Imaging (SPARCO)

SPARCO (SPectral Analysis of Resolved Compact Objects) decomposes the scene into a parametric model (stars, background) + a reconstructed image (environment).

Function	Description
`reconstruct_sparco(x0, data, ft; lambda_ref, ...)`	High-level SPARCO: set physical params, get image back
`chi2_sparco_flat_f(x, params, model, ft, data)`	Flat-model SPARCO chi-squared (forward only)
`optimize_sparco_flat_parameters(params, model, x, ft, data)`	Optimize flat-model SPARCO parameters (image fixed)
`model_and_image_to_vis(model, params, x_img, ft, data)`	Combined visibility: `V_model + W(λ) · V_image`

OITOOLS.reconstruct_sparco — Function

reconstruct_sparco(x_start, data, ft;
                   lambda_ref, star_flux=0.5, bg_flux=0.0,
                   d_env=0.0, star_di=4.0,
                   star_spectral_index_fixed=true,
                   fit_model_first=true,
                   regularizers=[], weights=[1,1,1],
                   maxiter=200, rounds=3, verb=false, ...)

High-level SPARCO image reconstruction with the standard star + background + image model.

Builds the flat-model dict internally from the physical parameters, then calls reconstruct_hybrid.

Arguments

x_start: starting image (nx × nx)
data: OIdata or Vector{OIdata}
ft: Fourier transform plans from setup_nfft
lambda_ref: reference wavelength in metres (required)
star_flux: initial stellar flux fraction at λ₀ (default 0.5)
bg_flux: initial background flux fraction at λ₀ (default 0.0)
d_env: environment spectral index (default 0.0)
star_di: stellar spectral index (default 4.0 = Rayleigh-Jeans)
star_spectral_index_fixed: fix star_di during reconstruction (default true)
fit_model_first: run a model-only fit before image reconstruction (default true)

Returns a named tuple (params, image, model, free_params, chi2).

source

OITOOLS.chi2_sparco_flat_f — Function

chi2_sparco_flat_f(x_img, params, model, ftplan, data;
                   w_name="env,f", weights=[1,1,1], verb=true, vonmises=false)

Evaluate chi2 for the SPARCO-flat model (no gradient).

source

OITOOLS.optimize_sparco_flat_parameters — Function

optimize_sparco_flat_parameters(params_start, model, x_img, ft, data;
                                w_name="W", weights=[1,1,1], lb, ub,
                                method=:LN_NELDERMEAD, maxeval=2000)

Optimize SPARCO-flat model parameters with the image held fixed. Supports gradient-free (:LN_NELDERMEAD) and gradient-based (:LD_LBFGS, etc.) methods. For :LD_* methods, the parameter gradient is extracted from chi2_sparco_flat_fg. Returns (minchi2, params_opt, ret).

source

OITOOLS.model_and_image_to_vis — Function

model_and_image_to_vis(model, params, x_img, ftplan, data; w_name="W")

Compute the combined visibility Vtotal = Vmodel + W(λ) · V_image.

Returns a NamedTuple (; V_total, V_model, V_image, W).

source