Hi.
I have been trying to work with the Rdonlp2 package. There is a control
parameter called epsx that has the following description:
epsx(1e-5) - successful temination is indicated if the Kuhn-Tucker criteria
are satisfied within the value
I have tried to use the donlp2.control() function to adjust the value of
epsx, e.g.:
>cntlx1_3<-donlp2.control(epsx=1e-03)
>cntlx1_3$epsx
[1] 0.001>
I have adjusted it larger (as above) and smaller, but it has not had any
effect on the solution when I run donlp2(...,control=cntlx1_3). And the
console output from running the donlp2() function reports a value of epsx,
which is always "epsx= 1.000e-05".
Am I doing something wrong? Is there a way to adjust the value of epsx?
I've been comparing donlp2() with quadprog() on a quadratic programming
problem; the objective function values are pretty close, but Id like to see
if I can get them any closer by adjusting epsx. Any help is appreciated.
Thanks!
-- TMK --
212-460-5430 home
917-656-5351 cell