High Performance Computing 1
Method
•
–The interpolation operator In(i-1) takes an approximate solution x(i-1) for P(i-1) and converts it to an approximation x(i) for the problem P(i) on the next finer grid: ( b(i), x(i) ) = In(i-1)( b(i-1), x(i-1) ) Its implementation also requires just a weighted average with nearest neighbors.
–The solution operator S(i) take a problem P(i) and approximate solution x(i), and computes an improved x(i). x_improved(i) = S(i)( b(i), x(i) ) We often a variation of Jacobi's method as the solution operator (recall symmetry of Jacobi, in contrast to, say, GS)
•