Array Comprehensions 101

Creating array comprehensions in Julia is both easy and efficient. You can construct a vector or array (the terms are interchangeable in Julia) by using an inline, reversed for-loop structure within brackets. This syntax will likely resonate with programmers coming from Python, where similar patterns exist:

myv = [x + 1 for x in 1:5]

Here, the expression preceding the "for" keyword dictates the output of the comprehension. Each element in the new vector is generated by this simple algorithm:

myv = [x + 1 for x in 1:5]

Result:

5-element Vector{Int64}:

2

3

4

5

6

You can also use standard iterators and methods to achieve similar results, providing a tuple for output variable names:

myv = [x + y for (x, y) in zip(5:10, 1:5)]

Multi-dimensional arrays can be constructed in the same way:

[(x, y) for x = 1:3, y = 1:2]

Result:

3×2 Array{Tuple{Int64,Int64},2}:

(1,1) (1,2)

(2,1) (2,2)

Moreover, comprehensions can be annotated, enhancing performance as discussed in my earlier article on the use of annotations. Combining annotated comprehensions with standard ones can significantly enhance Julia's already powerful comprehension features.

Array Comprehension Advanced Features

Things become even more intriguing with comprehensions, as they offer unexpected capabilities. For instance, you can introduce conditions either at the beginning or the end of the comprehension:

[if x > 5 x else 1 end for x in 1:20]

[x^2 for x in 0:9 if x > 5]

Another unique aspect of comprehensions is the ability to use "begin" and "end" to expand their functionality, allowing for code that doesn't need to remain in-line. This feature is particularly useful in cases involving arithmetic, enabling you to maintain a shorter compile time without sacrificing readability:

[begin

s = sum(1:20)

n = length(1:20)

x = s / n

end for x in 1:20]