a non-smooth optimization problem
- From: Hagen Thom <hth@xxxxxxxxxxxxxxxxxxxxxx>
- Date: Wed, 19 Oct 2005 21:34:42 +0200
Hi,
(in the following k and N and q are fixed integers)
Assume I have fixed reel numbers
A_1,...,A_N
and fixed SMOOTH functions
g_1(theta),...,g_N(theta)
with a finite dimensional parameter theta in a subset of R^q
I'd like to minimize the following expression
sum_{n=1}^N sum_{j=0}^k 1_{[a_j,a_{j+1}[}(A_n) g_n(theta_j)
where 1_M is the indicator function of a set M; the
sets M are intervals.
The expression is simply smooth if only
theta_1,...,theta_k
are free variables and the a_j fixed; classical calculus
works in principle.
But I'd like to minimize the expression in the free
variables
theta_1,...,theta_k
as well as the k free variables a_j, j=1,...,k, with
0=a_0<a_1<...<a_k<a_{k+1}=infinity
In the a_j, j=1,...,k, the function is not smooth.
I have no idea, after what I could search.
Have such problems a "name" or classification?
I would appreciate hints (like references)
for a starting point.
Best regards.
.
- Follow-Ups:
- new attempt
- From: Hagen Thom
- Re: a non-smooth optimization problem
- From: Peter Spellucci
- new attempt
- Prev by Date: Re: Optimization
- Next by Date: Re: Recommend please : free math library
- Previous by thread: Correction for: Interpolating function with two important characteristics
- Next by thread: Re: a non-smooth optimization problem
- Index(es):