🚀 go-pugleaf

MAILER-DAEMON
(CEO_POA) Remote Name Server unavailable.Subject: Returned mail:    ----- Mail rejected by MailI. -----moo-jae pak@dg2.oclc.org

   ----- Mail rejected by MailI. -----
From: apl-l@JUPITER.SUN.CSD.UNB.CA
To:  APL-L@UNBVM1.BITNET
Subject:      apl-l UseNet batch for Wed Jun 23 23:59:09 ADT 1993

X-Ceo_Options:  The above mePrevious comments:

See document for message.


CEO document contents:
Contents of apl-l UseNet batch for Wed Jun 23 23:59:09 ADT 1993:

  1. Subject: Re: Is there a clever way to do this?
     From: Randy A MacDonald <71477.330@CompuServe.COM>

  2. Subject: Re: UN: ISIAPL WS UTILITIES
     From: wchang@phage.cshl.org ( CSHL)

  3. Subject: Re: Is there a clever way to do this?
     From: vpcsc4@sfsuvax1.sfsu.edu (Emmett McLean)

  4. Subject: Call for Participation: APL Quote-Quad
     From: Robert_Brown@MTS.cc.Wayne.edu

  5. Subject: Re: Is there a clever way to do this?
     From: reiterc@lafcol.lafayette.edu (Reiter Clifford A)

  6. Subject: J:  Even and Odd
     From: reiterc@lafcol.lafayette.edu (Reiter Clifford A)

===> Start of article 1

Newsgroups: comp.lang.apl
Path: csd.unb.ca!torn!watserv2.uwaterloo.ca!watserv1!71477.330@CompuServe.COM
From: Randy A MacDonald <71477.330@CompuServe.COM>
Subject: Re: Is there a clever way to do this?
Message-ID: <930623043110_71477.330_DHG38-1@CompuServe.COM> Originator:
daemon@watserv1.uwaterloo.ca
Sender: news@watserv1.uwaterloo.ca
Organization: University of Waterloo
Date: Wed, 23 Jun 1993 04:31:10 GMT


In article <1993Jun21.085652.15880@csus.edu>, vpcsc4@sfsuvax1.sfsu.edu (Emmett
McLean) writes:
>
>   I am seeking a verb, call it foo, which does the following :
>
>       b
>   1 1 0 1 1 1 1 0 1 1
>       m
>   7 8 9 0 1 2 3 4 5 6
>   2 3 4 5 6 7 8 9 1 5
>   1 0 5 6 4 2 2 3 1 1
>
>       b puzzle m
>   7j8 9 0j1 2j3 4 5j6
>   2j3 4 5j6 7j8 9 1j5
>   1j0 5 6j4 2j2 3 1j1
>
>   Can anyone think of a good way to implement it?

 In short my way is:

     f1 =. j./&(2&{.)  NB.  build complex out of a singleton Real
                       NB.  or an ordered pair Real,Im

     f2 =. -.+.~./\    NB.  not or not_eq/\
                       NB.  this turns the b into a cutting list

     a1 =. '((x.f.)&[)((y.f.);.1)]' : 2
                       NB. THIS IS MY *FIRST* CONJUNCTION...

                       NB.  this is a modified cut, it uses the left NB.  verb
                       to change the left noun into a cutting NB.  noun, and
                       applies the right verb to the NB.  pieces of the right
                       noun.

    puzzle =. f2 a1 f1 " 1


In <205p0i$76n@csi.jpl.nasa.gov> (Sam Sirlin) mentions using j./ and $ although
how he actually would use them would be interesting to see. his solution does
have the virtue of not needing an "exotic" conjunction like ;. (cut) so a
(j-1)th generation APL would handle his solution. Sharp/Reuters/Soliton's APL
would be able to use the cut though.

A question, how did you arrive at b and m? the semantics of b seemed awkward...


|\/|
|\\| Randy MacDonald




===> Start of article 2

Path:
csd.unb.ca!torn!howland.reston.ans.net!newsserver.jvnc.net!newsserver.cshl.org!
newsserver.cshl.org!not-for-mail From: wchang@phage.cshl.org ( CSHL)
Newsgroups: comp.lang.apl
Subject: Re: UN: ISIAPL WS UTILITIES
Date: 23 Jun 1993 13:22:12 -0400
Organization: Cold Spring Harbor Lab, NY
Lines: 37
Distribution: world
Message-ID: <20a3g4INNh6b@phage.cshl.org>
References: <60.2216.4394.0N17C13D@canrem.com>
NNTP-Posting-Host: phage.cshl.org
Cc: wchang@cshl.org


