Decomposition

Work out what simple blocks a part is made of. Given a solid, decomposition returns the maximal volumes it decomposes into: lumps of material with no inside corner anywhere on them, none of which is contained in another.

A part usually reads as material in more than one way — a slab with an upright on it is equally an upright with a slab around its foot — so the volumes overlap by design, and every reading is returned rather than one being chosen.

How it works

The method in pictures, start to finish, on a part small enough to hold in your head. What the algorithm does and why, before any question of how to call it.

To the walkthrough

How far a sheet reaches

The one question the method turns on, worked out on the part it is first shown on: what bounds the sheet a face is carried on as, and what the part reads like when nothing does.

To the sheets

Running it

Decomposing your own file, reading every field of the result, the four settings, and what to do when coverage comes back low.

To the how-to

Learning path

1. Read How a part is broken into simple blocks first. The result is confusing without it — in particular the volumes overlapping, so their shares add to more than the whole part.

2. Then How far a sheet reaches, which is the one step of the method that has a choice in it, and the one that decides whether a part reads as a handful of blocks or as a pile of slabs.

3. Then Decomposing a part for the call itself and the settings.

4. Judge a run by covered_fraction, not by how many volumes came back.

Key names

• Decomposer — the way in: a part, the settings, and a method for every question

• decompose() — the same run as one value, for a caller with nothing to ask

• Decomposition — the volumes, and what the run could not do

• decompose_into_cells() — the cutting stage on its own

• CellDecomposition — the cells, and the cuts refused

• DecompositionError — no result could be produced

• IndecomposableError — there was nothing to do