Impedance matrix localization produces a sparse moment method matrix

Basis and testing functions for moment-method calculations that make the resulting matrix sparse are discussed. Recent numerical results have verified theoretical predictions that a moment-method matrix with N/sup 2/ elements can be replaced by one with 100N nonzero elements. This technique is unlike some other techniques (e.g. k-space, differential equation, etc.) in that this is done without increasing N from its usual moment-method value. This technique also differs from many of these techniques (e.g. k-space. Rokhlin´s factorization, etc.) because the resulting matrix is truly sparse, permitting solutions using efficient preconditioners for iterative techniques and possibly sparse direct techniques. For large moment-method problems (N>10000), the resulting savings in memory and execution time are both greater than a factor of 100. This impedance matrix localization technique applies to any moment-method formulation.<>