Here is the comparison for Zeldovich test using different reconstruction methods after the shock is developed, all cases use HLLC and VL, the Dual Energy parameters are:

\(\eta_1 = 0.001\) \(\eta_2 = 0.030\)

Here using PPMP:

PPMC Reconstruction:

From the CHOLLA paper:

Our implementation of this method closely follows that outlined in Stone et al. (2008).

\[\begin{aligned} \boldsymbol{W}_{L, A} &=\frac{1}{2}\left(\boldsymbol{w}_{i}+\boldsymbol{w}_{i-1}\right)-\frac{1}{6}\left(\delta \boldsymbol{w}_{i}-\delta \boldsymbol{w}_{i-1}\right) \\ \boldsymbol{W}_{R, A} &=\frac{1}{2}\left(\boldsymbol{w}_{i+1}+\boldsymbol{w}_{i}\right)-\frac{1}{6}\left(\delta \boldsymbol{w}_{i+1}-\delta \boldsymbol{w}_{i}\right) \end{aligned}\]

But from Stone et al. (2008) equation 46:

\[\begin{array}{l}{\boldsymbol{w}_{L, i}=\left(\boldsymbol{w}_{i}+\boldsymbol{w}_{i-1}\right) / 2-\left(\delta \boldsymbol{w}_{i}^{m}+\delta \boldsymbol{w}_{i-1}^{m}\right) / 6} \\ {\boldsymbol{w}_{R, i}=\left(\boldsymbol{w}_{i+1}+\boldsymbol{w}_{i}\right) / 2-\left(\delta \boldsymbol{w}_{i+1}^{m}+\delta \boldsymbol{w}_{i}^{m}\right) / 6}\end{array}\]

The signs are different. Changing the signs in ppmc_cuda.cu to match Stone (2008), I get the next result:

Seems similar to the evolution using PPMP, looks like there is an inconsistency between reconstruction methods.