1: \begin{abstract}

2:   In \cite{wang2023towards}, a dual-consistent

3:   dual-weighted residual-based $h$-adaptive method has been proposed

4:   based on a Newton-GMG framework, towards the accurate calculation of a

5:   given quantity of interest from Euler equations. The performance of

6:   such a numerical method is satisfactory, i.e., the stable

7:   convergence of the quantity of interest can be observed in all

8:   numerical experiments. In this paper, we will focus on the

9:   efficiency issue to further develop this method, since efficiency is

10:   vital for numerical methods in practical applications such as the

11:   optimal design of the vehicle shape. Three approaches are studied

12:   for addressing the efficiency issue, i.e., i). using convolutional

13:   neural networks as a solver for dual equations, ii). designing an

14:   automatic adjustment strategy for the tolerance in the $h$-adaptive

15:   process to conduct the local refinement and/or coarsening of mesh

16:   grids, and iii). introducing OpenMP, a shared memory parallelization

17:   technique, to accelerate the module such as the solution

18:   reconstruction in the method. The feasibility of each approach and

19:   numerical issues are discussed in depth, and significant

20:   acceleration from those approaches in simulations can be observed

21:   clearly from a number of numerical experiments. In convolutional

22:   neural networks, it is worth mentioning that the dual consistency

23:   plays an important role to guarantee the efficiency of the whole

24:   method and that unstructured meshes are employed in all

25:   simulations.

26: \end{abstract}