phlower.tl.L1Norm_decomp
- phlower.tl.L1Norm_decomp(adata: AnnData, graph_name: str | None = None, eigen_num: int = 100, L1_mode='sym', mysym='a', check_symmetric: bool = True, isnorm=True, iscopy: bool = False)
graph hodge laplacian decomposition if sym:
call eigsh to for symmetric matrix eigen decomposition
L_1^s = D_2^{-1/2} L_1_norm D_2^{1/2} eigen vector of which is D_2^{-1/2} u_r
- else:
call eigs to for non-symmetric matrix eigen decomposition, and only keep the real part of eigen values and vectors L1_norm = D_2 B_1^ op D_1^{-1} B_1 + B_2 D_3 B_2^ op D_2^{-1}
Parameters
- adata: AnnData
AnnData object store the graph in uns slot
- graph_name: str
graph name for the graph with many holes
- eigen_num: int
number of eigenvalues to be calculated
- L1_mode: str
“sym” or “RW”
- check_symmetric: bool
check if the matrix is symmetric
- isnorm: bool
normalize the graph hodge laplacian
- iscopy: bool
copy the adata or not