Abstract: Basing on regularization method,we propose an efficient method for calibrating time-varying parameters in non-Gaussian one-factor short-term rate model with constraineds.Stability and convergence are proved with variational principle。For computational convenience,penalty method and smoothing operator are applied to transform a constrained inverse problem into one without constraints.Finally,theoretical analysis is shown valid by means of numerical simulations.Empirical results is earried out.

Key words: non-Guassian shor-term rate model, regularization method, penalty method, smoothing operator