Conformability

From APL Wiki
Jump to navigation Jump to search

Conformability describes the conditions that must be satisfied by arguments to dyadic scalar functions. The arguments must match in shape, taking into account scalar extension and, when supported, singleton extension or leading axis agreement. The criteria which determine the result shape are often considered part of conformability as well. The Each operator also checks conformability when called dyadically, and some functionality, such as multiple assignment, might use modified conformability rules.

Two arguments are said to conform when they have matching shapes, or one of them is extendible to match the other's shape. Possible rules for extensibility are:

The result shape associated with such arguments can then be determined:

  • If the shapes matched, it is that shape
  • If exactly one argument was extended, it is the other argument's shape
  • Otherwise, it is the shape of the argument with the largest rank (this condition is only possible with singleton extension).

Frame conformability

Main article: Frame agreement

The Rank operator and function rank check for conformability not on the entire shapes of arguments, but only on their frames. The same procedure applies to determine whether two frames conform, and if function f's result shape is known, then the result shape of f⍤l r can be found by appending it to the shape obtained from conformability on the frames computed from ranks l and r. Rather than copying elements of the extended argument to match the other one, cells are copied.

Frame conformability may not allow singleton extension even when ordinary conformability supports it: Dyalog APL, SHARP APL, and A+ have singleton extension for scalar dyadics but not for Rank (and J doesn't implement singleton extension). Of these, J and A+ use leading axis agreement while SHARP and Dyalog use only empty-frame extension, the equivalent of scalar extension. However, NARS2000 implements Rank with singleton extension for frames.

External links

APL features [edit]
Built-ins Primitives (functions, operators) ∙ Quad name
Array model ShapeRankDepthBoundIndex (Indexing) ∙ AxisRavelRavel orderElementScalarVectorMatrixSimple scalarSimple arrayNested arrayCellMajor cellSubarrayEmpty arrayPrototype
Data types Number (Boolean, Complex number) ∙ Character (String) ∙ BoxNamespaceFunction array
Concepts and paradigms Conformability (Scalar extension, Leading axis agreement) ∙ Scalar function (Pervasion) ∙ Identity elementComplex floorArray ordering (Total) ∙ Tacit programming (Function composition, Close composition) ∙ GlyphLeading axis theoryMajor cell search
Errors LIMIT ERRORRANK ERRORSYNTAX ERRORDOMAIN ERRORLENGTH ERRORINDEX ERRORVALUE ERROREVOLUTION ERROR