Arrays, indexing, and broadcasting

The idea

Ordinary Julia arrays use one-based indices, and multidimensional arrays accept one index per dimension. An array is mutable: assigning to an indexed position changes the existing object. Assigning an array to another variable creates another reference to that same object. copy creates a separate outer array, but nested mutable elements are still shared, so choose a deeper copying policy only when the data model requires it.

Broadcasting applies an operation across compatible shapes. A dotted call such as abs.(values) applies a function elementwise, and a dotted operator such as values .+ 2 combines every element with a scalar. Plain matrix operations have their own algebraic meaning, so dots are a deliberate part of the expression. eachindex provides suitable indices for generic array iteration instead of assuming every array uses a particular index range.

original = [2, 4, 6]
separate = copy(original)
separate[1] = 9
shifted = original .+ 3
println(original[1])
println(join(shifted, ","))
grid = [1 2; 3 4]
println(grid[2, 1])

2
5,7,9
3

Changing the copied outer array leaves the original numbers untouched. Broadcasting adds three to each original entry, and the matrix lookup selects row two, column one.