1: \begin{abstract}

2: The multigrid-reduction-in-time (MGRIT) technique has proven to be successful

3: in achieving higher run-time speedup by exploiting parallelism in time.

4: The goal of this article is to develop and analyze a MGRIT algorithm,

5: using FCF-relaxation with time-dependent time-grid propagators,

6: to seek the finite element approximations of unsteady fractional Laplacian problems.

7: {\color{red}The multigrid with line smoother proposed in [L. Chen, R. H. Nochetto, et al.,

8: Math. Comp. 85 (2016) 2583--2607] is chosen to be the spatial solver.}

9: Motivated by [B. S. Southworth, SIAM J. Matrix Anal. Appl. 40 (2019) 564--608],

10: we provide a new temporal eigenvalue approximation property and then

11: deduce a generalized two-level convergence theory which removes the previous unitary diagonalization assumption

12: on the fine and coarse time-grid propagators

13: required in [X. Q. Yue, S. Shu, et al.,

14: Comput. Math. Appl. 78 (2019) 3471--3484].

15: Numerical computations are included to confirm the theoretical predictions and demonstrate

16: the sharpness of the derived convergence upper bound.

17: \end{abstract}

18: