On Mar 28, 2012, at 12:10 PM, Mathias Vanwolleghem wrote:
I would like to try out the possibilities of this library in a
Python code. However I noticed that the Python API does not allow to
pass extra parameters to the objective function, contrary to the C
API.
Did I read it well that the objective function can only take two
arguments: the optimization variable numpy array, and the gradient ?
Is there a way to nevertheless pass extra function parameters.
You can pass as many extra parameters as you want to the Python
objective, using a lambda. (Because Python has proper closures and
lexical scoping, it doesn't need hacks like the extra void* parameter
in C or "functor" objects in C++.)
There is an example of this in the NLopt tutorial -- look at the
"myconstraint" function, where we pass two extra parameters "a" and
"b" via a lambda.
e.g. if your objective is a function f(x,grad,p) that takes an extra
parameter p, then you would do:
opt.set_min_objective(lambda x,grad: f(x,grad,somep))
to pass "somep" for the "p" parameter.
--SGJ
_______________________________________________
NLopt-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
