SkrCCA optimization function for multiple groups (Single Slide) - First Component Uses flat kernel structure for consistent data access

SkrCCA optimization function for multiple groups (Single Slide) - First Component Uses flat kernel structure for consistent data access

Usage

optimize_bilinear(
  X_list,
  flat_kernels,
  sigma,
  max_iter = 1000,
  tol = 1e-05,
  step_size = 1,
  sdev2_list = NULL
)

Arguments

X_list

Named list of data matrices (cell by PC matrix)

flat_kernels

Flat list of kernel matrices with names like "kernel|sigma0.1|TypeA|TypeB"

sigma

Sigma value (numeric)

max_iter

Maximum number of iterations

tol

tolerance of accuracy

step_size

Step size for damped power iteration. Default 1 (standard power iteration). Values in (0,1) blend old and new weights for smoother convergence, which can help with many cells or many CCs.

sdev2_list

Optional named list of squared standard deviations per cell type, used for weighted normalization when scalePCs = TRUE. Default NULL (unweighted).

Value

Named list w_list containing the first weight vector component.