triqs_Nevanlinna.solver.Solver.optimize

Solver.optimize(grid, eta, target, nk=15, maxiter=1000, gtol=0.001, verbose=False)[source]

Optimize the spectral function over a Hardy-function basis.

Adds a Hardy-function basis to the Nevanlinna interpolant and minimizes target over its coefficients, smoothing the real-frequency Green’s function while preserving the input Matsubara data. Requires a prior call to solve().

Parameters:

gridtriqs.gf.MeshReFreq: Real-frequency mesh on which to evaluate.
etafloat: Lorentzian broadening added to the real frequencies.
targetcallable: Objective to minimize, called as target(G_w) with the evaluated real-frequency Green’s function and returning a scalar (e.g. a measure of negative spectral weight).
nkint, optional: Number of Hardy basis functions per orbital. Default 15.
maxiterint, optional: Maximum number of conjugate-gradient iterations. Default 1000.
gtolfloat, optional: Gradient-norm tolerance for convergence. Default 1e-3.
verbosebool, optional: If True, print the optimizer status on completion. Default False.

Returns:

triqs.gf.Gf: Real-frequency Green’s function evaluated with the optimized Hardy coefficients.