Shared-state elimination for the two anchored runs #
The local one-parity eliminator in SingleScaleInnerAssembly returns a
number and therefore forgets the final set of labeled copies. At an
interior anchor, however, the right outward run must see every copy already
used by the left outward run. This file supplies the continuation-passing
variant needed for that finite-Fubini step.
The terminal continuation receives both the final used set and the final
lattice point. In conditionedNXAnchoredRunsSum the continuation keeps the
former and deliberately discards the latter: the right run restarts at the
anchor point, so no artificial edge between the two outer endpoints is
introduced.
Continuation-passing version of conditionedNXParityChainSum.
At the empty schedule it exposes the final used-copy set and final point to
tail; all nonempty cases are exactly the existing conditioned eliminator.
Equations
- One or more equations did not get rendered due to their size.
- Anderson4D.XYCluster.conditionedNXParityChainSumThen ht hroot Nm mu z hz R [] x✝² x✝¹ x✝ = x✝ x✝² x✝¹
Instances For
The CPS eliminator is monotone in its terminal continuation.
A constant terminal continuation factors out of the CPS eliminator. This also records that the existing eliminator is the unit-terminal special case.
Unit continuation recovers conditionedNXParityChainSum exactly.
A nonnegative terminal continuation gives a nonnegative CPS sum.
Uniform target-product estimate with an arbitrary bounded terminal
continuation. The constant is exactly the one from
conditionedNXParityChainSum_le_targetProduct.
The same continuation estimate with the constant power made explicit.
Two outward runs with one genuinely shared used-copy state #
Run the left schedule first and pass its final used-copy set to the right
schedule. The right schedule restarts at anchorPoint; the terminal point
of the left schedule is intentionally ignored.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Both outward runs remain nonnegative when they share the used set.
The shared-state two-run sum is bounded by the product of the two local target products. The same universal constant is used on both sides.
Paper-facing normalization of the two-run bound: one factor C per
elimination block, followed by the unscaled left and right target products.