Vocabulary/atco

>> <<   Back to: Vocabulary Thru to: Dictionary

[x] u@:v y

At

Conjunction

Rank Infinity -- operates on [x and] y as a whole -- WHY IS THIS IMPORTANT?

---

Forms the composition (u@:v) of two verbs u and v .

The result is a tacit verb equivalent to  u [x] v y (where [x] represents an optional x-argument)

   y =: i. 6        NB. a sample list (the contents don't matter)
   <: # y           NB. count the items in y and subtract 1
5
   compo =: <:@:#   NB. Make a new verb: the "composition" of (<:) and (#)
   compo y
5

You can safely chain verbs using (@:) with minimal need for parentheses

   1 2 3 +/@:*:@:- 2 2 2   NB. sum of squared differences
2
   +/ *: 1 2 3 - 2 2 2     NB. same thing, without using (@:)
2

See: More Information for a visual comparison of At (@:), Atop (@), Compose (&) and Appose (&:).

---

Common uses

1. Implement: f(g(x)) -- the mathematical composition of the two functions: f and g.

   mean=: +/ % #
   cat=: ,&1"1
   ]z=: i.2 3       NB. sample noun
0 1 2
3 4 5
   cat z            NB. appends 1 to each row of z
0 1 2 1
3 4 5 1

   mean@:cat z      NB. mean of the COLUMNS of cat z
1.5 2.5 3.5 1

   mean@:cat b.0    NB. rank of (mean@:cat)
_ _ _

But see Rank in a hurry: an insidious rank problem for how and when these different methods give different results.

---

Related Primitives

Atop (@), Compose (&), Appose (&:), Hook ((u v)), Fork ((f g h))

---

More Information

1. Unlike Atop (@), the rank of (u@:v) is infinite, irrespective of the ranks of u and v. This means that u will be executed on the entire result of v, after individual results of v have been collected and filled.

An illustration of this explanation follows, extracted from the series of diagrams referenced below:

The difference between At (u@:v) and Atop (u@v) is shown in the first two columns of the diagram below:

The above diagram also shows Appose (&:) and Compose (&) for overall comparison.

These four are also visualized along with verb application and rank here: a series of flow diagrams

2. u@:v is called for when the rank of v is less than the ranks of an argument, but you want to apply u to the entire result of v.

Thus, in the "sum of squared-differences" example in the introduction, we needed (@:) not (@)

   1 2 3 +/@:*:@:- 2 2 2
2
   1 2 3 +/@*:@-   2 2 2   NB. different result using (@) in place of (@:)
1 0 1

Because x - y has rank 0, the entire verb  +/@*:@- was applied to each atom of x and y individually, making +/ useless because it is now applied to each number independently.

3. So what's the difference between Atop (@) and Compose (&) ?

None at all, for the monads (u@v) and (u&v)

  u&v y ↔ u v y
  u@v y ↔ u v y

But the dyads are different

  x u&v y ↔ (v x) u (v y)
  x u@v y ↔ u x v y

According to the J Dictionary -- &: is equivalent to & except that the ranks of the resulting function are infinite; the relation is similar to that between @: and @

---