Necessary and sufficient conditions for the polynomial convergence of the quasi-reversibility and finite-difference methods for an ill-posed Cauchy problem with exact data (Q266092)

The author analyses the convergence of finite difference methods with two and three layers for approximate solutions of the Cauchy problem in abstract form \[ \dot{x}(t) = A x(t), \qquad x(0)= f \qquad \text{on} \quad [0,T], \] where \( A: D(A)\subset X\rightarrow X \) is an unbounded closed operator in a Banach space \( X\), \(f\in D(A)\). Imposing two hard conditions on the operator \(A\) and using the theory of operator interpolation, the convergence with a polynomial rate with respect to the step \( \Delta \) for the finite difference methods is proved. Also, for the quasi-reversibility method to solve ill-posed Cauchy problems, the polynomial rate of convergence is proved with respect to the regularization parameter \( \varepsilon \) of the problem \[ \dot{x_{\varepsilon}}(t) = (A - \varepsilon A^{2})x_{\varepsilon}(t), \qquad x_{\varepsilon}(0)= f. \] Numerical results to illustrate the assertions of the proved theorems are not given.