date: 19:10 2/9/2003
host type: i686-pc-mingw32
lisp-implementation-type: Kyoto Common Lisp
lisp-implementation-version: GCL-2-5.0

Barton

----------------------------------------------------------------------

>Comment By: Wolfgang Jenkner (wjenkner)
Date: 2003-11-18 18:41

Message:
Logged In: YES 
user_id=581700

ad 1) Thanks :-)

I have added a TRANSLATE property for $QUANC8 (see the URLs in the
previous message below).  So the last example in the demo (concerning
the function p) really shows a speed difference between the
interpreted and the translated version as intended, at least for CLISP
and SBCL (I haven't tested it with GCL).  For SBCL the difference
seems quite impressive, actually.

This also requires a small patch to numer.lisp (see below)

ad 2) I still don't know why SBCL doesn't grok that type declaration.
I'll recompile SBCL and see.  It would also be a good thing to test
the code with CMUCL.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ cut ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Index: numer.lisp
===================================================================
RCS file: /cvsroot/maxima/maxima/src/numer.lisp,v
retrieving revision 1.1.1.1
diff -u -r1.1.1.1 numer.lisp
--- numer.lisp	8 May 2000 06:09:41 -0000	1.1.1.1
+++ numer.lisp	18 Nov 2003 17:06:15 -0000
@@ -155,6 +155,8 @@
 		   (OR LPROPS MPROPS)
 		   (OR MPROPS LPROPS))
 	       (GETL F '(OPERATORS)))))
+	((functionp f)
+	 (list 'expr f))
 	((consp f) ;(EQ (TYPEP F) 'LIST)
 	 (LIST (IF (MEMQ (CAR F) '(FUNCTION LAMBDA NAMED-LAMBDA))
 		   'EXPR
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ cut ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


----------------------------------------------------------------------

Comment By: Barton Willis (willisbl)
Date: 2003-11-18 16:49

Message:
Logged In: YES 
user_id=895922

1. Nice work.  

2. The load(qq.lisp) bug was already reported (but not
resolved) in bug 614631.  I think we should commit
the fix and close both bugs.

Barton

----------------------------------------------------------------------

Comment By: Wolfgang Jenkner (wjenkner)
Date: 2003-11-18 01:38

Message:
Logged In: YES 
user_id=581700

However, the double integral example in the demo won't work.  This
shows why $QUANC8 can't be a DEFUN...

I rewrote it as

(defmspec $quanc8 (form)
  (setq form (cdr form))
  (if (cdddr form)
      (apply #'call-quanc8
             (meval `((lambda) ((mlist) ,(cadr form)) ,(car form)))
             (mapcar #'meval (cddr form)))
      (apply #'call-quanc8 (mapcar #'meval form))))

All examples in the demo seem to work as expected (after lower-casing
them).  My current versions of the relevant files are

http://members.inode.at/wjenkner/maxima/qq.lisp

http://members.inode.at/wjenkner/maxima/qq.dem

This works in CLISP.  SBCL has some problem with the type declarations
in QUANC8 (without them it works fine).


----------------------------------------------------------------------

Comment By: Barton Willis (willisbl)
Date: 2003-11-17 21:42

Message:
Logged In: YES 
user_id=895922

The change

;(DEFUN ($QUANC8 TRANSLATED-MMACRO) (F A B
&OPTIONAL (C NIL C-P))
;  (IF (NOT C-P)
;      `((CALL-QUANC8) ,F ,A ,B)
;      `((CALL-QUANC8) ((LAMBDA) ((MLIST) ,A) ,F) ,B ,C)))

(defun $quanc8 (f a b &optional (c nil c-p))
  (if (not c-p)
      (call-quanc8 f a b)
    (call-quanc8 `((lambda) ((mlist) ,a) ,f) b c)))

allows qq.lisp to load correctly.   Here are a few simple
tests

(C1) load("c:/maximacvs/maxima/share/numeric/qq.lisp");
(D1) 		   
c:/maximacvs/maxima/share/numeric/qq.lisp
(C2) QUANC8_ABSERR : 1.0e-12;
(D2) 				    1.0E-12
(C3) quanc8('(1/x),x,1,2);
(D3) 			       0.69314718056001
(C4) %-log(2.0d0);
(D4) 			     6.283862319378386E-14
(C5) f(x) := block([], mode_declare(x,float), 1/x)$
(C6) compile(f)$
Compiling gazonk0.lsp.
End of Pass 1.  
End of Pass 2.  
OPTIMIZE levels: Safety=2, Space=2, Speed=2
Finished compiling gazonk0.lsp.
(C7) quanc8(f,1,2);
(D7) 			       0.69314718056001

Barton

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=104933&aid=842863&group_id=4933