Spatial and temporal discretizations

When defining a field in MEDCoupling, the notions of spatial and temporal discretizations play a crucial role.

The spatial discretization details the relationship between the field and its support mesh and the temporal discretization gives an indication of the time coverage represented by the field.

Spatial discretization

A field can be supported by:

• the nodes (vertices) of the mesh: this is built with the ON_NODES keyword in the constructor of a field.

• the cells (or "elements") of the mesh: built with the ON_CELLS keyword in the constructor of a field.
• or more complex items:
  • Gauss points: built with ON_GAUSS_PT
  • Gauss points on nodes per element: built with ON_GAUSS_NE
  • Kriging points: built with ON_NODES_KR

The spatial discretization is at the center of the interpolation mechanisms, since one of the main interpolation paramter is indeed specifying from which source discretization to which target discretization one wants to go. For example:

• a P0->P0 interpolation means that a field on cells will be transfered to another cell-based field;
• a P1->P0 interpolation means that a field on nodes this time will be transfered to a cell-based field.
• etc ...

Finally, in the code itself, the class MEDCouplingFieldDiscretization is the concrete representation of this concept.

Temporal discretization

Similarly to the spatial discretization, a field object in MEDCoupling has a time discretization representing the time range that is covered by the data. It is also specified in the constructor of a field.

It can be one of:

• NO_TIME, in this case no time is attached to the field, and no time-related operation is permitted (for example unable to call getValueOn())
• ONE_TIME, the field data represent a single time step.
• LINEAR_TIME, the field data contains two arrays, stamped with two different time points. A linear interpolation of the field values between those two time steps is then possible.
• CONST_ON_TIME_INTERVAL, the field data contains a single array of data, but a start- and end-time can be specified, thus declaring that the field represent a constant set of data during this time interval. All time evaluation function then just check that the given time fits in the interval.

Finally, in the code itself, the class MEDCouplingTimeDiscretization is the concrete representation of this concept.