>> exponential_test 26-Feb-2012 15:48:33 EXPONENTIAL_TEST: Demonstrate Mercer's theorem and the KL expansion for the exponential kernel. Using interval [0,10] Requested 20 eigenmodes, computed 20 I Lambda(I) 1 1.87288 2 1.21306 3 0.914777 4 0.688918 5 0.525614 6 0.408551 7 0.323876 8 0.261617 9 0.214975 10 0.179373 11 0.151711 12 0.129865 13 0.112355 14 0.0981303 15 0.0864339 16 0.0767106 17 0.0685474 18 0.0616322 19 0.0557263 20 0.0506447 Frobenius norm of I - Psi' * Psi = 5.3888e-12 Truncated estimate of K(s,s) = 1 for S in the interval. S K(s,s) estimate 0 0.743478 0.5 0.769579 1 0.706087 1.5 0.655966 2 0.636134 2.5 0.653042 3 0.702602 3.5 0.771539 4 0.841113 4.5 0.892305 5 0.911086 5.5 0.892305 6 0.841113 6.5 0.771539 7 0.702602 7.5 0.653042 8 0.636134 8.5 0.655966 9 0.706087 9.5 0.769579 10 0.743478 Index Cumulative Eigenvalue sum 1 0.187288 2 0.308594 3 0.400072 4 0.468964 5 0.521525 6 0.56238 7 0.594768 8 0.62093 9 0.642427 10 0.660364 11 0.675535 12 0.688522 13 0.699757 14 0.709571 15 0.718214 16 0.725885 17 0.73274 18 0.738903 19 0.744476 20 0.74954 Use a fixed number of eigenfunctions = 10 but vary the correlation length RHOBAR. (We used RHOBAR = 1 above.) The sum of the eigenvalues, divided by (B-A), discloses the relative amount of the total variation that is captured by the truncated expansion. RHOBAR VARSUM 4 0.947442 2 0.89598 1 0.660364 0.5 0.564253 0.25 0.42638 0.125 0.250992 0.0625 0.138373 0.03125 0.0793884 0.015625 0.0485088 0.0078125 0.0317328 EXPONENTIAL_TEST: Normal end of execution. 26-Feb-2012 15:48:51 >>