In computer programming, array slicing is an operation that extracts a subset of elements from an array and packages them as another array, possibly in a different dimension from the original.
Common examples of array slicing are extracting a substring from a string of characters, the "ell" in "hello", extracting a row or column from a two-dimensional array, or extracting a vector from a matrix.
Depending on the programming language, an array slice can be made out of non-consecutive elements. Also depending on the language, the elements of the new array may be aliased to (i.e., share memory with) those of the original array.
For "one-dimensional" (single-indexed) arrays — vectors, sequence, strings etc. — the most common slicing operation is extraction of zero or more consecutive elements. Thus, if we have a vector containing elements (2, 5, 7, 3, 8, 6, 4, 1), and we want to create an array slice from the 3rd to the 6th items, we get (7, 3, 8, 6). In programming languages that use a 0-based indexing scheme, the slice would be from index 2 to 5.
Reducing the range of any index to a single value effectively eliminates that index. This feature can be used, for example, to extract one-dimensional slices (vectors) or two-dimensional slices (rectangular matrices) from a three-dimensional array. However, since the range can be specified at run-time, type-checked languages may require an explicit (compile-time) notation to actually eliminate the trivial indices.
General array slicing can be implemented (whether or not built into the language) by referencing every array through a dope vector or descriptor — a record that contains the address of the first array element, and then the range of each index and the corresponding coefficient in the indexing formula. This technique also allows immediate array transposition, index reversal, subsampling, etc. For languages like C, where the indices always start at zero, the dope vector of an array with d indices has at least 1 + 2d parameters. For languages that allow arbitrary lower bounds for indices, like Pascal, the dope vector needs 1 + 3d entries.