Re: Hildreth-d'Esopo
- From: spellucci@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx (Peter Spellucci)
- Date: Sun, 8 Jun 2008 15:05:08 +0200 (CEST)
In article <a6d4ba7e-fa4e-497e-8a63-ac0a74b39490@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
chi <kravtsovdenis@xxxxxxxxxxxxxx> writes:
Thank you Peter,
This is helpful. So I guess box constraints (x_l < x < x_u) could be
applied here as well? If you are saying it is from stone age, could
you recommend any other more appropriate method for this type of
problem (I'm currently experimenting with box constrained Levenberg).
Thank you.
what is "good" depends on your specific problem (Hildreth-Esopo
has the advantage of being really simple, robust and requiring nothing
than a code for evaluation of the residual b-A*x, no explicit use of the
matrix storage required)
look here:
http://plato.asu.edu/guide.html
for large scale cases the reflective Newton's method of Coleman and Li might
be a choice.
hth
peter
.
- References:
- Hildreth-d'Esopo
- From: chider
- Re: Hildreth-d'Esopo
- From: Peter Spellucci
- Re: Hildreth-d'Esopo
- From: chi
- Hildreth-d'Esopo
- Prev by Date: Re: all solutions manual 2008
- Next by Date: Re: An integral
- Previous by thread: Re: Hildreth-d'Esopo
- Next by thread: Re: String searching and lienar optimization
- Index(es):
Relevant Pages
|
