[Csdp] CSDP initialization

[Brian Borchers]

On Sun, Nov 6, 2016 at 8:04 PM, Dehghanpoor, Golnoosh <
g.dehghanpoor at wustl.edu> wrote:

> Hi,
>
>
> I was wondering if in the first iteration of the CSDP algorithm, the
> analytical center is calculated (the weight of objective function assigned
> to aero and just the logarithmic term in the barrier function), or the
> weight of the logarithmic term in the barrier function, is just set to be a
> very large number? (and if that's the case what's its range?). Thanks for
> your time.
>

No.  CSDP uses a heuristic to compute a reasonably central initial
solution. Alternatively, the user can supply an initial solution (but this
isn't recommended.)

Starting from the initial solution, at each iteration CSDP computes a
primal-dual affine step to estimate how far mu can be reduced and then uses
the estimated reduction in mu as the target in computing a primal-dual
barrier step.  This is a conventional predictor-corrector approach.

-- 
Brian Borchers                          borchers at nmt.edu
Department of Mathematics      http://www.nmt.edu/~borchers/
New Mexico Tech                       Phone: (575) 322-2592
Socorro, NM 87801                   FAX: (575) 835-5366
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/csdp/attachments/20161108/561c288f/attachment.html>