Convergence rates of moving mesh methods for moving boundary partial integro-differential equations from regime-switching jump-diffusion Asian option pricing (Q2297071): Difference between revisions

The paper studies the following system of two-dimensional partial integro-differential equations (PIDEs) satisfied by the value function of the Asian option under the state-dependent regime-switching jump-diffusion models \[ \begin{split} \displaystyle \dfrac{\partial V(S,I,t,i)}{\partial t}+\dfrac{1}{2}\sigma_i^2S^2\dfrac{\partial^2 V(S,I,t,i)}{\partial S^2}+(r_i-\delta_i-\lambda_i\kappa_i)S\dfrac{\partial V(S,I,t,i)}{\partial S}\\ +S\dfrac{\partial V(S,I,t,i)}{\partial I}+\lambda_i \int_{-\infty}^{+\infty}V(e^{y}S,I,t.i)f_i(y)dy+\sum_{i=1}^{d}q_iV(S,I,t,I)\\ -(r_i+\lambda_i)V(S,I,t,i)=0, \,\, i \in \mathcal{D}, \end{split} \] subject to the following conditions: \[ V(S,I,T,i)=\max(I/T-K,0), \,\,\, V(S,-\infty,t,i)=0, \,\,\, i\in \mathcal{D}, \] where \(S\) and \(I\) are dummy variables, \(i \in \mathcal{D} \equiv \{1, 2, \dots , d\} \) are nonnegative constants. Each equation in the above system of equations is a two-dimensional problem and since there is no diffusion in the \(I\) direction, it causes many difficulties in the numerical solutions and analysis with the standard finite difference methods. The authors recast the above two-dimensional problem into a moving boundary problem of one-dimensional PIDEs and analyze the convergence rates of the moving FDMs. By converting the semi-infinite domain into an appropriate finite domain, a uniform mesh for time and a moving mesh for space are defined. Here, the moving mesh for space is driven by the time-dependent boundary, not by the use of the monitor function. Moving FDMs are constructed to solve the moving boundary problem. The proof of the local truncation error estimation has given. Also, the error analysis of the moving FDM has carried out. Several numerical examples are presented to verify the convergence rates of the moving FDM.