TY - GEN
T1 - Investigation of the Spectral Properties of Generalized Equivalence Integral Equations
AU - Sivakumar, Suryakumar
AU - Brick, Yaniv
AU - Adrian, Simon B.
N1 - Publisher Copyright:
© 2025 IEEE.
PY - 2025/1/1
Y1 - 2025/1/1
N2 - Method of moments integral equation solvers for electrically large problems require the use of fast methods to avoid an O(N2) computational complexity (N being the number of problem unknowns, which is assumed here to scale with the frequency, for a fixed mesh density). To this end, it is convenient and common to use algebraic compression that exploits the ε-rank deficiency (ε being the best low-rank approximation error bound) of the off-diagonal moment matrix block, such as the adaptive cross approximation (ACA). If the blocks, which are organized to represent interactions between well-separated source and testing sub-domains of the scatterer Ω, exhibit ranks that scale slow with the increase in number of domain basis functions, the complexity can reach O(N log N). However, for electrically large problems, this is guaranteed only for reduced dimensionality (i.e., elongated, quasi-planar) geometries, whereas for broadside interactions the ε-rank scales, asymptotically, linearly with O(N). Fortunately, for an essentially-convex Ω, this limitation can be overcome by formulating the problem as a generalized integral equation, with a modified kernel that includes an auxiliary component that significantly reduces the broadside interactions and highlight reduced-dimensionality interactions, thus enabling the design of fast solvers (A. Sharshevsky, Y. Brick, and A. Boag, “Direct solution of scattering problems using generalized source integral equations,” IEEE Transactions on Antennas and Propagation, vol. 68, no. 7, pp. 5512-5523, Jul. 2020. DOI: 10.1109/TAP.2020. 2975549). The auxiliary components are produced by sources placed inside Ω, which can be designed in various manners. Their design may have influence on the spectral properties of the generalized integral equation and the conditioning of the corresponding moment matrix that, to date, have not been studied.
AB - Method of moments integral equation solvers for electrically large problems require the use of fast methods to avoid an O(N2) computational complexity (N being the number of problem unknowns, which is assumed here to scale with the frequency, for a fixed mesh density). To this end, it is convenient and common to use algebraic compression that exploits the ε-rank deficiency (ε being the best low-rank approximation error bound) of the off-diagonal moment matrix block, such as the adaptive cross approximation (ACA). If the blocks, which are organized to represent interactions between well-separated source and testing sub-domains of the scatterer Ω, exhibit ranks that scale slow with the increase in number of domain basis functions, the complexity can reach O(N log N). However, for electrically large problems, this is guaranteed only for reduced dimensionality (i.e., elongated, quasi-planar) geometries, whereas for broadside interactions the ε-rank scales, asymptotically, linearly with O(N). Fortunately, for an essentially-convex Ω, this limitation can be overcome by formulating the problem as a generalized integral equation, with a modified kernel that includes an auxiliary component that significantly reduces the broadside interactions and highlight reduced-dimensionality interactions, thus enabling the design of fast solvers (A. Sharshevsky, Y. Brick, and A. Boag, “Direct solution of scattering problems using generalized source integral equations,” IEEE Transactions on Antennas and Propagation, vol. 68, no. 7, pp. 5512-5523, Jul. 2020. DOI: 10.1109/TAP.2020. 2975549). The auxiliary components are produced by sources placed inside Ω, which can be designed in various manners. Their design may have influence on the spectral properties of the generalized integral equation and the conditioning of the corresponding moment matrix that, to date, have not been studied.
UR - https://www.scopus.com/pages/publications/105036640538
U2 - 10.23919/CNC-USNC-URSI64444.2025.11419944
DO - 10.23919/CNC-USNC-URSI64444.2025.11419944
M3 - Conference contribution
AN - SCOPUS:105036640538
T3 - 2025 IEEE CNC-USNC-URSI North American Radio Science Meeting (Joint with AP-S Symposium) - Proceedings
SP - 3194
BT - 2025 IEEE CNC-USNC-URSI North American Radio Science Meeting (Joint with AP-S Symposium) - Proceedings
PB - Institute of Electrical and Electronics Engineers
T2 - 2025 IEEE CNC-USNC-URSI North American Radio Science Meeting (Joint with AP-S Symposium)
Y2 - 13 July 2025 through 18 July 2025
ER -