In article <60.2216.4394.0N17C13D@canrem.com> richard.levine@canrem.com
(Richard Levine) writes: >(1) From the documentation for the ISI utilities--
>
>-Iverson Software provides a process for transferring functions and
>-variables between different implementations of APL.  The objects can be
>-transferred between ISI products APLIWIN, APLI386, and APLIPC, >-and
APL-PLUS, APL2, and I-APL.
>
>-You will need to access two isiinout workspaces - one to transfer out, >-the
other to transfer in.  There is a workspace provided for each APL. >-All
documentation is in isiinout.saw.
>
>-Functions, file formats, etc. follow the standards and guidelines in >-the
APL Workspace Interchange Standard, Version 0 (WSIS0). >

Very intriguing, and probably useful.

Workspace transfer may work with ISO APL, but not with nested array APLs. For
example, the expression "A uparrow B" in APL2 cannot be parsed (hence
translated) without knowing whether A is a function (uparrow is first) or a
variable (uparrow is take).  A heuristic approach will work most of the time,
but it is possible for a local variable to shadow a function with the same
name, somewhere down the chain of function calls. (This is a well-known and
unfortunate side effect of overloading uparrow together with the dynamic
scoping rule.)

The question is, are there enough people using ISO APL for an ISO-only
transliteration to be useful?  I think all of us here use nested arrays.

Anyway, we have already spent a lot of bandwidth over this.

Perhaps we can have a Birds-of-a-Feather session at APL93, to _decide_ on a
uniform way of transliterating (or translating) APL?  My feeling is that this
would be a crucial first step toward unifying APLs (ha ha).

-- Bill Chang (wchang@cshl.org)

===> Start of article 3

Newsgroups: comp.lang.apl
Path:
csd.unb.ca!torn!howland.reston.ans.net!math.ohio-state.edu!sdd.hp.com!elroy.jpl
.nasa.gov!decwrl!csus.edu!sfsuvax1.sfsu.edu!vpcsc4 From:
vpcsc4@sfsuvax1.sfsu.edu (Emmett McLean)
Subject: Re: Is there a clever way to do this?
Message-ID: <1993Jun23.173338.23798@csus.edu>
Sender: news@csus.edu
Organization: San Francisco State University
References: <930623043110_71477.330_DHG38-1@CompuServe.COM> Date: Wed, 23 Jun
1993 17:33:38 GMT
Lines: 121


 Thank you for the help everyone.
 I am reviewing the solutions.

 Now,

 Randy A MacDonald  writes:
>
> In short my way is:
>
>     f1 =. j./&(2&{.)  NB.  build complex out of a singleton Real
>                       NB.  or an ordered pair Real,Im
>
>     f2 =. -.+.~./\    NB.  not or not_eq/\
>                       NB.  this turns the b into a cutting list
>
>     a1 =. '((x.f.)&[)((y.f.);.1)]' : 2
>                       NB. THIS IS MY *FIRST* CONJUNCTION...
>
>                       NB.  this is a modified cut, it uses the left
>                       NB.  verb to change the left noun into a cutting
>                       NB.  noun, and applies the right verb to the
>                       NB.  pieces of the right noun.
>
>    puzzle =. f2 a1 f1 " 1
>
  This is great and is along the lines of what I thought might be the fastest
  solution.

  First conjunction?  Gee. I'd say you're already an expert. Nice hack. This is
  actually the kind of approach to the problem I thought existed but couldn't
  formulate.
>
>In <205p0i$76n@csi.jpl.nasa.gov> (Sam Sirlin) mentions using j./ and $
>although how he actually would use them would be interesting to see. >his
solution does have the virtue of not needing an "exotic" conjunction >like ;.
(cut) so a (j-1)th generation APL would handle his solution.
>Sharp/Reuters/Soliton's APL would be able to use the cut though. >
>A question, how did you arrive at b and m? the semantics of b seemed
>awkward...
>
>

   The problem came up in interfacing the LAPACK
   eigenvalue-eigenvector solver with J.

   For real matrices with complex eigenvalues I would get an n by n real matrix
   which had to adjusted into an eigenvector matrix.

