Recent content by farooq117

in fact, in my case, a quadratic programming package isn't really an option.. the problem that i described is one sub-problem in one iteration of a gradient-based algorithm for distributed optimization.. if there is a centralized optimizer, then perhaps it can use such a quadratic programming...

thanks..

there is another approach.. this gives a quadratic cost with coupled linear constraints.. i'd be very grateful if you could check the working.. the problem is to find the maximum of <a,x> - (L/2)||x-u||_inf^2 s.t.: x >= 0, ||x||_inf <= R. this is the same as finding the minimum of...

i have an approach in mind.. i thought i'd get your opinion on it.. the problem is to find the maximum of the following: <a,x> - (L/2)||x-u||_inf^2 s.t.: x >= 0, ||x||_inf <= R. The is the same as finding the minimum of (L/2)||x-u||_inf^2 - <a,x> s.t.: x >= 0, ||x||_inf <= R. I use...

well.. the actual problem is to find the argument which maximizes the following expression: <a,x> - (L/2)||x-u||_inf^2, s.t.: x >= 0, ||x||_inf <= R, where a and u are known vectors, x is the vector of decision variables, and R and L are known scalars. Furthermore, the vector u satisfies...

I have to minimize an expression of the following type: min <a,x>-L||x-u||_inf^2 s.t.: ||x||_inf <= R, where a is a vector of coefficients, x is the vector of decision variables, <.,.> denotes the scalar product, R and L are scalars, u is some constant (known) vector, and 'inf' denotes...