Dear Ray, I agree that the constant term does not affect the value of optimal x. However, it would be convenient to have the final value of f(x,z) directly in this case. No worries.
Thank you for your reply, Panagis On 18 September 2012 17:01, Ray Zimmerman <[email protected]> wrote: > Because of the form of the user-defined costs, I'm not sure you have full > control over a constant term, but that doesn't impact the optimization > anyway. You can certainly implement the quadratic function f(x) = -x^2 + 4x > - 4. You will simply have to add an offset of 2 to that cost and to the > final objective function value outside of the optimization. > > You can do this with the following parameters: > rhat = 2 > k = 0 > d = 2 > m = 1 > C = -1 > > -- > Ray Zimmerman > Senior Research Associate > 419A Warren Hall, Cornell University, Ithaca, NY 14853 > phone: (607) 255-9645 > > > > > On Sep 17, 2012, at 3:13 PM, Panagis Vovos wrote: > > Dear Ray, > > I mean an quadratic cost function, such as f(x)= - x^2 + 4x - 2. This > function f(x) has both negative and positive values for positive > values of x (attached picture). > > Thanks for the help, > > Panagis > > > > > On 17 September 2012 16:20, Ray Zimmerman <[email protected]> wrote: > > In order to be sure I understand what you are attempting, I think I'd still > > need an example of the kind of function you are looking for. > > > -- > > Ray Zimmerman > > Senior Research Associate > > 419A Warren Hall, Cornell University, Ithaca, NY 14853 > > phone: (607) 255-9645 > > > > > > On Sep 14, 2012, at 3:40 AM, Panagis Vovos wrote: > > > Dear Ladies and Gentlemen, > > > Judging from a reply I received already, I definetely have not made it > > perfectly clear what I mean by negative-psitive values for my > > user-defined cost function. I can get a function having both positive > > and negative values, BUT ONLY IF my variable can take negative values. > > This is because the user defined cost function is in the form > > Y=aX^2+bX, which means it has to verify Y=0 for X=0 (cross junction of > > the two axes). Unfortunatelly, most OPF variables cannot take negative > > values. > > > > Thanks again, > > > Panagis Vovos > > > > > Dear Ladies and Gentlemen, > > > I am trying to define my own cost function, that will be added to the > > regulat MATPOWER OPF cost function, using 5.3.1. However, after > > simulating (5.27)-(5.31) in Matlab, I am unable to produce a (ideally, > > quadratic) cost function that will span in both negtive and positive > > values. The best I can get is a positive or negative (but never both) > > cost function with a dead band. Does anyone know a way that I can > > achieve this according to MATPOWER standard approach or do I have to > > write my own code for the definition of the cost function, derivatives > > etc? > > > Thank you for your time, > > > Panagis Vovos > > > > <fx.jpg> > >
