Continuous dependence for backward parabolic operators with log-Lipschitz coefficients (Q834781): Difference between revisions

The authors considered the Cauchy problem for the following backward parabolic equation: \[ u_t + \sum^n_{ i,j=1}\partial_x{}_i{} (a_{i,j} (t, x)\partial_x{}_j{} u) +\sum^n_ {j=1} b_j (t, x)\partial_x{}_j{} u + c(t, x)u = 0,\quad (t, x) \in [0, T ] \times \mathbb R^n . \] It is well known that the Cauchy problem is an ill-posed problem in any reasonable functional space, due to the smoothing effect of forward parabolic operators. But the authors exploit the weighted energy estimates, similar to those of Glagoleva and Hurd, and obtain the desired results about continuous dependence of solutions on the initial data. Finally, in the Appendix they outline the construction of a counterexample to Hölder continuous dependence.