Output

Toucan writes an orientation table during initialization and rank-local microstructure regions as rolling substrate windows become complete. Output coordinates use the RDF origin/resolution and are written in the same spatial units as the input (meters for Toucan’s documented JSON interface).

Orientation table

Rank 0 writes Orientations.csv in the configured output directory. Each row contains:

dirID,d1x,d1y,d1z,d2x,d2y,d2z,d3x,d3y,d3z

The three vectors form one randomly generated orthonormal orientation basis. Rows are written in orientation-array order. The current orientation file labels those rows 1 through NumGrainOrientations, while grain arrays store their orientation index in dirID.

Region selection

Output/Selection/Dims controls geometry:

Value	Result
`XYZ`	One sampled three-dimensional region per completed window.
`XY`	One or more constant-z planes.
`XZ`	One or more constant-y planes.
`YZ`	One or more constant-x planes.
`None`	No microstructure-region files.

Stride samples indices independently along x, y, and z. Offset selects indices congruent to that offset modulo the stride. Plane filenames include .Slice.Xn, .Slice.Yn, or .Slice.Zn before the extension. Output is rank-local:

Rank.<rank>.Output.<window>.csv
Rank.<rank>.Output.<window>.xmf
Rank.<rank>.Output.<window>.<attribute>.bin

Slice suffixes appear between <window> and the extension. Rank and window numbers are zero-padded according to the MPI size and configured layer count. Internal MPI halo cells are omitted from the selected rank region.

CSV

CSV output produces one file per selected rank-local region. Its base columns are:

x,y,z,dirID

Coordinates are physical positions. Enabling UniqueIdentifiers appends repeatID and nucleatedRankID. Enabling UnderResolved appends underResolved as 0 or 1. A baseplate grain uses -1 for the CSV nucleating-rank field. CSV writes only substrate cells that have received a simulated grid state.

XDMF

XDMF output produces an .xmf metadata file plus raw binary arrays. The XDMF topology is a 3D rectilinear mesh with origin and spacing derived from the RDF header and output stride. The always-present node attribute is dirID.

Optional attributes are:

repeatID and nucleatedRankID when UniqueIdentifiers is enabled;
underResolved when UnderResolved is enabled.

Each attribute is stored as a contiguous signed 32-bit integer array in a sibling file named from the region base plus the attribute name and .bin. The x index varies fastest in these output arrays. Integer value -1 represents cells outside the transferred grid and baseplate nucleating-rank IDs.

Grain identity

dirID identifies crystallographic orientation and can repeat across grains. When unique identifiers are enabled, the tuple of direction ID, repeat ID, and nucleating rank is the propagated grain identity used by Toucan. Nucleated grains receive a repeat counter for their orientation on their MPI rank; baseplate grains use their generated repeat index and have no nucleating rank. The under-resolution field is set when the selected capture time precedes the captured cell’s liquidus-crossing time, then propagates with the grain state.