Truncated RAP-MUSIC (TRAP-MUSIC) source-localisation algorithm — sub-space truncation per iteration for robust correlated-source resolution. More...


Go to the source code of this file.
Classes | |
| struct | INVLIB::TrapMusicDipole |
| Result of a TRAP-MUSIC source scan. More... | |
| class | INVLIB::InvTrapMusic |
| TRAP-MUSIC (Truncated RAP-MUSIC) source localization. More... | |
Namespaces | |
| namespace | INVLIB |
| Inverse source estimation (MNE, dSPM, sLORETA, dipole fitting). | |
Truncated RAP-MUSIC (TRAP-MUSIC) source-localisation algorithm — sub-space truncation per iteration for robust correlated-source resolution.
SPDX-License-Identifier: BSD-3-Clause Copyright (c) 2026 MNE-CPP Authors
INVLIB::InvTrapMusic implements the TRAP-MUSIC algorithm of Mäkelä et al., NeuroImage 197, 616-626 (2019). It extends classic RAP-MUSIC by truncating the signal subspace by one dimension after each found source, which removes the rank-collapse failure mode that RAP-MUSIC exhibits when sources are strongly correlated. The class scans the leadfield for the grid point with the largest subspace correlation, projects it out, truncates the subspace and repeats until the requested source count is reached or the correlation drops below the user threshold. Returns a list of TrapMusicDipole records carrying grid index, correlation, 3-D position and orientation per found dipole.
Definition in file inv_trap_music.h.