The Standard ML Basis Library


The Array2 structure


Synopsis

signature ARRAY2  (* OPTIONAL *)
structure Array2 :> ARRAY2  (* OPTIONAL *)

The Array2 structure provides polymorphic mutable 2-dimensional arrays. As with 1-dimensional arrays, these arrays have the equality property that two arrays are equal if they are the same array, i.e., created by the same call to a primitive array constructor such as array, fromList, etc.; otherwise they are not equal. This also holds for arrays of zero length. Thus, the type ty array admits equality even if ty does not.

The elements of 2-dimensional arrays are indexed by pair of integers (i,j) where i gives the row index, and i gives the column index. As usual, indices start at 0, with increasing indices going from left to right and, in the case of rows, from top to bottom.


Interface

eqtype 'a array
type 'a region = {
                   base : 'a array,
                   row : int,
                   col : int,
                   nrows : int option,
                   ncols : int option
                 }
datatype traversal = RowMajor | ColMajor

val array : int * int * 'a -> 'a array
val fromList : 'a list list -> 'a array
val tabulate : traversal
                 -> int * int * (int * int -> 'a)
                   -> 'a array

val sub : 'a array * int * int -> 'a
val update : 'a array * int * int * 'a -> unit

val dimensions : 'a array -> int * int
val nCols      : 'a array -> int
val nRows      : 'a array -> int

val row : 'a array * int -> 'a Vector.vector
val column : 'a array * int -> 'a Vector.vector

val copy : {
               src : 'a region,
               dst : 'a array,
               dst_row : int,
               dst_col : int
             } -> unit

val appi : traversal
             -> (int * int * 'a -> unit)
               -> 'a region -> unit
val app  : traversal -> ('a -> unit) -> 'a array -> unit
val foldi : traversal
              -> (int * int * 'a * 'b -> 'b)
                -> 'b -> 'a region -> 'b
val fold  : traversal
              -> ('a * 'b -> 'b-> 'b -> 'a array -> 'b
val modifyi : traversal
                -> (int * int * 'a -> 'a)
                  -> 'a region -> unit
val modify  : traversal -> ('a -> 'a-> 'a array -> unit

Description

type 'a region = {
                   base : 'a array,
                   row : int,
                   col : int,
                   nrows : int option,
                   ncols : int option
                 }
This type specifies a rectangular subregion of a 2-dimensional array. If ncols equals SOME(w), with 0 <= w, the region includes only those elements in columns with indices in the range from col to col + (w - 1), inclusively. If ncols is NONE, the region includes only those elements lying on or to the right of column col. A similar interpretation holds for the row and nrows fields. Thus, the region corresponds to all those elements with position (i,j) such that i lies in the specified range of rows and j lies in the specified range of columns.

A region reg is said to be valid if it denotes a legal subarray of its base array. More specifically, reg is valid if

0 <= #row reg <= nRows (#base reg)
when #nrows reg = NONE, or
0 <= #row reg <= (#row reg)+nr <= nRows (#base reg)
when #nrows reg = SOME(nr), and the analogous conditions hold for columns.

datatype traversal = RowMajor | ColMajor
This type specifies a way of traversing a region. Specifically, RowMajor indicates that, given a region, the rows are traversed from left to right (smallest column index to largest column index), starting with the first row in the region, then the second, and so on until the last row is traversed. ColMajor reverses the roles of row and column, traversing the columns from top down (smallest row index to largest row index), starting with the first column, then the second, and so on until the last column is traversed.

array (r, c, init)
creates a new array with r rows and c columns, with each element initialized to the value init. If r < 0, c < 0 or the resulting array would be too large, the Size exception is raised.

fromList l
creates a new array from a list of a list of elements. The elements should be presented in row major form, i.e., hd l gives the first row, hd (tl l) gives the second row, etc. This raises the Size exception if the resulting array would be too large or if the lists in l do not all have the same length.

tabulate trv (r, c, f)
creates a new array with r rows and c columns, with the (i,j)(th) element initialized to f (i,j). The elements are initialized in the traversal order specified by trv. If r < 0, c < 0 or the resulting array would be too large, the Size exception is raised.

sub (arr, i, j)
returns the (i,j)(th) element of the array arr. If i < 0, j < 0, nRows arr <= i, or nCols arr <= j, then the Subscript exception is raised.

