Some analysis of Tikhonov regularization for the inverse problem of option pricing in the price-dependent case (Q2492071): Difference between revisions

This is one of a series of papers on inverse option pricing written by members of the Chemnitz research group on inverse problems and published in the last years. In this paper, T.~Hein analyzes in detail in a Hilbert space setting the problem of identifying purely price-dependent volatility functions \(a\) from a function \(u\) of prices of European vanilla call options with varying strikes \(K \in (0,\infty)\) and fixed maturity. He discusses the forward operator \(F: a \mapsto u\) mapping between \(H^1(R)\) and \(L^2(R)\) when using logarithmized variables and makes assertions on its properties including weak closedness. Moreover, he investigates the Fréchet derivative of this operator and shows that the corresponding nonlinear operator equation \(F(a)=u\) is locally ill-posed. For the stable approximate solution of this equation Tikhonov regularization can be applied. As is done in \textit{T. Hein} and \textit{B. Hofmann} [Inverse Probl. 19, 1319--1338 (2003; Zbl 1086.91028)] for the case of identifying purely time-dependent volatilities also here a stringent convergence rate analysis gets possible. Note that if time- and price-dependent volatilities can be determined separately, then volatility surfaces (which are of main interest, for example for pricing exotic options) can be obtained approximately by using decoupling techniques [see \textit{I. Bouchouev} and \textit{V. Isakov}, Inverse Probl. 15, R95--R116 (1999; Zbl 0938.35190) and \textit{H. Egger, T. Hein} and \textit{B. Hofmann}, ibid. 22, 1247--1259 (2006; Zbl 1112.91032)].