Imaging (Gradient Descent)

reconstruct(x_start::AbstractArray{T,4}, data::AbstractMatrix{<:OIdata},
            ft::AbstractMatrix; ...)

Unified image reconstruction from optical interferometric data using VMLMB.

x_start is the initial image of shape (nx, nx, nwav, nepoch). data is a Matrix{OIdata} of size (nwav, nepoch) (from readoifits or readoifits_multiepochs). ft is the matching Matrix of FT plans from setup_ft.

The criterion minimised is (χ² + regularization) / ndof. The flux-normalisation chain-rule correction is applied to the combined (χ² + regularization) gradient per cell.

Keywords

• weights = [1.0, 1.0, 1.0]: relative weights for (V², T3amp, T3phi)
• regularizers = []: per-cell regularizers — a list of ["name", μ, ...] tuples applied identically to every (wavelength, epoch) cell
• transspectral_regularizers = []: cross-wavelength regularizers (per epoch), e.g. [["transspectral_tv", 0.1]]
• temporal_regularizers = []: cross-epoch regularizers, e.g. [["temporal_tvsq", 0.01]]
• epochs_weights = []: per-epoch scaling (default: uniform)
• use_diffphases = false: force differential-phase fitting
• verb = false: verbose output
• maxiter = 100: maximum VMLMB iterations
• vonmises = false: von Mises loss for closure phases
• ftol = (0, 1e-8), xtol = (0, 1e-8), gtol = (0, 1e-8): VMLMB tolerances

reconstruct(x_start::AbstractMatrix, data::OIdata, ft; ...)

Monochromatic convenience wrapper — delegates to the unified 4-D method. See the 4-D method for the full keyword list.