The symbol chosen to represent Mix depends on the current Migration Level.
If ⎕ML<2, Mix is represented by the symbol: ↑.
If ⎕ML≥2, Mix is represented by the symbol: ⊃.
Y may be any array whose items may be uniform in rank and shape, or differ in rank and shape. If the items of Y are non-uniform, they are extended prior to the application of the function as follows:
- If the items of Y have different ranks, each item is extended in rank to that of the greatest rank by padding with leading 1s.
- If the items of Y have different shapes, each is padded with the corresponding prototype to a shape that represents the greatest length along each axis of all items in Y.
For the purposes of the following narrative, y represents the virtual item in Y with the greatest rank and shape, with which all other items are extended to conform.
R is an array composed from the items of Y assembled into a higher-rank array with one less level of nesting. ⍴R will be some permutation of (⍴Y),⍴y.
K is an optional axis specification whose value(s) indicate where in the result the axes of y appear. There are three cases:
- K may be a scalar or 1-element vector whose value is a fractional number indicating the two axes of Y between which new axes are to be inserted for y. The shape of R is the shape of Y with the shape ⍴y inserted between the ⌊Kth and the ⌈Kth axes of Y
- K may be a scalar or 1-element vector integer whose value specifies the position of the first axis of y in the result. This case is identical to the fractional case where K (in this case) is ⌈K (in the fractional case).
- K may be an integer vector, with the same length as ⍴y, each element of which specifies the position in the result of the corresponding axis of the y.
If K is absent, the axes of y appear as the last axes of the result.
Simple Vector Examples
In this example, the shape of Y is 3, and the shape of y is 2. So the shape of the result will be a permutation of 2 and 3, i.e. in this simple example, either (2 3) or (3 2).
If K is omitted, the shape of the result is (⍴Y),⍴y.
↑(1 2)(3 4)(5 6) 1 2 3 4 5 6
If K is between 0 and 1, the shape of the result is (⍴y),⍴Y because (⍴y) is inserted between the 0th and the 1st axis of the result, i.e. at the beginning.
↑[.5](1 2)(3 4)(5 6)
1 3 5
2 4 6
If K is between 1 and 2, the shape of the result is (⍴Y),⍴y because (⍴y) is inserted between the 1st and 2nd axis of the result, i.e. at the end. This is the same as the case when K is omitted.
↑[1.5](1 2)(3 4)(5 6)
1 2
3 4
5 6
An integer K may be used instead. If ⎕ML≥2, ⊃ is used instead of ↑).
⎕ML←3
⊃(1 2)(3 4)(5 6)
1 2
3 4
5 6
⊃[1](1 2)(3 4)(5 6)
1 3 5
2 4 6
⊃[2](1 2)(3 4)(5 6)
1 2
3 4
5 6
Shape Extension
If the items of Y are unequal in shape, the shorter ones are extended:
⎕ML←3
⊃(1)(3 4)(5)
1 0
3 4
5 0
⊃[1](1)(3 4)(5)
1 3 5
0 4 0
More Simple Vector Examples:
]box on
Was OFF
'Andy' 'Geoff' 'Pauline'
┌────┬─────┬───────┐
│Andy│Geoff│Pauline│
└────┴─────┴───────┘
↑'Andy' 'Geoff' 'Pauline'
Andy
Geoff
Pauline
⎕ML←3
⊃('andy' 19)('geoff' 37)('pauline' 21)
┌───────┬──┐
│andy │19│
├───────┼──┤
│geoff │37│
├───────┼──┤
│pauline│21│
└───────┴──┘
⊃[1]('andy' 19)('geoff' 37)('pauline' 21)
┌────┬─────┬───────┐
│andy│geoff│pauline│
├────┼─────┼───────┤
│19 │37 │21 │
└────┴─────┴───────┘
⊃('andy' 19)('geoff' 37)(⊂'pauline')
┌───────┬───────┐
│andy │19 │
├───────┼───────┤
│geoff │37 │
├───────┼───────┤
│pauline│ │
└───────┴───────┘
Notice that in the last statement, the shape of the third item was extended by catenating it with its prototype.
Example (Matrix of Matrices)
In the following examples, Y is a matrix of shape (5 4) and each item of Y (y) is a matrix of shape (3 2). The shape of the result will be some permutation of (5 4 3 2).
Y←5 4⍴(⍳20)×⊂3 2⍴1
Y
┌─────┬─────┬─────┬─────┐
│1 1 │2 2 │3 3 │4 4 │
│1 1 │2 2 │3 3 │4 4 │
│1 1 │2 2 │3 3 │4 4 │
├─────┼─────┼─────┼─────┤
│5 5 │6 6 │7 7 │8 8 │
│5 5 │6 6 │7 7 │8 8 │
│5 5 │6 6 │7 7 │8 8 │
├─────┼─────┼─────┼─────┤
│9 9 │10 10│11 11│12 12│
│9 9 │10 10│11 11│12 12│
│9 9 │10 10│11 11│12 12│
├─────┼─────┼─────┼─────┤
│13 13│14 14│15 15│16 16│
│13 13│14 14│15 15│16 16│
│13 13│14 14│15 15│16 16│
├─────┼─────┼─────┼─────┤
│17 17│18 18│19 19│20 20│
│17 17│18 18│19 19│20 20│
│17 17│18 18│19 19│20 20│
└─────┴─────┴─────┴─────┘
By default, the axes of y appear in the last position in the shape of the result, but this position is altered by specifying the axis K. Notice where the (3 2) appears in the following results:
⍴⊃Y
5 4 3 2
⍴⊃[1]Y
3 2 5 4
⍴⊃[2]Y
5 3 2 4
⍴⊃[3]Y
5 4 3 2
⍴⊃[4]Y
INDEX ERROR
⍴⊃[4]Y
∧
Note that ⊃[4]Y generates an INDEX ERROR because 4 is greater than the length of the result.
Example (Vector K)
The axes of y do not have to be contiguous in the shape of the result. By specifying a vector K, they can be distributed. Notice where the 3 and the 2 appear in the following results:
⍴⊃[1 3]Y
3 5 2 4
⍴⊃[1 4]Y
3 5 4 2
⍴⊃[2 4]Y
5 3 4 2
⍴⊃[4 2]Y
5 2 4 3
Rank Extension
If the items of Y are unequal in rank, the lower rank items are extended in rank by prefixing their shapes with 1s. Each additional 1 may then be increased to match the maximum shape of the other items along that axis.
⎕ML←3
Y←(1)(2 3 4 5)(2 3⍴10×⍳8)
Y
┌─┬───────┬────────┐
│1│2 3 4 5│10 20 30│
│ │ │40 50 60│
└─┴───────┴────────┘
⍴⊃Y
3 2 4
⊃Y
1 0 0 0
0 0 0 0
2 3 4 5
0 0 0 0
10 20 30 0
40 50 60 0
In the above example, the first item (1) becomes (1 1⍴1) to conform with the 3rd item which is rank 2. It is then extended in shape to become (2 4↑1 1⍴1) to conform with the 2-row 3rd item, and 4-column 2nd item.. Likewise, the 2nd item becomes a 2-row matrix, and the 3rd item gains another column.