I've got two ways to amend a subarray in J but I don't like either of them.
(Imagine selecting a rectangle in a paint program and applying some arbitrary operation to that rectangle in place.)
t =. i. 10 10 NB. table to modify
xy=. 2 3 [ wh =. 3 2 NB. region i want want to modify
u =. -@|. NB. operation to perform on region
I can fetch the subarray and apply the operation in one step with cut (;.0):
st =. ((,./xy),:(|,./wh)) u;.0 t
Putting it back is easy enough, but seems to require building a large boxed array of indices:
(,st) (xy&+each,{;&:i./wh) } t
I also tried recursively splitting and glueing the table into four "window panes" at a time:
split =: {. ; }. NB. split y into 2 subarrays at index x
panes =: {{ 2 2$ ; |:L:0 X split&|:&.> Y split y [ 'Y X'=.x }}
glue =: [: ,&>/ ,.&.>/"1 NB. reassamble
xy panes t
┌────────┬────────────────────┐
│ 0 1 2│ 3 4 5 6 7 8 9│
│10 11 12│13 14 15 16 17 18 19│
├────────┼────────────────────┤
│20 21 22│23 24 25 26 27 28 29│
│30 31 32│33 34 35 36 37 38 39│
│40 41 42│43 44 45 46 47 48 49│
│50 51 52│53 54 55 56 57 58 59│
│60 61 62│63 64 65 66 67 68 69│
│70 71 72│73 74 75 76 77 78 79│
│80 81 82│83 84 85 86 87 88 89│
│90 91 92│93 94 95 96 97 98 99│
└────────┴────────────────────┘
NB. then split the lower right pane again,
NB. extract *its* upper left pane...
s0 =. 1 1 {:: p0 =. xy panes t
s1 =. 0 0 {:: p1 =. wh panes s0
NB. apply the operation and reassemble:
p1a =. (<u s1) (<0 0) } p1
glue (<glue p1a) (<1 1) } p0
The first approach seems to be the quicker and easier option, but it feels like there ought to be a more primitive way to apply a verb at a sub-array without extracting it, or to paste in a subarray at some coordinates without manually creating the array of indices for each element.
Have I missed a better option?