1: \begin{abstract}

2: In this paper, we propose an accurate finite difference method to discretize the two and three dimensional fractional Laplacian $(-\Dt)^{\fl{\ap}{2}}$ in the hypersingular integral form and apply it to solve the fractional reaction-diffusion equations. 

3: The key idea of our method is to split the strong singular kernel function of the fractional Laplacian. 

4: Hence, we first formulate the fractional Laplacian as the weighted integral of a central difference quotient  and then approximate it  by the weighted trapezoidal rule.  

5: It is proved that for $u \in C^{3,\fl{\ap}{2}}({\mathbb R}^d)$, our method has an accuracy of ${\mathcal O}(h^2)$,  uniformly for any $\ap \in (0, 2)$, while for $u \in C^{1,\fl{\ap}{2}}({\mathbb R}^d)$, the accuracy is ${\mathcal O}(h^{1-\fl{\ap}{2}})$. 

6: As $\ap \to 2^-$, the convergence behavior of our method is consistent with that of the central difference approximation of the classical Laplace operator. 

7: This  study  would fill the gap in the literature on numerical methods for the high dimensional factional Laplacian. 

8: In addition, we apply our method to solve the fractional reaction-diffusion equations and present a fast algorithm for their efficient computations. 

9: The computational cost of our method is ${\mathcal O}(M\log M)$, and the storage memory is ${\mathcal O}(M)$,  with $M$ the total number of spatial unknowns. 

10: Moreover, our method is simple and easy to implement. 

11: Various examples, including the two-dimensional fractional Allen--Cahn equation, and two- and three-dimensional fractional Gray--Scott equations, are provided to demonstrate the effectiveness of our method. 

12: It shows that our method is accurate and efficient in solving the higher dimensional fractional reaction-diffusion equation, and it can be easily applied to solve other fractional PDEs.

13: 

14: \end{abstract}

15: