I made arrayfire bindings a while back
https://github.com/Pascal-J/Jfire
its a gpu library that probably has the simper routines youwant.
From: bill lam
To: Programming forum
Sent: Wednesday, December 13, 2017 7:02 PM
Subject: Re: [Jprogramming] J-like on CU
Every programming language which I have mastered, I have read a
reference manual for at least once.
In languages which I dabble, I have not bothered - but I do not use
those much, either. (Generally speaking, if a language reference
manual is not something I can obtain cheaply, I have ignored the
I would imagine that for most sentences, once data was in GPU, it
would stay there (until some operation, such as foreigns, required
moving it back out).
That said, this would not be a small project and GPU limitations may
require imposing some seemingly arbitrary limitations - especially in
early
Well.. keep in mind that numbers are symbols - symbols with an elegant
recursive definition, for that matter.
Anyways, if you want a symbolic result, you could do something like
this (closely related to the other implementation, if you look
closely):
dss=:-:<:3 #.inv 15 21 19 11 5 7
nms=: ;:'n ne
Hi,
I think reading the dictionary is an essential part of learning .J.
Why? Firstly, it is short compared to most dictionaries.
Secondly, its very terseness begets questions thus keeping your mind awake
and inducing experimentation at the console.
An historical note : J was conceived by langua
I had tried making use of opencl inside j engine. From my experience
1. Double precision can be slower than single precision by an order of
magnitude because gpu has very few number of alu for double precision
operations. More expensive ones that designed for scientific computing
instead of gaming
> I am surprised how many people have not read the reference manual -
> not just in the J community but among technical people in general.
>
> I really don't know what to make of this.
This, to me, is akin to wondering why many people, particularly young
children, can communicate orally fairly wel
That seems to be the case.
On Wed, Dec 13, 2017 at 5:11 PM, Erling Hellenäs
wrote:
> Stated differently, the i. in the manual page for i. is different from the
> i. you use in your programs and write at the terminal. The relation between
> them is
> i.(terminal) <=> i.(manual page)"1 /Erling
>
>
Which presumably implies that the specifications and claims regarding all
verbs in the vocabulary are, unless otherwise stated, only in terms of
their corresponding intrinsic ranks (because the extension to arguments of
higher rank occurs systematically). In other words, my guess was right
and, in
Comments inline
On Wed, Dec 13, 2017 at 4:55 PM, Raul Miller wrote:
> This?
>
Well, kind of ... if all you want are the answers, which makes good
business sense.
On the other hand, doing it symbolically would take some variant of a
parser.
PS Really like your 3D coordinates (and their gener
Stated differently, the i. in the manual page for i. is different from
the i. you use in your programs and write at the terminal. The relation
between them is
i.(terminal) <=> i.(manual page)"1 /Erling
On 2017-12-13 20:53, Erling Hellenäs wrote:
Yes, as far as I understand when you execute a mo
This?
dss=:-:<:3 #.inv 15 21 19 11 5 7
nms=: ;:'n ne se s sw nw'
a11=:3 :0
({:,>./)+/||:+/\dss{~nms i.y
)
a11 10 6 1 1 #;:'se ne s sw'
15 16
Thanks,
--
Raul
On Wed, Dec 13, 2017 at 4:22 PM, Jimmy Gauvin wrote:
> HI,
>
> I have been thinking about Brian's "cancellation" approach and ma
You might try looking at what is happening with J's older brother APL:
http://www.dyalog.com/uploads/conference/dyalog16/presentations/U08_SIMD_Boolean_Array_Algorithms_handouts.pdf
http://www.elsman.com/pdf/Dyalog16.pdf
https://futhark-lang.org/publications/fhpc16.pdf
On Wed, Dec 13, 2017 at 3:
HI,
I have been thinking about Brian's "cancellation" approach and maybe we can
sort of formalize it.
Associative relations for cancelling paths:
0 = N + S = S + N
0 = NE + SW = SW + NE
0 = NW + SE = SE + NW
Associative relations for simplifying paths:
N = NE + NW = NW + NE
S = SE + SW = SE +
Definitely worth reading http://www.jsoftware.com/help/dictionary/dictb.htm
Especially this part: "Finally, each verb has three intrinsic ranks:
monadic, left, and right. The definition of any verb need specify only
its behaviour on cells of the intrinsic ranks, and the extension to
arguments of h
Nevermind, I see that my suggestion does not help.
On Wed, Dec 13, 2017 at 3:57 PM, Brian Schott
wrote:
> Wouldn't the "issue" be documented better if instead of saying
>
> The shape of i.y is |y
>
> it said
>
> The shape of i.y is |"i. y
>
>
>|"i.]2 2 $ 1 2 3 4
> 1 2
> 3 4
>
>
>
>
-
Wouldn't the "issue" be documented better if instead of saying
The shape of i.y is |y
it said
The shape of i.y is |"i. y
|"i.]2 2 $ 1 2 3 4
1 2
3 4
On Wed, Dec 13, 2017 at 3:41 PM, Raul Miller wrote:
> This issue is documented at http://www.jsoftware.com/help/
> dictionary/dictb.htm
>
>
This issue is documented at http://www.jsoftware.com/help/dictionary/dictb.htm
Thanks,
--
Raul
On Wed, Dec 13, 2017 at 1:18 PM, Jose Mario Quintana
wrote:
> Apparently, TongKe Xue got an answer to his question (and even more); but,
> I have at least one of my own.
