This is more of an expanding list of Q idioms I have had to either assemble or remember or some combination.
- Cross product of two lists is faster in table form
q)show x:til 3
0 1 2
q)show y:2#7
7 7
q)x cross y
0 7
0 7
1 7
1 7
2 7
2 7
q) ([] x) cross ([] y)
x y
—
0 7
0 7
1 7
1 7
2 7
2 7
\t til[1000] cross 1000#5
85
\t flip value flip ([] til 1000) cross ([] x1:1000#5)
33
/and if you are happy working with it in table form
\t ([] til 1000) cross ([] x1:1000#5)
5
- Take last N observations for a column in a tableBoth of these can be used to create a list of indexes and the columns can be simply projected on to the list of indexes
/a) Intuitive way take last n from c
q) N:10; C:til 100000;
q) {[n;c]c{y-x}[til n] each til count c}[N;C] /timing 31 milli
0
1 0
2 1 0
3 2 1 0
4 3 2 1 0
….
/b) Faster way using xprev and flip
q) \t {[n;c] flip (1+til n) xprev \\: c}[10;c] /timing 9 millesec
- Create a Polynomial Function from the coefficients
poly:{[x] (‘)[wsum[x;];xexp/:[;til count x]]};
f:poly 0 1 2 3;
f til 5
0 6 34 102 228
- Count Non Null entries
All in K:
fastest:{(#x)-+/^x}
slower:{+/~^x}
slowest:{#*=^x}
Q translation:
fastest:{count[x]-sum null x}
slower:{sum not null x}
slowest:{count first group null x}
- Camel case char separated symbols:
camelCase:{[r;c]`$ssr[;r;””] each @'[h;i;:;]upper h @’ i:1+ss'[;r] h:string x}
most common case
q)c:` sv/: `a`b cross `e`fff`ggk cross `r`f
`a.e.r`a.e.f`a.fff.r`a.fff.f`a.ggk.r`a.ggk.f`b.e.r`b.e.f`b.fff.r`b.fff.f`b.ggk.r`b.ggk.f
camelCase[“.”;c]
`aER`aEF`aFffR`aFffF`aGgkR`aGgkF`bER`bEF`bFffR`bFffF`bGgkR`bGgkF
6. Progress style bars:
p:0;do[10; p+:1; system “sleep .1”;1 “\r “,p#”#”];-1 “”;
{[p]1 “\r “,(a#”#”),((75-a:7h$75*p%100)#” “),string[p],”%”;if[p=100;-1 “”;]}
7. AutoCorr:
autocorr:{x%first x:x{(y#x)$neg[y]#x}/:c-til c:count x-:avg x}
8. Index of distinct elements:
idistinct:{first each group x}
iabdb 