MathGroup Archive 2005

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: extra data in expressions

  • To: mathgroup at smc.vnet.net
  • Subject: [mg53409] Re: [mg53367] extra data in expressions
  • From: "Wolf, Hartmut" <Hartmut.Wolf at t-systems.com>
  • Date: Tue, 11 Jan 2005 01:30:38 -0500 (EST)
  • Sender: owner-wri-mathgroup at wolfram.com

>-----Original Message-----
>From: joshuahorowitz at comcast.net [mailto:joshuahorowitz at comcast.net] 
To: mathgroup at smc.vnet.net
>Sent: Saturday, January 08, 2005 8:40 AM
>Subject: [mg53409] [mg53367] extra data in expressions
>
>Here's a rather abstract foundations question:
>
>I once unthinkingly assumed that an expression was determined 
>by its head and parts. This is of course very wrong: any two 
>expressions with head Real, no matter how different, have head 
>Real and no parts. Same goes for Symbol, Integer, String, and 
>all other "atomic" expression heads. This means that certain 
>expressions can (and must) hold hidden extra data.
>
>My question is whether there are any consistent, logical 
>semantics for this extra data. Can an expression contain both 
>parts AND extra data?
>Which built-in symbols (or rather, expressions with head one 
>of these built-in symbols) contain extra data; is there a list 
>somewhere? Can user-defined symbols contain user-defined extra 
>data? Why does Head[String @@ 1] return Integer? What else 
>could it possibly return?
>
>I'm starting to suspect that "everything is an expression" 
>involves some pretty ugly exceptions.
>
>

Things stay simple, if we avoid implementation fantasies like "hold
hidden extra data".

A Mathematica expression either is atomic or composed by a head (which
in turn may be any Mathematica expression) and a sequence (which might
be empty) of elements (each in turn may be any Mathematica expression).

The Mathematica function Head returns the head-component if it is
composed, else returns a Symbol which characterizes the atom (think of
it as a "data type"), which is most important for pattern matching
purposes.

In[23]:= Head["hello world"[1][2]]
Out[23]=
"hello world"[1]

The implementation of an atom (even if it appears to be "composed" to
our eyes, as e.g. Rational[7,8] or 2.1 + 0.2\[ImaginaryI]) is completely
unaccessible to us.


You can recognize an atomic expression by the function AtomQ:

In[15]:= AtomQ /@ 
{3, 2.1 + 0.2\[ImaginaryI], 7/8, Anything, "{1,2}", \[Pi], {1, 2}, 1[2]}
Out[15]=
{True, True, True, True, True, True, False, False}


In[31]:= Head /@ 
{3, 2.1 + 0.2\[ImaginaryI], 7/8, Anything, "{1,2}", \[Pi], {1, 2}, 1[2]}
Out[31]=
{Integer, Complex, Rational, Symbol, String, Symbol, List, 1}


Access of components only is possible if the expression is composed:

In[17]:= Off[First::"normal"]

In[19]:=
First /@ {3, 2.1 + 0.2\[ImaginaryI], 7/8, Anything, "{1,2}", \[Pi], {1,
2}, 
    1[2]}
Out[19]=
{First[3], First[2.1 + 0.2*I], First[7/8], 
  First[Anything], First["{1,2}"], First[Pi], 1, 2}


(* but Part[expression, 0] just gives Head[expression] *)



Apply at atoms reproduces the atom (for any "function"):

In[14]:=
Anything @@@ {3, 2.1 + 0.2\[ImaginaryI], 7/8, Anything, 
    "{1,2}", \[Pi], {1, 2}, 1[2]}
Out[14]=
{3, 2.1 + 0.2*I, 7/8, Anything, "{1,2}", Pi, 
  Anything[1, 2], Anything[2]}


(This explains why Head[String @@ 1] returns Integer.)

--
Hartmut Wolf


  • Prev by Date: Re: GUIKit: standalone error
  • Next by Date: Re:Re: FullSimplify with Assumptions
  • Previous by thread: Re: extra data in expressions
  • Next by thread: Plotting dates on the x-axis