[GiNaC-list] Term ordering and compiling C++ code
alexei.sheplyakov at gmail.com
Mon May 24 20:51:07 CEST 2010
On Mon, May 24, 2010 at 02:17:41AM -0400, jros wrote:
> In this example
> I suppose that no sharing is implied like e2=e1+c, that would be
> computationally expensive. But, if instead
> then e1 is referenced by e2?? (I suppose yes).
No. GiNaC tries to keep trees (which what GiNaC::add essentially is) as
flat as possible (see expairseq::construct_from_epvector). This save
*a lot* of memory and makes collecting similar terms more efficient.
So +( +(a, b), c) gets transformed into +(a, b, c), and nothing is
shared between e1 and e2.
More information about the GiNaC-list