1: \begin{abstract}

2: In this paper, we propose two mixed precision algorithms for Block-Jacobi preconditioner(BJAC): 

3: %for a Neumann polynomial preconditioner based on the block-Jacobi iteration: 

4: a fixed low precision strategy and an adaptive precision strategy. 

5: %We demonstrate that these two mixed precision preconditioners, when employed within Krylov subspace methods, can efficiently solve linear systems arising from diffusion and radiation hydrodynamics equations. We integrate these mixed precision preconditioners into the JXPAMG solver and 

6: We evaluate the performance improvement of the proposed mixed precision BJAC preconditioners combined with the preconditioned conjugate gradient algorithm using problems including diffusion equations and radiation hydrodynamics equations.

7: Numerical results show that, compared to the uniform high precision PCG algorithm, the mixed precision preconditioners can achieve speedups from $1.3\times$ to $1.8\times$ without sacrificing accuracy.

8: Furthermore, we observe the  phenomenon of convergence delay in some test cases for the mixed precision preconditioners, and further analyse the matrix features associate with the convergence delay behavior.

9: 

10: 

11: %Experimental results reveal that the preconditioned conjugate gradient algorithm with these mixed precision preconditioners can achieve speedups ranging from 1.3× to 1.8× compared to the uniform precision algorithm, without sacrificing accuracy. 

12: 

13: %Furthermore, we observe convergence delays in certain problems and conduct analysis of the convergence behavior of these mixed precision preconditioners. Our findings indicate that the convergence delays are associated with the multiscale feature and diagonal dominance of the linear systems.

14: 

15: \end{abstract}

16: