TetGen#

.ele + .node read + write eager

Summary of the specification#

TetGen splits a mesh across sidecar files that share a basename. .node starts with a counts line — number of points, dimension, number of attributes, boundary-marker flag — then one line per point: an index, its coordinates, its attributes, and its marker if the flag is set. .ele starts with its own counts line — number of tetrahedra, nodes per tetrahedron, region-attribute flag — then one line per element: an index and its node indices, with an optional trailing region attribute. Indices may be 0- or 1-based; the first index in the file decides.

Specification at a glance#

.node header	npoints dim nattrs boundary_marker_flag
.node line	index x y z [attrs…] [marker]
.ele header	ntetrahedra nodes_per_tet region_attr_flag
.ele line	index n1 n2 n3 n4 [region]
index base	0 or 1, inferred from the first index
companions	.face, .edge, .neigh use the same convention

Read the full TetGen specification ↗

Reading#

import polyxios as px

mesh = px.read("model.ele")
mesh.vertices          # (n, 3)
mesh.element_types     # element groups found in the file

Writing#

px.write(mesh, "out.ele")

This codec takes no format-specific options.

Quirks worth knowing#

Pass either path — px.read("bar.ele") or px.read("bar.node") — and the paired file is located by basename.
Boundary markers become vertex tags and region attributes become element attributes, so both labelling schemes are kept.
Both 0- and 1-based indexing are handled; the base is inferred rather than assumed.