Thanks, Bo.
This variation on comb works pretty well, except it doesn't test for the
empty case, ie x = 0:
NB. these boolean verbs are quicker than multiply and add for this case
twice =: 1 & (32 b.)
or =: 23 b.
Smpscomb =: 4 : 0
k=. 2^i.>:d=.y-x
z=. <"0 k
for. i.<: x do.
z=. k or &.> ,&.>/\. twice&.> z
end.
#:/:~;z
)
ts
6!:2 , 7!:2@]
ts'3 Smps 20'
0.6071034 2.873121e8
ts'3 Smpscomb 20'
0.0001229 66976
3 (Smps -: Smpscomb) 20
1
The sort adds a little time and space; it is only necessary if you
require the same
ordering as your original example.
Cheers,
Mike
On 01/08/2021 18:31, 'Bo Jacoby' via Programming wrote:
Thanks to Mike!
comb=: 4 : 0
NB.*comb v combinations of size x from i.y
comb=: 3 : 0
comb~ y
:
k=. i.>:d=. y-x
z=. (d$<i.0 0),<i.1 0
for. i.x do. z=. k ,.&.> ,&.>/\. >:&.> z end.
3 comb 6
I need to study in order to make it do what I want. I am aware that my
programming is embarassingly inefficient.
Thanks to Raul!
2 (((= +/"1) # ]) f) 6 NB. fork, not hook!
I wonder why it is hard to code the hook. It is not important, but it is
strange.
I am studying the bessel correction. I turns out that the mean of the bessel
corrected variances of all samples is equal to the bessel corrected variance of
the population.
E =. +/%# NB. mean
Va =. E@:*:-*:@:E NB. Variance
Bc =. -.&.:%@#*Va NB. Bessel corrected variance
S =. ((((=+/"1)#])#&2#:[:i.2&^) #)#] NB. samples
Bc n=.i.8
6
E Bc |:4 S n
6
E Bc |:5 S n
6
E Bc |:6 S n
6
E Bc |:7 S n
6
E Bc |:8 S n
6
Den søndag den 1. august 2021 16.06.21 CEST skrev 'Mike Day' via Programming
<[email protected]>:
Nice work, especially persuading 13 : to return a tacit version.
I had reached the same solution as Pascal Jasmin’s, so not worth a comment in
itself.
However, it is just about worth pointing out that the filtering approach gets
expensive for larger values of y; eg 3 Smps 20 would entail filtering 2^20
rows to obtain 1140. In that case, I reckon it’s better to use the 1s indices
provided by, eg, 3 comb 20 .
comb is in addons/stats/base/combinatorial .
In case it matters!
Mike
Sent from my iPad
On 31 Jul 2021, at 20:48, Raul Miller <[email protected]> wrote:
That's a bit tricky in this case, but it can be done:
First off, the problem:
13 : '(#~x=+/"1)f y' NB. select row
4 : '(#~x=+/"1)f y'
But we can work around this by giving 13 : a variant expression:
13 : 't#~x=+/"1 t=.f y'
[ (] #~ [ = [: +/"1 ]) [: f ]
And, in this example, we can turn this into a hook through inspection:
Smps=: (] #~ [ = [: +/"1 ]) f
This could be simplified further, of course:
2 ((] #~ (= +/"1)) f) 6
and
2 (((= +/"1) # ]) f) 6
do the same computation.
I hope this helps,
--
Raul
On Sat, Jul 31, 2021 at 12:43 PM Srdjan Stipic <[email protected]> wrote:
You can use “13 : ‘…’” construct to get tacit verb.
On Sat 31. Jul 2021 at 18:00, 'Pascal Jasmin' via Programming <
[email protected]> wrote:
2 ([ (] #~ [ = +/"1@:]) f@]) 6
On Saturday, July 31, 2021, 11:52:10 a.m. EDT, 'Bo Jacoby' via Programming
<[email protected]> wrote:
Consider this code
f=.#&2#:[:i.2&^ NB. 0-1 rows
Smps=. 4 : '(#~x=+/"1)f y' NB. select rows
2 Smps 6 NB. for example
0 0 0 0 1 1
0 0 0 1 0 1
0 0 0 1 1 0
0 0 1 0 0 1
0 0 1 0 1 0
0 0 1 1 0 0
0 1 0 0 0 1
0 1 0 0 1 0
0 1 0 1 0 0
0 1 1 0 0 0
1 0 0 0 0 1
1 0 0 0 1 0
1 0 0 1 0 0
1 0 1 0 0 0
1 1 0 0 0 0
I want to rewrite it tacit
Smps=.((=+/)"1 f)#f@]
This works but is without the hook, so "f" is repeated.
How to write the hook tacitly?
I believe that this is easy, but I am stuck.
Thank you!
Bo.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
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
