> Can you post your commands? It would be an interesting exercise trying
to optimize for performance.
> -Dan
I will be happy to share my commands, with the disclaimer that I'm a novice
at J. My approach is simply what I've been able to cobble together in
isolation.
Suppose you have the following array of factors, each row being a box:
1.75481 1.62128 1.9037 2.32099 2.23944 2.92083 0.929577 3.41011
1.87385 1.50193 1.46014 1.23175
1.33699 1.25722 1.25292 1.45213 1.66981 1.11555 1.94697 1.65239 1.81117
2.06941 2.37717 0
1.09631 1.30063 1.37888 1.06136 1.11959 1.35294 1.38781 1.20239 1.22937
1.38261 0 0
1.19252 1.19422 1.05293 1.26402 1.14045 1.23251 1.24019 1.16252 1.20354
0 0 0
1.07837 0.998656 1.03209 1.06962 1.24631 1.06288 1.05953 1.07418 0
0 0 0
1.03052 1.02557 1.05389 1.10402 1.02959 1.03608 1.06828 0 0
0 0 0
1.00987 1.02625 1.08751 0.964162 1.03218 1.04666 0 0 0
0 0 0
1.00419 1.04348 0.997288 1.01199 1.00334 0 0 0 0
0 0 0
1.00139 1 1.02448 1 0 0 0 0 0
0 0 0
1 1 1 0 0 0 0 0 0
0 0 0
1 1 0 0 0 0 0 0 0
0 0 0
1 0 0 0 0 0 0 0 0
0 0 0
I want to find every possible product down the columns and store the
results in a box. The only exception is that I don't want any products
where the result is zero. In other words, every nonzero term in the first
row multiplied by every nonzero term in the second row multiplied by .....
Because there are 12 nonzero factors in the first row, 11 in the second,
and so on, there will be 12! total products in the resulting box.
Here's the gerund I came up with to do it:
convolvefactors=:multiplyfirsttwoboxesandappend`justmultiplyfirsttwoboxes@
.arethereonlytwoboxes
where:
arethereonlytwoboxes=:{.@(=&2@$)
justmultiplyfirsttwoboxes=:<@(sparse@(>@(0&{)) */ sparse@(>@(1&{)))
multiplyfirsttwoboxesandappend=:<@(sparse@(>@(0&{)) */ sparse@(>@(1&{))) ,
2&}.
sparse=:5 & $.@ $.
In this case, convolvefactors(^:11) applied to the array above will give
the result I want. (But remember, each row must be a box. I displayed it
as an array above because I couldn't figure out how to paste the boxes).
This function, convolvefactors(^:11), terminates in about 30 minutes on my
little Mac Mini. It doesn't even get out of the gate on my Windows 7
laptop; I get an immediate "Out of Memory" error.
By the way, I stumbled upon the "sparse" function above when I was trying
to get rid of the 0's. I realized after the fact that I could have done
something like this:
removezeros=:(#~(~:&0)) (&.>)
Cheers,
Jeremy
On Sun, May 26, 2013 at 9:26 AM, Jeremy Smith
<[email protected]>wrote:
> Hi everyone,
>
> I was trying to take 9 boxes and multiply them all together, storing the
> output in a single box. The first of the 9 boxes contains 12 numbers, the
> second contains 11, and so on, so that the output of this calculation will
> be a single box with 12*11*10*9*8*7*6*5*4=79,833,600 numbers.
>
> When I do this on Jgtk701 for Mac, the operation completes successfully
> and requires about 10 seconds. On my windows machine I receive an "Out of
> Memory" error. I'm using exactly the same series of commands.
>
> Do these errors vary by machine, or operating system? My windows laptop
> has 8GB of RAM compared to my Mac's puny 4GB.
>
> Thanks,
> Jeremy
>
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
