### Table 5. Numerical results for the rst example with the iterative splitting method and BDF3.

### Table 7. Numerical results for the second example with the iterative splitting method and BDF3.

### Table 2 Communication cost of a single iteration of Newton apos;s method for the CARE. 5 Numerical Experiments

### Table 5: Numerical results for problem 1 (ny = 10; nx = 2) N method Iter Feva Flops Flops per Iter

### Table 3: Numerical results for problem 3{5 problem N method Iter Feva Flops( ops on fun. evaluation) Prob. 3 10 TR-NY 17 27 177898(17700)

### Table 4: Numerical results for problem 6 n (N ? 1) method Iter (Feva) Flops( ops on fun. evaluation) 10 TR-NY 9 56293(3373)

### Table 2: Numerical results for problem 2 (ny = 4; nx = 2) N method Iter Feva Flops( ops on fun. evaluation) 10 TR-NY 13 14 320289(40485)

### Table 3.1: Travel-time problem. Comparison of iteration numbers between time-marching and GS with sweeping methods. Lax-Friedrichs numerical Hamiltonian. The third order

### Table 2 Number of iteration for the iterated penalty method (l = 1, k = 6, m = 10, j = 3)

29

"... In PAGE 6: ... We numerically test the dependence of the iteration number of the iterated penalty method on the penalty parameter r. We can see from Table2 that for r between 1 and 10, the number of iteration seems to be the least. But with bigger r, we have to increase the smoothing parameter m or use W-cycle multigrid methods to get a more accurate multigrid solution for each penalty problem.... ..."

### Table 1 Iterations (k) required for convergence of block methods.

"... In PAGE 15: ... Convergence is asserted when the norm of the residual is less than 10?7. Table1 contains numerical results for the block Jacobi and Gauss Seidel methods applied to the backward Euler equations. The number of iterations required to con- verge to the desired solution increases linearly with the number of timesteps in parallel as predicted by the theory.... ..."