Or, more specifically, = is comparing atoms (the things that go into
arrays, which includes everything in a box - including the shape of
things in the box).
Thanks,
--
Raul
On Wed, Dec 20, 2017 at 12:48 AM, Devon McCormick wrote:
> Or, to put it another way, equality (=) which compares values
Or, to put it another way, equality (=) which compares values is different
from equivalence (-:) which accounts for shape as well.
On Wed, Dec 20, 2017 at 12:35 AM, bill lam wrote:
> = is rank 0
> -: is rank infinity
>
> Try 0= 1$0 and 0-: 1$0
> and note the shape of results
>
> On Dec 20, 2017
= is rank 0
-: is rank infinity
Try 0= 1$0 and 0-: 1$0
and note the shape of results
On Dec 20, 2017 1:16 PM, "Dabrowski, Andrew John"
wrote:
> Here's my latest exasperation.
>
> zero =: < (1$0)
>
> zero
>
> ┌─┐
>
> │0│
>
> └─┘
>
> > zero
>
> 0
>
> $ > zero
>
> 1
>
> zero =
Here's my latest exasperation.
zero =: < (1$0)
zero
┌─┐
│0│
└─┘
> zero
0
$ > zero
1
zero = <0
0
(>zero) = 0
1
Why in God's name does J consider a singleton list 0 equal to the scalar
0 but consider the boxed versions of each distinct?
Wouldn't it type less if
key=. 128{.x,128#{.a.
or
key=. 128{.!.({.a.)x
On Dec 20, 2017 12:01 PM, "Scott Locklin" wrote:
> This verb might be useful to someone at some point. Pretty
> straightforward stuff, but fiddley and no point in reinventing the wheel
> if someone else needs it.
>
> https://
@dabrowski: In my experience, J beats Java-server-mode. J gets C level
performance. I've never gotten C-level performance via Java.
@bill: I'm happy to use "J restricted to floating points" -- I have no
attachment to doubles.
I guess no one has an argument for J+GPU beating wrappers-around-CUDA -
This verb might be useful to someone at some point. Pretty
straightforward stuff, but fiddley and no point in reinventing the wheel
if someone else needs it.
https://en.wikipedia.org/wiki/Hash-based_message_authentication_code
NB. hmac signature
NB. x key, y needs signature
hmac=: 4 : 0
key=. x
One weakness of J for GPU is that J doesn't support single precision or
half precision float but which are what GPU being commonly used for. The
overhead of conversion from/to double precision may or may not be
significant, it depends on applications, ymmv.
On Dec 20, 2017 4:39 AM, "TongKe Xue" w
Someone else should have ask that, but I forget. The image should already
saved as ~temp/viewmat.png, its default size should also be capable of
being changed by users.
On Dec 20, 2017 9:49 AM, "Louis de Forcrand" wrote:
Hi,
Does anyone know how to save a viewmat image to disk, and if so, how t
Hi,
Does anyone know how to save a viewmat image to disk, and if so, how to chose
the resulting image’s size?
Thanks,
Louis
--
For information about J forums see http://www.jsoftware.com/forums.htm
Here's another approach:
f=: [ * i.@<.@%~
60 -.&(f&1000) 24
60 180 300 420 540 660 780 900
Of course, you could easily replace f with it definition (in
parenthesis), since it's referenced only once.
But this way it's easy to see what f is doing:
60 f 1000
0 60 120 180 240 300 360 420 4
I should add: the tail-recursion-like idiom is F^:test^:_ arg
Thanks,
--
Raul
On Tuesday, December 19, 2017, Raul Miller wrote:
> J's induction operation ( ^: ) is very similar to tail recursion (in terms
> of data flow) except that it does not have full recursion as one of its
> failure mod
One of the reasons I bound to arrayfire was that it provided one interface to
AMD (opencl) and nvidia (cuda faster on it than opencl), and also to paralel
cpu libraries if no gpu is available.
A practicality downside is that arrayfire routines have a compilation step such
that the first time
J's induction operation ( ^: ) is very similar to tail recursion (in terms
of data flow) except that it does not have full recursion as one of its
failure modes.
--
Raul
On Tuesday, December 19, 2017, Dabrowski, Andrew John
wrote:
> Is J a good language for using recursion? In particular does
J has elegant recursion facilities, but it doesn't support
tail-recursion removal. Not yet, anyway; we have a plan for doing so
but it's low-priority.
Henry Rich
On 12/19/2017 5:10 PM, Dabrowski, Andrew John wrote:
Is J a good language for using recursion? In particular does it do tail
rec
GPU programming interfaces:
https://developer.nvidia.com/nvidia-gpu-programming-guide
Skip
Skip Cave
Cave Consulting LLC
On Tue, Dec 19, 2017 at 3:22 PM, Henry Rich wrote:
>
>Given that much of said works reduces to cuBlas + cuDNN, it seems like a
>> GPU-backed-J, although more concise, w
Jimmy's solution is nice, and fast as well. It answers the question that
was posed, but I usually like to list the resulting numbers, just as a
check.
Of course, it's easy to get the resulting numbers using Jimmy's approach
(which is also faster than the bulk elimination approach.)
a#~*24|a=
Is J a good language for using recursion? In particular does it do tail
recursion optimization?
--
For information about J forums see http://www.jsoftware.com/forums.htm
Even when running java in server mode?
On 12/19/2017 03:39 PM, TongKe Xue wrote:
Hi,
In my experience, on the CPU, J beats Java. I suspect this is due to
Java's GC and J's ability to via "higher representation of ranks/loops" to
run highly optimized code.
Is there any reason to believe that
Explicit (and faster for larger range) :
+/*24|60*i.>:<.1000%60
8
In this specific case is tacit necessary? useful?
On Tue, Dec 19, 2017 at 1:52 PM, 'Skip Cave' via Programming <
programm...@jsoftware.com> wrote:
> How many numbers from 1 to 1000 are divisible by 60 but not by 24?
>
Given that much of said works reduces to cuBlas + cuDNN, it seems like a
GPU-backed-J, although more concise, would end up calling the same
functions.
I expect you're right. I don't know what the interfaces to GPUs look
like, but the goal would be to have the rank operator (gpufunc"2 for
e
> I.1=#.|:0=24 60 |/i.1001
Slightly simpler:
I. wrote:
> Here's one approach:
>
>I.1=#.|:0=24 60 |/i.1001
> 60 180 300 420 540 660 780 900
>
> Thanks,
>
> --
> Raul
>
>
> On Tue, Dec 19, 2017 at 1:52 PM, 'Skip Cave' via Programming
> wrote:
> > How many numbers from 1 to 1000 are divisible
Hi,
In my experience, on the CPU, J beats Java. I suspect this is due to
Java's GC and J's ability to via "higher representation of ranks/loops" to
run highly optimized code.
Is there any reason to believe that GPU-backed-J would beat Tensorflow on
Tensor / Deep Learning work ?
Given that
Thanks for the several nice solutions to the Quora problem. On Quora, most
of the solutions to this type
of problem are like Mike Day's approach, where they use logic to finesse
the answer.
I prefer the "brute force" approach, listing and eliminating candidates in
bulk. J's array-handling primitiv
Here's one approach:
I.1=#.|:0=24 60 |/i.1001
60 180 300 420 540 660 780 900
Thanks,
--
Raul
On Tue, Dec 19, 2017 at 1:52 PM, 'Skip Cave' via Programming
wrote:
> How many numbers from 1 to 1000 are divisible by 60 but not by 24?
>
> Here's my somewhat inelegant explicit solution:
>
>
Ok, yes - rank is itself a rank zero array.
Empty shape, zero rank.
Thanks,
--
Raul
On Tue, Dec 19, 2017 at 2:11 PM, Jimmy Gauvin wrote:
> not empty rank rather o rank
>
> On Tue, Dec 19, 2017 at 2:01 PM, Raul Miller wrote:
>
>> An empty tank zero array would be inconsistent.
