Agathe Herrou
sous la direction de Nicolas Bonneel, Julie Digne et Bruno Lévy
Power diagrams are equivalent to optimal transport plans [Aurenhammer et al, 1998]
Power diagrams are equivalent to optimal transport plans [Aurenhammer et al, 1998]
Voronoi diagram: all the sites have the same influence
Voronoi diagram: all the sites have the same influence
Limited capacities: adapt the assignment
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Cell size increases as its weight decreases
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Collectively adjust the weights so that the areas match the capacities
Power diagrams are equivalent to optimal transport plans [Aurenhammer et al, 1998]
Cost of transporting a cell to its site
Constraint: difference between the prescribed masses and the actual masses
Weights act as Lagrange multipliers
$F$ is concave in $(\phi)$: stationary point is a maximum
Algorithm | Author(s) | Speed | Proven convergence speed | Requirements |
---|---|---|---|---|
Gradient ascent | Aurenhammer et al, 1998 | Slow | No | None |
Multiscale | Mérigot, 2011 | Fast | No | None |
Quasi-Newton (L-BFGS) | Lévy, 2014 | Fast | No | None |
Newton - KMT | Kitagawa et al, 2017 | Very fast | Yes | No empty cells |
Sample mesh
Compute transport
Compute barycenters
Compute mesh
Filter triangles
[Lévy 2014]
Sampling size | Algorithm | Computation time |
---|---|---|
500 | Symmetric | 83s |
500 | Non-symmetric | 2s |
6500 | Non-symmetric | 79s |
where:
Stationary point
$$
\begin{cases}
\nabla F(X) + J_B(X)^\top\cdot\Lambda = 0\\
B(X) = 0
\end{cases}
$$
Take a step and linearize $$ \begin{cases} \nabla F(X + \delta X) = \nabla F(X) + H_F(X)\delta X + o(\lVert \delta X \rVert)\\ B(X + \delta X) = B(X) + J_B(X)\delta X + o(\lVert \delta X \rVert) \end{cases} $$
Reynolds transport theorem [Reynolds 1903]
Reynolds transport theorem [Reynolds 1903]
Reynolds transport theorem [Reynolds 1903]
Let: