1. Basic concepts

Your’re invited to start here.

This section explains in a concise way fundamental concepts (incarnated by classes) for a better understanding of examples in FAQ.

The objects presented in this section are objects dedicated to data manipulation.

Later in advanced section, other objects will be presented to more complex data, as composite of objects presented here.

1.1. medcoupling Fields

The most fundamental object is field. Fields are incarnated by MEDCouplingFieldDouble.

A field is an object able to give a value on every points geometrically covered by its support. The field support is a mesh.

Depending on the physics relative to the field, medcoupling fields proposes different spatial discretizations : cells, nodes, Gauss points.

The spatial discretization defines an algorithm giving the value of field on point given the position of a point inside the domain covered by the mesh and an array of float, integers located on specific location of mesh.


1.2. medcoupling Meshes

Meshes are incarnated by MEDCouplingMesh (and their subclasses). A mesh contains cells and nodes (indifferently called points). A cell has a geometric type (TRI3, QUAD4, TETRA4).

All cells contained in a medcoupling mesh must have the same dimension. This dimension is called meshdimension.

The set of points of a medcoupling mesh are stored into coordinates array. The number of components of coordinates of coordinates array is called the space dimension.

The space dimension is always greater or equal to mesh dimension.

1.3. medcoupling Arrays

One of the challenges faced by medcoupling is the reduction of memory footprints. In medcoupling, memory expensive attributes are arrays or composite of arrays.

To do so, medcoupling arrays represent contiguous arrays in the most compact way to guaranty at most locality.

The type of elements contained in arrays is fix. Today int32, float32 and float64 arrays are available. Elements in arrays are grouped into fixed size packets called tuple.

This size of every tuple is called number of components.

Consequently number of elements in an array is equal to number of tuples times number of components.

A typical usage of medcoupling array is for coordinates of points storage. Number of components will be equal to space dimension and number of tuples will be equal to number of points.

A medcoupling array has a name. And each component of array has also a name. The component name use the following convention to put an optional unity (“X cote [mm]”).

If you are already a fan of numpy (you’re right it’s an amazing/wonderful standard package), medcoupling arrays behaves just like numpy arrays and anyway there are zero copy gateways between medcoupling arrays and numpy.

medcoupling arrays implement different algorithms like reordering, cloud comparisons, arithmetic, geometry helpers in addition to algorithms proposed by numpy.