rampedpyrox.calc_L_curve

rampedpyrox.calc_L_curve(model, timedata, ax=None, plot=False, nLam=150, lam_max=100.0, lam_min=0.001)[source]

Function to calculate the L-curve for a given model and timedata instance in order to choose the best-fit smoothing parameter, lambda.

Parameters:
  • model (rp.Model) – rp.Model instance containing the A matrix to use for L curve calculation.
  • timedata (rp.TimeData) – rp.TimeData instance containing the time and fraction remaining arrays to use in L curve calculation.
Keyword Arguments:
 
  • ax (None or matplotlib.axis) – Axis to plot on. If None and plot = True, automatically creates a matplotlip.axis instance to return. Defaults to None.
  • plot (Boolean) – Tells the method to plot the resulting L curve or not.
  • lam_min (int) – Minimum lambda value to search. Defaults to 1e-3.
  • lam_max (int) – Maximum lambda value to search. Defaults to 1e2.
  • nLam (int) – Number of lambda values to consider. Defaults to 150.
Returns:

  • lam_best (float) – The calculated best-fit lambda value.
  • axis (None or matplotlib.axis) – If plot = True, returns an updated axis handle with plot.
Raises:
  • ScalarError – If lam_max or lam_min are not int or float.
  • ScalarError – If nLam is not int.

See also

Daem.calc_L_curve()
Instance method for calc_L_curve.

References

[1] D.C. Forney and D.H. Rothman (2012) Inverse method for calculating
respiration rates from decay time series. Biogeosciences, 9, 3601-3612.
[2] P.C. Hansen (1994) Regularization tools: A Matlab package for analysis
and solution of discrete ill-posed problems. Numerical Algorithms, 6, 1-35.