Example 2.4 (Scott 1981, PRG-19, §2) — eliminating the first run of 1's, `g : #
B → B`
Scott's second approximable mapping reads a binary sequence left to right and
"eliminates the first
consecutive run of 1's while copying all the other digits." Thus
g(0ⁿ 1ᵏ 0 x) = 0ⁿ⁺¹ x (for k > 0), g(1^∞) = ⊥, g(0ⁿ 1^∞) = 0ⁿ.
It is instructive because it turns total inputs (e.g. 1^∞) into partial
outputs.
We give the guaranteed finite output out : Σ* → Σ* for each finite input
prefix via a two-state
scan: out copies leading 0's and, on the first 1, hands over to del, which
swallows the run
of 1's and — on the terminating 0 — emits that single 0 and copies the rest
verbatim. The
elementwise value on the finite element ↑(σΣ*) is ↑((out σ)Σ*), so the
neighbourhood relation is
X g Y ↔ ∃ σ, X = σΣ* ∧ (out σ)Σ* ⊆ Y ∧ Y ∈ B.
out is prefix-monotone (out_mono: σ <+: σ' → out σ <+: out σ'), which gives
Definition
2.1(iii); (i) and (ii) are the principal-filter facts for the cone (out σ)Σ*.
Constructive.
Inside the first run of 1's: swallow 1's; on the first 0, emit it and
copy the rest.
Equations
Instances For
The guaranteed output prefix: copy leading 0's; on the first 1, eliminate
the run via del.
Equations
Instances For
Prefix-monotonicity of the output. Extending the input prefix can only
extend the guaranteed
output: σ <+: σ' → out σ <+: out σ'. This is the heart of Definition 2.1(iii)
for g.
Example 2.4 — the run-eliminating mapping g : B → B. X g Y iff X = σΣ* and the cone of
the guaranteed output (out σ)Σ* approximates Y. Definition 2.1: (i) out [] = [] so Δ g Δ;
(ii) a fixed cone has a unique prefix, and the value is the principal filter
↑((out σ)Σ*), closed
under ∩; (iii) out_mono shrinks the output cone as the input cone shrinks.
Equations
- One or more equations did not get rendered due to their size.