>>
>> The numb
numaliquotsbelow =: <.@(%~ <:)
1000 -&(60&numaliquotsbelow - (60*.24)&numaliquotsbelow) 60
8
1000 -&(60&numaliquotsbelow - (60*.24)&numaliquotsbelow) 59
8
1000 -&(60&numaliquotsbelow - (60*.24)&numaliquotsbelow) 61
7
1000 -&(60&numaliquotsbelow - (60*.24)&numaliquotsbelow) 1
8
Henr
q:24 60
2 2 2 3
2 2 3 5
So we can see that odd multiples of 60 are not divisible by 24, while
even ones are.
1000 >.@% 60
17
shows that there are 16 multiples of 60, <: 1000 and greater than zero.
8 of these are even, divisible by 24, and 8 odd, not divisible by 24 .
60*1
not empty rank rather o rank
On Tue, Dec 19, 2017 at 2:01 PM, Raul Miller wrote:
> An empty tank zero array would be inconsistent.
>
> The number of elements in an array is the product of its dimensions, and
> the multiplicative identity is 1, not 0.
>
> Thanks,
>
> --
> Raul
>
> On Tuesday, Dec
An empty tank zero array would be inconsistent.
The number of elements in an array is the product of its dimensions, and
the multiplicative identity is 1, not 0.
Thanks,
--
Raul
On Tuesday, December 19, 2017, Jimmy Gauvin wrote:
> Hi,
>
> Thinking about nothing, specifically about the empty
How many numbers from 1 to 1000 are divisible by 60 but not by 24?
Here's my somewhat inelegant explicit solution:
#e=.c#~d=.-.0=24|/c=.a#~b=.0=60|/a=.1+i.1000
8
NB. What are the numbers?
e
60 180 300 420 540 660 780 900
How would I construct a more compact tacit solution?
Skip
Hi,
Thinking about nothing, specifically about the empty scalar or empty 0-cell.
Erling H. - " I experience an inconsistency in this model in the sense
that an array of rank greater than zero can be empty, a 0-cell can not."
The internal structure of a J array could support a rank 0 empty objec
The paper is here: http://www.jsoftware.com/papers/APLDictionary.htm /Erling
Den 2017-12-19 kl. 16:00, skrev John Baker:
Dictionary APL
--
For information about J forums see http://www.jsoftware.com/forums.htm
You might want to look up Ken Iverson’s 1980s era paper Dictionary APL. I
believe Roger maintains an online version. This document clearly explains a lot
of what you see in J.
Sent from my iPhone
> On Dec 19, 2017, at 7:22 AM, Raul Miller wrote:
>
> Well.. the issue I remember was that axis n
Well.. the issue I remember was that axis notation specifically and
bracket notation in general was inconsistent with the rest of the
syntax for APL.
There were at least two ways out of this - one would be to generalize
bracket notation (I think that K might have followed that path). The
other wou
Hi all!
IBM, APL2 Programming: Language Reference
http://publibfp.boulder.ibm.com/epubs/pdf/h2110611.pdf
You were right in that a scalar was considered an array.
There is a difference in that we used the AXIS notation instead of what
is Rank? See page 55.
This means the functions did not n
In particular I failed to realize that the optimal step might not be
unique. Before I boxed
p=. p,
Date: Mon, 18 Dec 2017 09:07:52 -0500
From: Raul Miller
To: Programming forum
Subject: Re: [Jprogramming] AoC Day 11 - something is wrong
Message-ID:
Content-Type: text/plain; charset="UT
I'm not completely sure what you are getting at here, actually.
When I was learning APL, I was taught that a scalar was a rank zero
array. And it seems to me that J uses the same design.
If you look at J's 3!:1 and 3!:2 foreigns (or you can use 3!:3 instead
of 3!:1 if you prefer), you'll see that
Hi all !
In APL we had the same array model as in Fortran. An array was an array
and you indexed it like in most other languages, except that you could
use an array as index and therefore could get very varying results. The
only problems I personally experienced with this model was that you
c
39 matches
Mail list logo
