Notes. The first column indicates the designation of the run. The second column corresponds to the relative amplitudes of the initial sine waves, namely ϵ_2D = ϵ_v/ϵ_x and ϵ_3D = (ϵ_v/ϵ_x, ϵ_z/ϵ_x) for two and three sine waves, respectively. The third column gives the value of ϵ_x. The fourth and fifth columns respectively give the spatial resolution ng of the grid used to solve the Poisson equation and the spatial resolution n_s of the mesh of vertices used to construct the initial tessellation (see Sect. 3 for details). The sixth column indicates the scale factor at the shell-crossing estimated by LPT extrapolated to the infinite order (Saga et al. 2018), while the seventh one provides the value a_sc measured in the Vlasov runs (see Appendix B.1) and which is actually used in Sect. 4.2. The eighth column indicates the value â_sc of the expansion factor of the closest available simulation snapshot to the collapse time for the comparisons performed in Sect. 4. Finally, the last column indicates the value of the expansion factor used for the analyses performed beyond the collapse time in Sect. 5. We note that with the normalisation of the scale factor used here, the value of a_sc has to be multiplied by |ϵ_x| to obtain a more intuitive collapse time, which would be equal to unity in the purely 1D case.