[Om] Adding DLMF links to CDs [Re: How to translate csymbol/@definitionURL]

[Lars Hellström]

Paul Libbrecht skrev:
> I encountered another usage recently... probably not *exactly*  
> definition oriented: search canonicalization rules.
> Examples of such are provided in Abou Youssef and Moody Altamimi's  
> paper at extrememarkup for content-MathML but almost each rule could  
> be considered an FMP.
> 
> What's interesting there is that really not all FMPs qualify as  
> canonicalization rules (e.g. the associativity can, but probably only  
> the one which "flattens"), that it goes in one sense and not the other

That rewrite rules are directed is pretty much unavoidable in Rewriting 
Theory.

> (applying a definition such as the binomial coefficient's factorial  
> expression certainly shouldn't go backwards for a normal search  
> comparison, coming from 1 to sin^2 x + cos^2 x would be horribly  
> surprising).
> 
> I am not sure what is the criterion, except length shortening,

The choice of criterion for orienting one's given identities into 
rewrite rules can be a Very Hard Problem (not unlike the problem of 
"proving theorem X"). Length shortening can be a good principle to 
start with, but it is not unusual that one has to make slight 
deviations from it.

Since you seem to be unfamiliar with rewriting, I should perhaps also 
point out that you often need to add to your system derived rules that 
no sane person would pick as axioms, just to ensure that the normal 
(canonical) forms with respect to it are unique.

Lars Hellström

> but  
> FMPs clearly are concerned, it's... just another set of FMPs.
> More important, I believe, is that we give IDs to FMPs!