ufl team mailing list archive

[Martin Sandve Alnæs <question212068@xxxxxxxxxxxxxxxxxxxxx>]

Question #212068 on UFL changed:
https://answers.launchpad.net/ufl/+question/212068

Martin Sandve Alnæs proposed the following answer:
Ignoring the non-scalar case, and limiting this to pure "chains" of
Product/Division, is probably fairly easy to implement as an algorithm
you can choose to apply, or that a form compiler can choose to apply.
This should also be safe from a floating point point of view. I
deliberately do not allow such deep simplifications automatic at
construction of expressions however, because they will be O(n) in the
size of the expression tree, and they may also obfuscate otherwise clear
cases of subexpression reuse.

-- 
You received this question notification because you are a member of UFL
Team, which is an answer contact for UFL.