conatus.expert.dmd

conatus.expert.dmd(X, *, dt, rank, inner_product, amplitudes='first', method='exact', Y=None, times=None, tol=None, max_nfev=None)[source]

DMD of the sequence X = [x_0 … x_m] (pairs x_j → x_{j+1}), or of pairs (X, Y).

amplitudes (how the amplitudes b are fitted; the modes and eigenvalues do not depend on it): declared default "first", the standard least-squares fit of the first snapshot b = argmin ‖F(x_0 − Φ b)‖ (Schmid 2010), valid for sequences and pairs; "optimal" fits every snapshot of a sequence (Jovanović et al. 2014). The choice is printed by explain().

method="optimized" needs tol and max_nfev (declared), uses the whole sequence (Y not allowed) and optionally non-uniform times (default: j·Δt, declared by dt).

Parameters:

  • dt (float)

  • rank (int)

  • inner_product (DataInnerProduct)

  • amplitudes (str)

  • method (str)

  • tol (float | None)

  • max_nfev (int | None)

Return type:

DMD