Lines Matching refs:metric

684 `DMPLEX` supports mesh adaptation using the *Riemannian metric framework*.
685 The idea is to use a Riemannian metric space within the mesher. The
686 metric space dictates how mesh resolution should be distributed across
688 that it is a *unit mesh* when viewed in the metric space. That is, the image
690 metric space has edges of unit length.
692 One of the main advantages of metric-based mesh adaptation is that it allows
698 See {cite}`alauzet2010` for further details on metric-based anisotropic mesh
701 The two main ingredients for metric-based mesh adaptation are an input mesh
702 (i.e. the `DMPLEX`) and a Riemannian metric. The implementation in PETSc assumes
703 that the metric is piecewise linear and continuous across elemental boundaries.
707 DMPlexMetricCreate(DM dm, PetscInt field, Vec *metric);
710 A metric must be symmetric positive-definite, so that distances may be properly
717 This routine may also be used to enforce minimum and maximum tolerated metric
755 $p$ and the target metric complexity, which is analogous to the vertex count.
770 Two different metric-based mesh adaptation tools are available in PETSc:
786 Once a metric has been constructed, it can be used to perform metric-based
790 DMAdaptMetric(DM dm, Vec metric, DMLabel bdLabel, DMLabel rgLabel, DM dmAdapt);