Semantic entries of the native Chan–He work stack #
An entry pairs the seven unsigned stack words with the three signed integer lists stored at
their bases. Memory readback reuses Block.Realizes; set contents use List.toFinset; and the
mathematical task is the existing Chan–He recursion. This is a proof invariant for native code,
not an additional machine or execution semantics.
Main definitions #
Entry: fuel, three native bases and their represented integer lists.Entry.words,Entry.task: native serialization and the existing mathematical task.Entry.Stored,Entry.Loaded: block readback and the current record's register ABI.
Main results #
Entry.image_read: the injective-array API represents exactly the listed finite set.Entry.read_injOn: duplicate-free lists meet the actual modulus/node call premise.Entry.Stored.input: exact signed-encoded native array readback.Entry.loaded_words: the seven register words match the pending entry's serialization.
The native worklist invariants and composition proofs are original.
One pending native record and the integer arrays represented by its pointers and lengths.
- fuel : ℕ
Remaining recursion levels.
The three data bases.
The lists stored at those bases, in their actual enumeration order.
Instances For
Seven unsigned stack words in the exact native pop order.
Instances For
The unchanged finite recursion task represented by an entry.
Instances For
The current seven record registers contain this entry's actual unsigned metadata.
Instances For
The finite-set input of each existing array kernel is exactly the entry's listed set.
Readback transfers along a memory equality on the three live blocks.
A prefix frame preserves every pending list whose block ends before the fresh allocation.
The seven actual register words match the exact work-stack serialization.
The native child-record words represent exactly the changed data block, with the two other columns and their original enumeration orders preserved.
The size and magnitude assumptions are the existing finite three-set admissibility notion; duplicate-free enumeration is the additional native-array obligation.
Instances For
Replacing a column by a duplicate-free subset preserves the existing size/magnitude promises.
A child reuses both untouched data blocks and the actual newly written heavy block.
Actual heavy lists and bases give exactly the existing ordered three-child worklist.
The native word-stack length records exactly seven cells per pending entry.