1: \begin{abstract}

2: We aim at the development and analysis of the numerical schemes for approximately solving the backward diffusion-wave problem,

3: which involves a fractional derivative in time with order $\alpha\in(1,2)$.

4: From terminal observations at two time levels, i.e., $u(T_1)$ and $u(T_2)$,

5: we simultaneously recover two initial data $u(0)$ and $u_t(0)$ and hence the solution $u(t)$ for all $t > 0$.

6: First of all, existence, uniqueness and Lipschitz stability of the backward diffusion-wave problem were established

7: under some conditions about $T_1$ and $T_2$.

8: Moreover, for noisy data, we propose a quasi-boundary value scheme to regularize the "mildly" ill-posed problem,

9: and show the convergence of the regularized solution.

10: Next, to numerically solve the regularized problem, a fully discrete scheme is proposed

11: by applying finite element method in space and convolution quadrature in time.

12: We establish error bounds of the discrete solution in both cases of smooth and nonsmooth data.

13: The error estimate is very useful in practice since it indicates the way

14: to choose discretization parameters and regularization parameter, according to the noise level.

15: The theoretical results are supported by numerical experiments.

16: \end{abstract}

17: