[GiNaC-list] extending the symbol

[nomad]

Hi Alexei,

Can you please attach a copy of your replies to my email address? I disable
email delivery whenever I subscribe to forums as most of the times I post
infrequently.


Alexei Sheplyakov-2 wrote:
> 
>> I am adding a light weight support for vector algebra in ginac. I am
>> aware that it can be done through Clifford algebra, but the output is
>> almost incomprehensible to debug.
> 
> I don't quite understand what is exactly wrong. Could you elaborate,
> please?
> 

If vector and scalar products of vectors are not given their own identities
but are implemented using the clifford algebra as in the post
http://www.ginac.de/pipermail/ginac-list/2005-June/000671.html then the
output no longer bears any resemblance to the original definition. For
example the consider the following output
   
"-C.0*B.2*A.1-C.1*A.2*B.0+C.2*B.0*A.1+C.0*A.2*B.1+C.1*B.2*A.0-C.2*B.1*A.0"
There is no resemblance to the original definition which could be printed
nicely
   "A . (B x C)"
To emphasize my point, I dont really care how the triple product is
represented inside. But when things go wrong (due to bugs) I would like to
easily debug the code. For example, if I was expecting A . (C x B), the
second output will immediately point out that to me.

On the other hand if scalar and vector products of vectors are given their
own identities, then there is no need for clifford algebra any more to
evaluate them.


Alexei Sheplyakov-2 wrote:
> 
>> My problem is this. When a request
>> like inner_product(lhs, rhs) is made, where one of lhs and rhs is not
>> a vector, I would like to replace the inner_product with a regular
>> product. How can I achieve this?
> 
> if (typeid(ex_to<basic>(lhs)) != typeid(your_class)) {
> 	// do the replacement
> } else if (typeid(ex_to<basic>(rhs)) != typeid(your_class)) {
> 	// same here
> } else {
> 	// do something nontrivial
> }
> 

Yes. This would be sufficient.


Alexei Sheplyakov-2 wrote:
> 
>> I am planning to do it as follows
>> 1) extend a symbol to a vector
> 
> I don't think it's a good idea. Symbol is something simple and in a sense
> "atomic". Also it's quite performance and memory critical. You'd better
> use the clifford class.
> 

I need to think about this.

thanks,
Krishna.

-- 
View this message in context: http://www.nabble.com/extending-the-symbol-tp19818944p19849081.html
Sent from the Ginac - General mailing list archive at Nabble.com.