view_offset Namespace Reference

Functions
	_decode (el)
	read_vtu (path)
	boundary_edges (tris, inside)
	to3d (pts2)
	compact (nodes, edges)
	add_run (label, vtu, color, seen_inputs)
	main (argv)

Variables
dict	VTK_DTYPE
list	PALETTE

Detailed Description

Polyscope viewer for topological_offset results.

Every layer is registered separately so it can be toggled in the UI: the input complex, and
per target distance the offset boundary and the offset region itself.

    pip install polyscope numpy
    python3 view_offset.py <outdir> [<outdir> ...]

Each <outdir> is a run_configs.py output directory -- it is scanned for */out.vtu. Pass the
construction and the optimized directory together to compare them in one window:

    python3 view_offset.py plot_con plot_opt

Layers are named "<dir>/<run> offset" and "<dir>/<run> band", plus one "input <tag>" per input
tag. Every layer starts hidden, so the window opens empty and you switch on exactly what you want
to compare. Everything is drawn in the z=0 plane and viewed
orthographically along z, with planar navigation.

Function Documentation

_decode()

view_offset._decode (   el )

protected

VTK inline 'binary' is base64 of [UInt64 nbytes][payload], uncompressed.

add_run()

view_offset.add_run	(		label,
			vtu,
			color,
			seen_inputs
	)

Register the offset band, its boundary, and every input tag present as its own layer.

Which cell array means what is dataset-dependent -- the annots model calls the input
complex tag_1 and the offset tag_5, the disk example is the other way round -- so nothing
is hardcoded: `offset_tag` is written by the offset itself and names the band, and every
other tag_* array is registered as an input layer under its own name.

boundary_edges()

view_offset.boundary_edges	(		tris,
			inside
	)

Edges incident to exactly one `inside` triangle -- the boundary of that region.

compact()

view_offset.compact	(		nodes,
			edges
	)

Keep only the nodes the edges actually use, and reindex.

Polyscope draws a sphere at EVERY node of a curve network, so handing it the whole mesh
vertex array renders thousands of stray dots for vertices no edge touches.

read_vtu()

view_offset.read_vtu

(

path

)

-> (points Nx2, triangles Mx3, {cell array name: values})

Variable Documentation

PALETTE

list view_offset.PALETTE

Initial value:

=  [
    (0.12, 0.44, 0.71),
    (0.18, 0.62, 0.49),
    (0.79, 0.54, 0.11),
    (0.71, 0.28, 0.24),
    (0.49, 0.35, 0.66),
    (0.35, 0.35, 0.35),
]

VTK_DTYPE

dict view_offset.VTK_DTYPE

Initial value:

=  {
    "Float64": np.float64,
    "Float32": np.float32,
    "Int64": np.int64,
    "UInt64": np.uint64,
    "Int32": np.int32,
    "UInt8": np.uint8,
}