Subsections


Multiples

A multiple consists of a number of elements, each of which have the same mode (sometimes known as the base mode). The mode of a multiple consists of the mode indicant for each element preceded by brackets, and is said “row of mode”. For example, here is an identity declaration of a row of CHAR multiple:

   []CHAR a = "abcd"

The phrase on the left-hand side of the equals symbol is read “row of car a”. The phrase on the right-hand side of the equals symbol is the denotation of a value whose mode is []CHAR. Spaces can, of course, appear before, between or after the brackets.

Multiples of mode []CHAR are so common that this denotation was devised as a kind of shorthand. The maximum number of elements in a multiple is equal to the maximum positive integer (max int), although in practice, your program will be limited by the available memory. The denotation of a []CHAR may extend over more than one line. There are two ways of doing this. You can simply write the denotation on more than one line in which case every character “between” the starting and ending quote characters is included except the newline characters, or you can split the denotation with quote characters at the end of one line and at the start of the continuation of the denotation on the next line. Here are two declarations which exemplify these rules:

   []CHAR long1 = "The first stage in the develo
   pment of a new program consists of analysing 
   the problem that the program must solve.";
   []CHAR long2 = "The first stage in the "
              "development of a new "
              "program consists of "
              "analysing the problem "
              "that the program must "
              "solve."

Notice that the second method is neater because you can indent the subsequent parts of the denotation. Everything “between” the second and third quote characters and “between” the fourth and fifth quote characters is ignored, although you should not put anything other than spaces or tabs and newlines there. If you want to place a quote character (") in the denotation, you must double it, just as in the character denotation. Here are two []CHAR denotations, each containing two quote characters:

   []CHAR rca = """Will you come today?""",
          rcb = "The minority report stated "
                "that ""in their opinion""";

The repeated quote characters are different from the quote characters which chain the two parts of the denotation of rcb.

Row-displays

Multiples of other modes cannot be denoted as shown above, but use a construct called a row-display. A row-display consists of none or two or more units separated by commas and enclosed by parentheses (or BEGIN and END). Here is the identity declaration for a written using a row-display:

   []CHAR a = ("a","b","c","d")

It is important to note that the units in the row-display could be quite complicated. For example, here is another declaration for a multiple with mode []CHAR:

   []CHAR b = ("a","P",REPR 36,"""")

In each of these two declarations, the number of elements is 4.

Here are identity declarations for a multiple of mode []INT and a multiple of mode []REAL:

   []INT  c = (1, 2+3, -2**4, 7, -11, 2, 1);
   []REAL d = (1.0, -2.9, 3e4, -2e-2, -5)

Note that the last unit of the row-display for c has the same value as the first unit. In a multiple of mode []INT, the individual elements can have any value of mode INT: that is to say, any integer or formula yielding an integer. In d, the unit yielding the last element is written as a formula yielding a value of mode INT. Since the context of the row-display is strong (because it occurs on the right-hand side of an identity declaration), this context is passed on to its constituent units. Thus, the context of the formula is also strong, and so the value yielded by the formula is widened to yield -5.0.

An empty row-display can be used to yield a flat multiple (one with no elements). For example, here is an identity declaration using an empty row-display:

   []REAL empty = ()

The denotation for a flat []CHAR is used in the identity declaration

   []CHAR none = ""

A multiple can also have a single element. However, a row-display cannot have a single unit (because it would be an enclosed clause, which is a different construct). In this case, we use a unit (or a formula, which is another kind of unit) for the only element, and the value of that unit is coerced to a multiple with a single element using the rowing coercion. For example,

   []INT ri = 4

yields a multiple with one element. An enclosed clause can be used instead:

   []INT ri1 = (4)

since an enclosed clause is also a unit (see section 10.4).

Rowing can only occur in strong contexts (and the right-hand side of an identity declaration is a strong context). Here is another example:

   []CHAR rc = "p"

A row-display can only be used in a strong context. Because the context of an operand is firm, a row-display cannot appear in a formula (but there is a way round this, see section 10.5). The shorthand denotation for a []CHAR is not a row-display and so does not suffer from this limitation.