   From the vector of consisting of the imaginary part of the eigenvectors, I
   had a boolean vector with which to create the eigenvectors, so, for example
   :

   ] realnbyn =. ? 4 4 $ 10
0 7 3 6
7 9 3 2
9 7 7 6
0 6 8 2

   ]eigenvec =. (1 1 0 0) foo realnbyn
0j7 0j_7 3 6
7j9 7j_9 3 2
9j7 9j_7 7 6
0j6 0j_6 8 2

   Or :

   ]eigenvec =. (0 1 1 0 ) foo realnbyn
0 7j3 7j_3 6
7 9j3 9j_3 2
9 7j7 7j_7 6
0 6j8 6j_8 2

   Since I was interested in methods as compared to an actual solution I framed
   the problem in simpler terms. (I hope I don't regret this later as I review
   Randy's solution.)

   My solution turned out to be like what Sam Sirlin suggested:

   5!:2 <'foo'
++-+--------------------------------------------------------------------+
||:|vr =: |: y.                                                         |
|| |b=: x.                                                              |
|| |z=: i. ({:@$vr),0                                                   |
|| |i=: 0                                                               |
|| |n=: # b                                                             |
|| |loop) $.=. >(i=n){cont;end                                          |
|| |cont) $.=. >((1&=)i{b){real;complex                                 |
|| |real) z=: z,. i { vr                                                |
|| |i=:>:i                                                              |
|| |$.=.loop                                                            |
|| |complex)                                                            |
|| |z=:z,.($$ j./"1@(]*($ $(1 1 1 _1)"_))@((# $ 2:) # ]))@|: (i,i+1){ vr| ||
|i=:>:>: i                                                           |
|| |$.=. loop                                                           |
|| |end) y.=.z                                                          |
++-+--------------------------------------------------------------------+


 I had to break down foo into little bits
 and pieces before I had an exact model to translate into C, but here is what
 the translitoration became:

    b=raze(ne(zero,mag(wi)));z=reshape(v2(n,0),mtv);
    for(i=0;i<n;i++){
     if(0==i0(from(sc(i),b)))
      z=overr(z,from(sc(i),vr));
     else{
      t=repeat(repeat(sc(n),sc(2)),cant1(from(v2(i,i+1),vr)));
      t=tymes(t,reshape(v2(2*n,2),over(v2(1,1),v2(1,-1))));
      z=overr(z,reshape(v2(n,2),df1(cant1(t),slash(ds(CJDOT))))); i++; }
    }}

  For me, for now, the functional approaches are harder to transliterate. But
  also more interesting. But if the conjunction/functional approach is faster
  I'll (try) to implement it.

  Emmett


===> Start of article 4

Newsgroups: comp.lang.apl
Path:
csd.unb.ca!torn!watserv2.uwaterloo.ca!watserv1!Robert_Brown@MTS.cc.Wayne.edu
From: Robert_Brown@MTS.cc.Wayne.edu
Subject: Call for Participation: APL Quote-Quad
Message-ID: <687488@MTS.cc.Wayne.edu>
Originator: daemon@watserv1.uwaterloo.ca
Sender: news@watserv1.uwaterloo.ca
Organization: University of Waterloo
Date: Wed, 23 Jun 1993 17:20:38 GMT



I am the Executive Editor of the APL Quote-Quad, and I am currently working on
the June, 1993 edition of Quote-Quad.

A reveiw of the items to be published indicates a shortage of text suitable for
publication.  There is little backlog of content to be published, but I do not
believe that there is any lack of topics to be covered.  Clearly the
discussions that take place in this newsgroup indicate that we should have NO
PROBLEM filling 32 pages every quarter, and I would be delighted to receive
more in the way of articles, notes, algorithms, letters, and stories that ae
based on APL or related languages.

I will consider any text submitted to me for publication.  You may send me text
via e-mail (Please send all electronic correspondence to bob@acm.org). If you
have an article, or wish to write one, but find it difficult or awkward to send
via e-mail, please send me a short note, and I'll send you a mailing address
and other information about article submission.  If there is not enough space
for publication in this issue of Quote-Quad, your text will be considered for
future issues.  Please understand that the existence of text in my hands makes
Quote-Quad publication that much more prompt and timely.

We are interested in publishing APL "Success Stories", articles regarding
specific projects or applications, proposals for evolving APL, teaching APL,
APL coding and style standardization, use of other computing facilities in
conjunction with APL (even e-mail!), sets of APL utilities, and news of
APL-related events and meetings.

If you think you have an idea for an article, and you want to develop it,
please get in touch with me.  If you think your idea is too "trivial", or does
not "fit in" with one of the above categories, you might be wrong, so please
send it along to me.

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

Another level of participation is also possible with APL Quote-Quad, that of
helping with the publication itself.  If you would like to become a member of
the editorial group, write a regular column, or perform some other activity,
this is an excellent time to come forward.  We have need for expertise in at
least some of the follwing areas: image processing, layout, proofreading,
editing articles in areas of specialization, and soliciting for articles.

