# [Rd] Do not misuse R-devel {was "Quadratic Programming"}

Martin Maechler maechler at stat.math.ethz.ch
Fri Dec 14 18:13:03 CET 2007

Please use the R-help mailing for such *Questions*.

R-devel has a very different purpose.
Probably you should also read the posting guide,
--> http://www.r-project.org/posting-guide.html

Regards,
Martin Maechler, ETH Zurich

>>>>> "dGdVS" == de Gosson de Varennes Serge (4100) <serge.de.gosson.de.varennes at forsakringskassan.se>
>>>>>     on Fri, 14 Dec 2007 09:15:26 +0100 writes:

dGdVS> Hi all!  I have a little question concerning
dGdVS> quadprog. To make it simple I'll start by stating the
dGdVS> problem:

dGdVS> I want to minimize

dGdVS> 		h(d,delta)=0.5d^T B d +nabla(f(x))^T d
dGdVS> +rho*delta^2

dGdVS> With respect to d\in R^n and delta \in R. I obviously
dGdVS> have constraints (depending on both d and delta).

dGdVS> Solve.QP does give me a good result for d but I
dGdVS> cannot obtain anything for delta. Simce dim(Dmat)=n
dGdVS> and sol<-rep(0,n) it isn't particularly surprising.
dGdVS> To set a diagonal matrix

dGdVS> 		(B 0 ) Amat= (0 rho )

dGdVS> Is a crapy idea. Does anyone have an idea?

dGdVS> Yours,

dGdVS> Serge

dGdVS> "Beatus qui prodest quibus potest" - (Lycklig är
dGdVS> den som hjälper andra)

dGdVS> Serge de Gosson de Varennes Försäkringskassan
dGdVS> Swedish Social Insurance Agency +46-76 11 40 799
dGdVS> serge.de.gosson.de.varennes at forsakringskassan.se

dGdVS> 	[[alternative HTML version deleted]]

dGdVS> ______________________________________________
dGdVS> R-devel at r-project.org mailing list
dGdVS> https://stat.ethz.ch/mailman/listinfo/r-devel