6

I don't know if the question trivial or not (if so I will delete it), but here is what I have:

I know that 

x x
(*  x^2  *)

Looking at the full form gives

FullForm[Times[x, x]]
(* Power[x, 2] *)

It is clear that the Times head has been changed to Power head. I just wonder how does this happen. Can I control this behavior to stop this conversion between heads?

m_goldberg
  • 107,779
  • 16
  • 103
  • 257
Basheer Algohi
  • 19,917
  • 1
  • 31
  • 78

2 Answers2

7

We can use this function to see that the conversion is not made during parsing:

parseString[s_String, prep : (True | False) : True] := 
  FrontEndExecute[FrontEnd`UndocumentedTestFEParserPacket[s, prep]]

parseString["x x"]

{BoxData[RowBox[{"x", "x"}]], StandardForm}

We can use this to see that the conversion does not take place while converting boxes to StandardForm:

ToExpression[RowBox[{"x", "x"}], StandardForm, HoldComplete] // FullForm

HoldComplete[Times[x,x]]

We can see that the rule is applied as part of the normal evaluation sequence:

Times[x, x] // TracePrint

x x

Times

x

x

(x^2)

Power

x

2

You cannot generally "stop this conversion" but you can temporarily disable the rules for Times using Block:

Block[{Times},
 ToString[x x]
]

"Times[x, x]"

You can also Hold or HoldComplete expressions and apply your own rules.

Neither of these methods lets Times otherwise behave as desired. I have lamented this problem myself with regard to Subtract and Divide as the "equivalent" forms they are converted into are neither equivalent nor as fast.

Community
  • 1
Mr.Wizard
  • 271,378
  • 34
  • 587
  • 1,371
  • Mr.Wizard, do you have new insight in the problem with the short forms you linked to? You said in one of the comments you might post a partial solution at some point. I'm sure many people would be interested (including me)! – sebhofer Aug 03 '14 at 09:42
  • @sebhofer I wish I did. I tried a few things and I didn't come up with anything I wanted to use. However it's been a long time and I didn't try too hard as I was hoping someone else had a better idea. – Mr.Wizard Aug 03 '14 at 14:44
4

Mathematica embraces the notion of canonical form. Internally it wants all expressions to be reduced to their canonical forms.

Times[x, x] is not a canonical form, so it gets reduced to Power[x, 2], which is.

I share Mr.Wizard's occasional frustration over the choices of canon made by Wolfram Research, but I think it is far too late to expect any change. So grin and bear it, as we say here in the US.

m_goldberg
  • 107,779
  • 16
  • 103
  • 257