SIGAPL is in the process of electing new officers.  A time of reorganization
and revitalization is also a good chance for new people to come forward to
continue and enhance our efforts.  I look forward to your response.

+---------------------------------------+----------------------------------+ |
Robert G. Brown                      |  "We don't believe in wasting    |
|  Executive Editor, APL Quote-Quad     |   valuable Bandwidth on a lot    |
|  REPLY TO: bob@acm.org                |   of fancy message footers"      |
+---------------------------------------+----------------------------------+

===> Start of article 5

Path:
csd.unb.ca!torn!spool.mu.edu!howland.reston.ans.net!darwin.sura.net!news.udel.e
du!udel!news.intercon.com!eddie.mit.edu!magnesium.club.cc.cmu.edu!news.sei.cmu.
edu!cert.org!netnews.upenn.edu!netnews.cc.lehigh.edu!lafcol!reiterc From:
reiterc@lafcol.lafayette.edu (Reiter Clifford A) Newsgroups: comp.lang.apl
Subject: Re: Is there a clever way to do this?
Message-ID: <1993Jun23.175029.21315@Lehigh.EDU>
Date: 23 Jun 93 17:50:29 GMT
References: <205p0i$76n@csi.jpl.nasa.gov>
Sender: usenet@Lehigh.EDU
Organization: Lehigh University
Lines: 41
Nntp-Posting-Host: lafcol.lafayette.edu
X-Newsreader: TIN [version 1.1 PL9]


Sam Sirlin (sam@csi.jpl.nasa.gov) wrote:
: In article <1993Jun21.085652.15880@csus.edu>, (Emmett McLean) writes: : |>
: |>   I am seeking a verb, call it foo, which does the following :
: |>
: |>       b foo m
: |>   7j8 9 0j1 2j3 4 5j6
: |>   2j3 4 5j6 7j8 9 1j5
: |>   1j0 5 6j4 2j2 3 1j1
: |>
: |>   Can anyone think of a good way to implement it?

: The traditional apl approach I think of immediately : is to do it in two
stages:

....

: Sam Sirlin

A variation on this is:
   b=.1 1 0 1 1 1 1 0 1 1
   m=.3 10$7 8 9 0 1 2 3 4 5 6 2 3 4 5 6 7 8 9 1 5 1 0 5 6 4 2 2 3 1 1
   expand=.(* +/\)@[ { 0&,@]   NB. like numeric APL expand along 1st axis
   by2=. ,&2@-:@$ $ ]          NB. makes a m by 2 matrix from length 2m vector
   sel=.;@({&(1 0;1))          NB. replaces each 0 in a boolean vector by 0 1
   foo=. (j./"1 @ by2 @ (sel@[ expand ]))"1

   m
7 8 9 0 1 2 3 4 5 6
2 3 4 5 6 7 8 9 1 5
1 0 5 6 4 2 2 3 1 1
   b foo m
7j8 9 0j1 2j3 4 5j6
2j3 4 5j6 7j8 9 1j5
  1 5 6j4 2j2 3 1j1

Cliff
--
Clifford A. Reiter
Mathematics Department, Lafayette College
Easton, PA 18042 USA,   215-250-5277

===> Start of article 6

Path:
csd.unb.ca!torn!spool.mu.edu!howland.reston.ans.net!darwin.sura.net!news.udel.e
du!udel!news.intercon.com!eddie.mit.edu!magnesium.club.cc.cmu.edu!news.sei.cmu.
edu!cert.org!netnews.upenn.edu!netnews.cc.lehigh.edu!lafcol!reiterc From:
reiterc@lafcol.lafayette.edu (Reiter Clifford A) Newsgroups: comp.lang.apl
Subject: J:  Even and Odd
Message-ID: <1993Jun23.182213.24160@Lehigh.EDU>
Date: 23 Jun 93 18:22:13 GMT
Sender: usenet@Lehigh.EDU
Organization: Lafayette College, Easton PA
Lines: 9
Nntp-Posting-Host: lafcol.lafayette.edu


I haven't had a chance to get a feel for J's Even and Odd conjunctions .. and
.:.  Anyone have some cute (short?) examples that would help me undersand the
potential of these conjunctions?
  Thanks,  Cliff

--
Clifford A. Reiter
Mathematics Department, Lafayette College
Easton, PA 18042 USA,   215-250-5277

===> End of articles for Wed Jun 23 23:59:09 ADT 1993