After finding coherence remained at some resonsnt frequences with the simple decoupling method, I tried to decouple the YAW from LEN and PIT.
The result is here. (The other one attached is coherence between YAW and LEN/PIT.)
We do not observe the peaks at 0.4-0.5 Hz (Whole chain mode), 0.57 Hz (IM L), 0.65 Hz (TM L) in YAW
I think it is good that we can also remove the 0.4-0.5 Hz peaks from YAW signal without using the data at those peaks. (I used that values only at 0.15 Hz(YAW), 0.57 Hz, and 0.65 Hz.)
My code for this calculation is in /kagra/Dropbox/Personal/Shoda/PRM/L2A/OL_diag.ipynb, it is a spagetti code though.
I will implement the new matrix when the suspension available.