Dimensions

One of the properties of a multiple is its number of dimensions. All the multiples declared so far have one dimension. The number of dimensions affects the mode. A two-dimensional multiple of integers has the mode

   [,]INT

(said “row-row-of-int”), while a 3-dimensional multiple of reals (real numbers) has the mode

   [,,]REAL

Note that the number of commas is always one less than the number of dimensions. In Algol 68, multiples of any number of dimensions can be declared.4.1

To cater for more than one dimension, each of the units of a row-display can also be a row-display. For example, the row-display for a multiple with mode [,]INT could be

   ((1,2,3),(4,5,6))

The fact that this is the row-display for a 2-dimensional multiple would be clearer if it were written

   ((1,2,3),
    (4,5,6))

For two dimensions, it is convenient to talk of “rows” and “columns”. Here is an identity declaration using the previous row-display:

   [,]INT e = ((1,2,3),
               (4,5,6))

The first “row” of e is yielded by the row-display (1,2,3) and the second “row” is yielded by (4,5,6). The first “column” of e is yielded by the row-display (1,4), the second “column” by (2,5) and the third “column” by (3,6). Note that the number of elements in each “row” is the same, and the number of elements in each “column” is also the same, but that the number of “rows” and “columns” differ. We say that e is a rectangular multiple. If the number of “rows” and “columns” are the same, the multiple is said to be square. Here is an identity declaration for a square multiple:

   [,]CHAR f = (("a","b","c"),
                ("A","B","C"),
                ("1","2","3"))

All square multiples are also rectangular, but the converse is not true. Note that in the row-display for a multi-dimensional multiple of characters, it is not possible to use the special denotation for a []CHAR.

The base mode of a multiple can be any mode, including another row mode. For example:

   [][]CHAR days =
      ("Monday","Tuesday","Wednesday",
       "Thursday","Friday","Saturday",
       "Sunday")

The mode is said “row of row of CHAR”. Note that days is one-dimensional, each element consisting of a one-dimensional []CHAR. The shorthand denotation for a []CHAR can be used in this case. Because the base mode is []CHAR, the individual []CHARs can have different lengths. Here is another example using integers:

   [][]INT trapezium = ((1,2),(1,2,3),(1,2,3,4))


Subscripts and bounds

Each element of a multiple has one integer associated with it for each dimension. These integers increase by 1 from the first to the last element in each dimension. For example, in the declaration

   []INT r1 = (90,95,98)

the integers associated with the elements are [1], [2] and [3] (see the next section for an explanation of why the integers are written like this). Remember that the first element in a row-display always has an associated integer of [1]. These integers are known as subscripts4.2 or indexers. Thus the subscript of 98 in r1 is [3]. In the two-dimensional multiple

   [,]INT r2 = ((-40, -30, -20),
                (100, 130, 160))

the subscripts for -40 are [1,1] and the subscripts for 160 are [2,3].

We say that the lower bound of r1 is 1, and its upper bound is 3. The multiple r2 has a lower bound of 1 for both the first and second dimensions, an upper bound of 2 for the first dimension (2 “rows”) and an upper bound of 3 for the second dimension (3 “columns”). We shall write the bounds of r1 and r2 as [1:3] and [1:2,1:3] respectively. The bounds of a flat multiple, unless specified otherwise (see the section on trimming), are [1:0].

The bounds of a multiple can be interrogated using the operators LWB for the lower bound, and UPB for the upper bound. The bounds of the first, or only, dimension can be interrogated using the monadic form of these operators. For example, using days defined above, LWB days yields 1, and UPB days yields 7. Where the multiple is multi-dimensional, the bounds are interrogated using the dyadic form of LWB and UPB: the left operand is the dimension while the right operand is the identifier of the multiple. For example, 1 UPB r2 yields 2 and 2 UPB r2 yields 3. The priority of the dyadic operators is 8.


Exercises

3.1
What is wrong with the following identity declarations? Ans[*]
(a)
()CHAR c1 = "Today"
(b)
[]CHAR c2 = 'Yesterday'
(c)
[]INT i1 = (1, 2.0, 3)
3.2
Using the identifier first 4 odd numbers, write an appropriate identity declaration. Ans[*]
3.3
Given the identity declarations
   []CHAR s = "abcdefgh";
   []REAL r = (1.4e2, 3.5e-1, -4.0);
   [,]INT t = ((2,3,5),
               (7,11,13),
               (17,19,23))

what is the value of the following: Ans[*]

(a)
UPB s
(b)
LWB r
(c)
2 UPB t - 1 LWB t + 1
3.4
Write the formulæ which give the upper and lower bounds of each of the following multiples: Ans[*]
(a)
[,,]REAL a
(b)
[]REAL b = ()

Sian Mountbatten 2012-01-19