update (arr, i, j, a)
sets the (i,j)(th) element of the array arr to a. If i < 0, j < 0, nRows arr <= i, or nCols arr <= j, then the Subscript exception is raised.

val dimensions : 'a array -> int * int
val nCols : 'a array -> int
val nRows : 'a array -> int
These functions return size information concerning an array. nCols returns the number of columns, nRows returns the number of rows, and dimension returns a pair containing the number of rows and the number of columns of the array. The functions nRows and nCols are respectively equivalent to #1 o dimensions and #2 o dimensions

row (arr, i)
returns row i of arr. If (nRows arr) <= i or i < 0, this raises Subscript.

column (arr, j)
returns column j of arr. This raises Subscript if j < 0 or nCols arr <= j.

copy {src, dst, dst_row, dst_col}
copies the region src into the array dst, with the element at position (#row src, #col src) copied into the destination array at position (dst_row,dst_col). If the source region is not valid, then the Subscript exception is raised. Similarly, if the derived destination region (the source region src translated to (dst_row,dst_col)) is not valid in dst, then the Subscript exception is raised.
Implementation note:

The copy function must correctly handle the case in which the #base src and the dst arrays are equal, and the source and destination regions overlap.



appi tr f reg
app tr f arr
These functions apply the function f to the elements of an array in the order specified by tr. The more general appi function applies f to the elements of the region reg and supplies both the element and the element's coordinates in the base array to the function f. If reg is not valid, then the exception Subscript is raised.

The function app applies f to the whole array and does not supply the element's coordinates to f. Thus, the expression app tr f arr is equivalent to:

let
  val range = {base=arr,row=0,col=0,nrows=NONE,ncols=NONE}
in
  appi tr (f o #3) range
end


foldi tr f init reg
fold tr f init arr
These functions fold the function f over the elements of an array arr, traversing the elements in tr order, and using the value init as the initial value. The more general foldi function applies f to the elements of the region reg and supplies both the element and the element's coordinates in the base array to the function f. If reg is not valid, then the exception Subscript is raised.

The function fold applies f to the whole array and does not supply the element's coordinates to f. Thus, the expression fold tr f init arr is equivalent to:

foldi tr (fn (_,_,a,b) => f (a,b)) init 
           {base=arr, row=0, col=0, nrows=NONE, ncols=NONE}


modifyi tr f reg
modify tr f arr
These functions apply the function f to the elements of an array in the order specified by tr, and replace each element with the result of f. The more general modifyi function applies f to the elements of the region reg and supplies both the element and the element's coordinates in the base array to the function f. If reg is not valid, then the exception Subscript is raised.

The function modify applies f to the whole array and does not supply the element's coordinates to f. Thus, the expression modify tr f arr is equivalent to:

let
  val range = {base=arr,row=0,col=0,nrows=NONE,ncols=NONE}
in
  modifyi tr (f o #3) 
end


See Also

Array, MONO_ARRAY2

Discussion

Note that the indices passed to argument functions in appi, foldi, and modifyi are with respect to the underlying matrix and not based on the region. This is different from the convention for the analogous functions on 1-dimensional slices.

Rationale:

It was clear that 2-dimensional arrays needed to be provided, but the interface is fairly rudimentary, largely due to the lack of experience with their uses in SML programs. Thus, we kept regions concrete, as opposed to the slice types, their 1-dimensional cousins. In addition, we felt it best, at this time, to avoid picking among the vast number of possible matrix functions.

Implementation note:

Unlike one-dimensional types, the signature for 2-dimensional arrays does not specify any bounds on possible arrays. Implementations should support a total number of elements that is at least as large as the total number of elements in the corresponding single dimension array type.


[ Top | Parent | Contents | Index | Root ]

Generated April 12, 2004
Last Modified June 11, 2000
Comments to John Reppy.


This document may be distributed freely over the internet as long as the copyright notice and license terms below are prominently displayed within every machine-readable copy.

Copyright © 2004 AT&T and Lucent Technologies. All rights reserved.

Permission is granted for internet users to make one paper copy for their own personal use. Further hardcopy reproduction is strictly prohibited. Permission to distribute the HTML document electronically on any medium other than the internet must be requested from the copyright holders by contacting the editors. Printed versions of the SML Basis Manual are available from Cambridge University Press. To order, please visit www.cup.org (North America) or www.cup.cam.ac.uk (outside North America).