Gunnar Andersson. Some New Randomized Approximation Algorithms. Doctoral dissertation, Department of Numerical Analysis and Computer Science, Royal Institute of Technology, May 2000.  Home/Search   Document Details and Download   Summary   Related Articles

....instance, i.e. an instance where the optimum solution satisfies a fraction 1 # of constraints. Our approximation algorithm for Max Co BIJ(d) is based on a semidefinite programming relaxation, similar to that used for Max BIJ(d) by Khot [12] combined with a rounding scheme used by Andersson [1, 2] to construct an approximation algorithm for Max d Section, the generalization of Max Bisection to domains of size d. Technically, we view this algorithmic result as the main contribution of this paper. Inapproximability results: For the Boolean case, d = 2, it is known that Max E3 NAE Sat can be ....

....by the positive constant K. 3 Approximation algorithm for Max Co BIJ(d) To construct an approximation algorithm for Max Co BIJ(d) we combine a modification of the semidefinite relaxation used by Khot [12] for the Max BIJ(d) problem with a modification of the randomized rounding used by Andersson [1, 2] for the Max d Section problem. Recall that a specific clause in the Max Co BIJ(d) problem is of the form (x, x # , #) where x and x # are variables in the Max Co BIJ(d) instance and # is a permutation, and that the clause is satisfied unless x = j and x # = #(j) for some j. In our semidefinite ....

Gunnar Andersson. Some New Randomized Approximation Algorithms. Doctoral dissertation, Department of Numerical Analysis and Computer Science, Royal Institute of Technology, May 2000.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC