Hi Mike,
Well, I *almost* got it right.
The culprit was a ] instead of a [ in as.
Now it's fixed on the wiki, as:
kpnd=: cs ($,) as |: *"0 _
cs=: ([: */&:> >.&# {.!.1&.> ;)&$
as=: ([: ~.@,@|: ;&i. +&> 0 , [)&(#@$)
Thanks for verifying :)
Jan-Pieter
On Wed, Oct 6, 2021, 16:13 'Michael Day' via Beta <[email protected]>
wrote:
> Thanks, J-P.
> The new version works when left hand rank >: right hand rank, but still
> fails otherwise:
>
> $2 3 4 kpnd&i. 7 8 9
> 14 24 36
> $2 3 4 kpnd&i. 7 8 9 10
> |index error: kpnd
> | $2 3 4 kpnd&i.7 8 9 10
> $2 3 4 kpnd&i.~ 7 8 9 10
> 14 24 36 10
> $2 3 4 kpnd&i. 7 8 9 10 11
> |index error: kpnd
> | $2 3 4 kpnd&i.7 8 9 10 11
> $2 3 4 kpnd&i.~ 7 8 9 10 11
> 14 24 36 10 11
>
> This is using J903 beta-u under Windows, but the verb "as" probably
> needs a bit
> more tweaking.
>
> Cheers,
>
> Mike
>
> On 05/10/2021 17:09, Jan-Pieter Jacobs wrote:
> > Hi Mike,
> > You are entirely right!
> > I discovered the same yesterday evening and cooked up a new version of
> the
> > axis shuffle that works right and happens to be beatiful (IMHO).
> >
> > It's on the Kronecker product wiki page.
> >
> > Jan-Pieter
> >
> > On Tue, Oct 5, 2021, 16:38 'Michael Day' via Beta <[email protected]>
> > wrote:
> >
> >> Probably not a beta problem, now, but that's where this thread started.
> >>
> >> It looks as if whatever was going wrong for you in beta-r and and beta-s
> >> is now ok.
> >> However, there seems to be an underlying problem when the ranks of the
> >> inputs
> >> differ by 2 or more.
> >>
> >> JVERSION
> >> Engine: j903/j64avx2/windows
> >> Beta-u: commercial/2021-10-01T12:08:24
> >> Library: 9.03.06
> >> Qt IDE: 1.9.3/5.15.2(5.15.2)
> >> Platform: Win 64
> >> Installer: J903 install
> >> InstallPath: c:/d/j903-beta
> >> Contact: www.jsoftware.com
> >>
> >> NB. using definitions as in the original post, of 1 Oct, copied below.
> >> A $@:kp B
> >> 6 6
> >> AA $@:kp B
> >> 6 6
> >> $each A;B
> >> +---+---+
> >> |3 3|2 2|
> >> +---+---+
> >> 3 3 $@kp&i. 2 2 NB. concentrating on input dimensions serves our
> >> purpose here...
> >> 6 6
> >>
> >> However, errors occur in some of the following examples:
> >> 2 3 $@kp&i.7 8 NB. resulting dimension as expected
> >> 14 24
> >> 2 3 $@kp&i.7 8 9 NB. resulting dimension as expected - with
> extension
> >> 14 24 9
> >> 2 3 $@kp&i.7 8 9 10 NB. fails!?
> >> |index error: kp
> >> | 2 3 $@kp&i.7 8 9 10
> >> 2 3 4 $@kp&i.7 8 9 10 NB. resulting dimension as expected
> >> 14 24 36 10
> >> 2 3 4 $@kp&i.7 8 9 10 11 NB. fails!?
> >> |index error: kp
> >> | 2 3 4 $@kp&i.7 8 9 10 11
> >> 2 3 4 5 $@kp&i.7 8 9 NB. resulting dimension as expected - with
> >> extension
> >> 14 24 36 5
> >> 2 3 4 5 $@kp&i.7 8 9 10 NB. resulting dimension as expected
> >> 14 24 36 50
> >> 2 3 4 5 6 $@kp&i.7 8 9 NB. fails?!
> >> |index error: kp
> >> | 2 3 4 5 6 $@kp&i.7 8 9
> >>
> >> FWIW, I won't bother you with my effort using direct definition ( {{
> >> ... }} ) ; it is far slower
> >> and more space consuming and less fun than kp; however, it does as
> >> least appear to work
> >> where kp fails in these examples!
> >>
> >> Cheers,
> >>
> >> Mike
> >>
> >> On 01/10/2021 08:23, Jan-Pieter Jacobs wrote:
> >>> Hi,
> >>> Yesterday I was playing with the Kronecker product, trying to extend it
> >> to
> >>> arbitrary dimensions (not that I really needed, but it seemed fun). It
> >>> works on this J:
> >>> JVERSION
> >>> Engine: j902/j32/android
> >>> Release-a: commercial/2020-12-24T11:35:03
> >>> Library: 9.02.08
> >>> Platform: Android 32 (armeabi-v7a)
> >>> Installer: unknown
> >>> InstallPath: /mnt/sdcard/Android/data/com.jsoftware.j.android/files
> >>> Contact: www.jsoftware.com
> >>>
> >>> But not on beta-r, nor beta-s on windows 64, avx2.
> >>>
> >>> The definition is as follows (based on
> >>> https://wiki.jsoftware.com/wiki/Essays/Kronecker_Product):
> >>>
> >>> kp=: cs ($,) dp |: *"0 _
> >>> NB. cs calculates the eventual size of pairwise joining axes
> >>> cs=: ([: */&:> >.&# {.!.1&.> ;)&$
> >>> NB. dp calculates the axis shuffle needed to interleave corresponding
> >> axes
> >>> of x and y
> >>> dp=: (+ {. [: ,@|:@i. 2 , >.)&(#@$)
> >>>
> >>> A=: =i. 3 NB. identity, shape 3 3
> >>> B=: >:i. 2 2 NB. shape 2 2
> >>> AA=: 0 13 26 e.~i. 3 3 NB. shape 3 3 3
> >>>
> >>> 'A kp B' and 'AA kp B' fail because 'A cs B' and 'AA cs B' throw a
> domain
> >>> error in J903-beta-r/s I don't understand, but do work in J902. Dissect
> >>> mentions an inconsistency, but eventually shows the right result.
> >>>
> >>> however, removing &$ from cs and feeding in shapes directly does work.
> >>>
> >>> 3 3 3 ([: */&:> >.&# {.!.1&.> ;) 2 2
> >>> 6 6 3
> >>>
> >>> Am I missing something or is this an interpreter bug?
> >>>
> >>> Any comments on the Kronecker product as such, does it make sense? The
> >> 2x2
> >>> case is at least consistent with the definitions in the essay.
> >>> If it is sound, I'd add it to the Essay.
> >>>
> >>> Best regards,
> >>> Jan-Pieter.
> >>> ----------------------------------------------------------------------
> >>> For information about J forums see http://www.jsoftware.com/forums.htm
> >>
> >> --
> >> This email has been checked for viruses by Avast antivirus software.
> >> https://www.avast.com/antivirus
> >>
> >> ----------------------------------------------------------------------
> >> For information about J forums see http://www.jsoftware.com/forums.htm
> >>
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
>
>
> --
> This email has been checked for viruses by Avast antivirus software.
> https://www.avast.com/antivirus
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