>
> The Dictionary is the of
Not that I know of.
There should be a J-like especially for the GPU implementations, but
no one has put the energy into building (let alone maintaining) such a
thing.
P.S. probably more a gene...@jsoftware.com topic than a
programm...@jsoftware.com topic.
Thanks,
--
Raul
On Wed, Dec 13, 2017
I will look closer at the documents you and Henry have pointed to
tomorrow. /Erling
On 2017-12-13 18:15, Raul Miller wrote:
That's not it - it's your claim that the dictionary did not document
this functionality that I was referring to.
Thanks,
--
Yes, as far as I understand when you execute a monadic rank 1 verb you
also execute an implicit rank adverb. At least that is how JWithATwist
works.
i. y <=> i."1 y
Same with all other J functions with a rank less than _ or _ _. However,
I think the scalar functions have different helper progra
Raul's code works fine until you try applying it to a "sparse" array.
This is the result of his transitive closure applied to the adjacency
matrix
for the example provided, the one with 7 nodes and 2 connected groups:
tc adjex12
1 0 1 1 1 1 1
0 1 0 0 0 0 0
1 0 1 1 1 1 1
1 0 1 1 1 1 1
1 0
Hi,
1. There are two questions in this email, but they're closely
related. Feel free to reply to either or both.
2. "J-like" here means NumPy/MatLab, but with J syntax. It only
needs support tensors of Floats -- no need for Complex, Chars, Boxes,
file IO, GUI, ...
3. I know about https://
I made it to learn FParsec, a parser, and more F#. I also used it in
discussions about some features in J. It is fully useful as lambda
expressions in F# and as an n-dimensional array class in F# with some
pretty cool features. However, I do not intend to support it or develop
it further. /Erli
Apparently, TongKe Xue got an answer to his question (and even more); but,
I have at least one of my own.
The Dictionary is the official reference and the (vocabulary) entry for i.
( http://www.jsoftware.com/help/dictionary/didot.htm ). reads in part:
"
Integers i. 1
The shape of i.y is |y , an
More details at
http://code.jsoftware.com/wiki/Vocabulary/EmptyArguments
I consider the treatment of empties to be as fixed as anything in J.
Its omission from the Dictionary is just an omission.
Henry Rich
On 12/13/2017 12:22 PM, Raul Miller wrote:
The issue you discovered here is also di
The issue you discovered here is also discussed at
http://www.jsoftware.com/help/jforc/empty_operands.htm
Properly speaking, it's an implementation issue rather than a
dictionary issue (and, thus, subject to change, given sufficient
motivation - but need a really good reason before making these ki
That's not it - it's your claim that the dictionary did not document
this functionality that I was referring to.
Thanks,
--
Raul
On Wed, Dec 13, 2017 at 9:46 AM, Erling Hellenäs
wrote:
> It seems you didn't read the parts of my manual I referenced. /Erling
>
>
> Den 2017-12-13 kl. 14:39, skre
The file:///C:/Users/Admin/Downloads/JWithATwistReferenceManual.pdf describes a
new language, JWithATwist, but what problem is this language supposed so solve?
Thanks.
Bo.
Den 17:12 onsdag den 13. december 2017 skrev Erling Hellenäs
:
Hi all !
It seems some related info did not make it
Hi all !
It seems some related info did not make it into my blog and it is not
needed in my manual. My tweets about this are advertently cryptic:
-I found some interesting features of the real #J dyadic rank operator.
-Let's say the rank is zero and the argument is empty, how do you find
the
This blogpost also contains related information.
https://erlhelinfotech.wordpress.com/2016/08/17/jwithatwist-scalar-operations/
/Erling
Den 2017-12-13 kl. 15:46, skrev Erling Hellenäs:
It seems you didn't read the parts of my manual I referenced. /Erling
Den 2017-12-13 kl. 14:39, skrev Ra
It seems you didn't read the parts of my manual I referenced. /Erling
Den 2017-12-13 kl. 14:39, skrev Raul Miller:
On Wed, Dec 13, 2017 at 3:46 AM, Erling Hellenäs
wrote:
I doubt you can find reasonably accurate descriptions of this functionality
(these four helper programs) anywhere else.
T
On Wed, Dec 13, 2017 at 3:46 AM, Erling Hellenäs
wrote:
> I doubt you can find reasonably accurate descriptions of this functionality
> (these four helper programs) anywhere else.
This suggests to me that you have not read
http://www.jsoftware.com/help/dictionary/dictb.htm
nor
http://www.jsoftwa
Yes.
(2 2 $ 1 2 3 4)
1 2
3 4
i.1 2
0 1
i.3 4
0 1 2 3
4 5 6 7
8 9 10 11
(i.1 2),: i.3 4
0 1 0 0
0 0 0 0
0 0 0 0
0 1 2 3
4 5 6 7
8 9 10 11
i. (2 2 $ 1 2 3 4)
0 1 0 0
0 0 0 0
0 0 0 0
0 1 2 3
4 5 6 7
8 9 10 11
I tried to describe this behavior in section "Th
34 matches
Mail list logo
