Multi-slide SkrCCA optimization - First Component

Handles both standard (multiple cell types) and within (single cell type) cases Uses flat kernel structure for consistent data access

Usage

optimize_bilinear_multi_slides(
  X_list_all,
  flat_kernels,
  sigma,
  slides,
  max_iter = 1000,
  tol = 1e-05,
  n_cores = 1,
  direct_solve = TRUE,
  step_size = 1,
  sdev2_list = NULL
)

Arguments

X_list_all

List of lists of data matrices

flat_kernels

Flat list of kernel matrices

sigma

Sigma value (numeric)

slides

Slide IDs

max_iter

Maximum number of iterations

tol

Convergence tolerance

n_cores

Number of cores for parallel computation

direct_solve

For single cell type, use direct eigenvalue solution

step_size

Step size for damped power iteration (default 1). Values in (0,1) blend old and new weights for smoother convergence.

sdev2_list

Optional named list of squared standard deviations per cell type for weighted normalization. Default NULL.

Value

Named list of weight vectors (first component)