Application of a Sparsity Pattern and Region Clustering for Near Field Sparse Approximate Inverse Preconditioners in Method of Moments Simulations

Table 1

Convergence performance for several SAI parameters.

	Precond. generation time (s)	Precond. size (MB)	Avg. number of iterations per direction ( pol.)	Avg. number of iterations per direction ( pol.)
No preconditioner	—	—	394.15	332.42
Conventional SAI	189	378.43	192.00	172.05
Proposed SAI	43	284.57	181.94	158.73
Sparsity distance () = 0.25 λ; Preprocessing threshold () = 0.03; Postprocessing threshold () = 0.03
Proposed SAI	192	281.78	129.81	115.98
Sparsity distance () = 0.5 λ; Preprocessing threshold () = 0.03; Postprocessing threshold () = 0.03
Proposed SAI	1208	304.33	111.78	103.21
Sparsity distance () = λ; Preprocessing threshold () = 0.03; Postprocessing threshold () = 0.03