Re: MathML as a symbolic math language


Hatto von Aquitanien wrote:
> Does MathML provide a useful enough abstract language specification to use
> it as the basis of a symbolic algebra processor? What does it lack? What
> flaws does it have? I'm specifically asking about content MathML, as
> opposed to presentational MathML.
>
> http://www.w3.org/TR/MathML2/chapter4.html

My impression is that the flexibility of MathML for expressing formulas
is at odds with the needs of CAS software for strong typing of
entities. The <apply> construct allows one to indicate an operation
between entities with no concern for type consistency.

You could restrict the domain of computation in some manner to mitigate
these design conflicts, but it's probably a better idea to let a CAS
package use (internally) data structures less portable (hopefully more
optimized for computational purposes) than MathML node fragments or
similar.

Perhaps you are not asking about internal structures at all. I think
MathML is being adopted for input/output purposes in some packages.

regards, chip

.