Are you saying you do not understand what a shadow price on a constraint is? Or that you do not understand a specific optimization algorithm (such as the interior point method used by MIPS) and how these multipliers are computed?
If it is the first, I will just say, it is the sensitivity of the objective function to the constraint. In other words, in the case of MU_PMIN, for example, a shadow price of $X/MW means that the objective function would decrease by $X*Y if you were to relax the PMIN limit by Y MW for some tiny value of Y. If it is the second, I suggest that you take a course or read a book on non-linear optimization. The algorithmic details of how these shadow prices are actually computed is different for each algorithm and beyond the scope of what I can explain in an e-mail. -- Ray Zimmerman Senior Research Associate 419A Warren Hall, Cornell University, Ithaca, NY 14853 phone: (607) 255-9645 On Mar 28, 2012, at 4:13 PM, Carol Francesca wrote: > Dear Dr. Zimmerman, > > I read the manual but I didn't understand how Pmin mu for dispatchable loads > as well as Pmax for generators are calculated. I really confused. If it is > possible please explain, because It is not clear for me how it has been > computed. > > Best Regards > > Carol Francesca > > > > On Mon, Mar 26, 2012 at 15:16, Ray Zimmerman <[email protected]> wrote: > MU_PMIN and MU_QMIN are shadow prices on the minimum generation limits for > real and reactive power. Constraint shadow prices, also called Kuhn-Tucker > multipliers, are a standard output of most all constrained optimization > solvers. See any book on constrained optimization for a description of their > meaning and how they are computed for any given algorithm. > > In MATPOWER, the shadow prices are computed by whichever solver is used. For > example, in the case of the default MIPS solver, they are included in the mu > variable in equation (A.32) in Appendix A. The value is computed in mips.m. > > I should probably mention also that in the case of trapezoidal generator > capability curves (see Section 5.4.3 in the manual), if one of the sloped > portions of the curve is binding, its shadow price is decomposed into > components for the corresponding real and reactive limits. > > -- > Ray Zimmerman > Senior Research Associate > 419A Warren Hall, Cornell University, Ithaca, NY 14853 > phone: (607) 255-9645 > > > > > On Mar 25, 2012, at 7:08 PM, Carol Francesca wrote: > >> Dear Prof. Zimmerman, >> >> I have a question: >> >> When I do runmarkt sometimes price and Lambdas are not equal and its because >> of network congestion and losses. >> In MATPOWER, how MU_PMIN and MU_QMIN are calculated? Could you please >> address the m.file which calculates these? >> How they affect the prices? There is no sufficient explanation about it in >> the manual. I am sorry but I cannot understand how they are calculated. >> >> Best Regards >> >> Carol > >
