Pair
Pair (⍮
) or Juxtapose is a primitive ambivalent function that constructs a vector of its arguments, present in Extended Dyalog APL, dzaima/APL, and BQN (as ⋈
). The monadic case is referred to as Half Pair in Extended Dyalog APL, and is similar to Enclose except that it gives a result of rank 1 rather than 0. This case is called Enlist in BQN, as well as K, where it is the monadic case of ,
(K has no dyadic Pair primitive). Dyadic Pair is similar to the Link primitive in SHARP APL and J, but Link has an asymmetric definition that adds its left argument to a vector right argument rather than creating a new vector.
Pair can be implemented as the dfn {⍺←,⊂ ⋄ ⍺ ⍵}
, or the catenation of the enclosed arguments, written ,⍥⊂
using Over.[1] Pair aids in constructing nested arrays in tacit programming, where stranding by juxtaposition is not available.
Common usage
Its plain usage is in pairing up two parallel values:
6 7 8(+⍮-)3 ┌───────┬─────┐ │9 10 11│3 4 5│ └───────┴─────┘
It can also be combined with Mix to increase rank rather than depth:
6 7 8(+↑⍤⍮-)3 9 10 11 3 4 5
History
Link precedes Pair significantly, as it was defined in SHARP APL in 1983,[2] and included in the initial version of J in 1990.
Pair appeared first in array languages based on the list model, where it also covers the functionality of Promote and Laminate. K, released in the 1990s, defines monadic ,
as enlist, but dyadic ,
as catenate. Ambivalent Pair appears in I as ;
since 2012; no name is given in the documentation, but the implementation uses "itemize" (the same name as J's Promote) and "cross".
Pair was defined as ⍮
in Extended Dyalog APL in 2018, and then adopted into dzaima/APL. It was added to BQN with the glyph ⋈
in 2021.
Documentation
- BQN (as
⋈
)
References
- ↑ "Composition and Enclosure" § Composition Operators. SATN-41, 1981-06-20.
- ↑ "Language Extensions of May 1983". SATN-45, 1983-05-02.