Skip to content

Commit

Permalink
paper: add figure references
Browse files Browse the repository at this point in the history
  • Loading branch information
freifrauvonbleifrei committed May 23, 2024
1 parent c292916 commit eed0a1b
Showing 1 changed file with 2 additions and 2 deletions.
4 changes: 2 additions & 2 deletions paper.md
Original file line number Diff line number Diff line change
Expand Up @@ -89,7 +89,7 @@ Instead of solving the problem on one grid that is very finely resolved in all d
the problem is solved on the so-called component grids which are all rather
coarsely resolved---each of them differently in the different dimensions.

![Combination scheme in two dimensions with $\vec{l}_{min} = (2,1)$ and $\vec{l}_{max} = (5,4)$, periodic boundary conditions](gfx/combischeme-2d.pdf)
![Combination scheme in two dimensions with $\vec{l}_{min} = (2,1)$ and $\vec{l}_{max} = (5,4)$, periodic boundary conditions. Figure first published in [@pollingerStableMassconservingHighdimensional2024].](gfx/combischeme-2d.pdf)

By updating each other's information throughout the simulation, the component grids
still obtain an accurate solution of the overall problem [@griebelCombinationTechniqueSolution1992].
Expand All @@ -111,7 +111,7 @@ parallelism can be re-used.
In addition, the number of process groups can be increased to leverage the
combination technique's embarrassing parallelism in the solver time steps.

![`DisCoTec` process groups: Each black square denotes one MPI rank. The ranks are grouped into the so-called process groups. Distributed operations in `DisCoTec` require either communication in the process group, or perpendicular to it---there is no need for global communication or synchronization, which avoids a major scaling bottleneck. The manager rank is optional.](gfx/discotec-ranks.pdf)
![`DisCoTec` process groups: Each black square denotes one MPI rank. The ranks are grouped into the so-called process groups. Distributed operations in `DisCoTec` require either communication in the process group, or perpendicular to it---there is no need for global communication or synchronization, which avoids a major scaling bottleneck. The manager rank is optional. Figure first published in [@pollingerStableMassconservingHighdimensional2024].](gfx/discotec-ranks.pdf)

Using `DisCoTec`, kinetic simulations were demonstrated to scale up to hundreds
of thousands of CPU cores [@pollingerStableMassconservingHighdimensional2024].
Expand Down

0 comments on commit eed0a1b

Please sign in to comment.