I've been reading "Antenna Theory" by Balanis, 2016 and the HOBBIES book by Zhang et al. This has made me curious about the choice of basis set. The constant value (aka constant pulse) seems rather a poor choice, at least for wire antennas.
The basis set used in HOBBIES seems more sensible, but there is no explanation of the reasoning behind the choice.
As neither says anything about the computation of the matrix inverse, I'm assuming that both NEC2/4 and HOBBIES use least squares.
Has anyone looked at applying the MoM using a wavelet basis? And more particularly using a sparse L1 solution instead of L2.