Integer-quotient tests for one twenty-first
For 𝑥 =1/21, nonmembership forces
the real greedy sequence eventually to select every exponent. The
integer quotient rows then agree with that sequence at growing cutoffs.
Let F21 denote the
conjunction consisting of a fatal greedy gap, finite skipped support
(hence cofinite selection), eventual alignment between the quotient rows
and the rational greedy prefix, and eventual visits to every doubling
block. Then
121∈A⟺F21 does not hold
(one_div_twenty_one_mem_iff_not_fatalAlignedBranch).
On the fatal branch the canonical quotient remainder 𝑠𝑅 is eventually strictly above the
closed capacity 2𝑅 (twentyOneFatalAlignedBranch_eventually_strict_supercapacity),
and thereafter the support appends 𝑅 +1 while 𝑠𝑅 follows one explicit affine
recurrence (twentyOneFatalAlignedBranch_eventually_affine_supercapacity).
Conversely, closed rows 𝑠𝑅 ≤2𝑅
along any unbounded sequence decay after normalisation and put 1/21 in the achievement set (twentyOneCofinalEvenQuotientGreedyDecay_of_closedRows;
one_div_twenty_one_mem_mersenneAchievementSet_of_cofinalGreedyDecay).
We give the definitions and the complete ordinary proof, including
the rounding estimate and the agreement through rank 2𝑅 needed in the short paper. Put 𝑇𝑀 =⌊2𝑀/21⌋ and 𝑞𝑀(𝑑) =⌊2𝑀/(2𝑑 −1)⌋.
Starting with 𝑇2𝑅, apply the
integer greedy rule to the weights 𝑞2𝑅(𝑑), in increasing order of 𝑑, selecting a weight exactly when it
does not exceed the current remainder. Let 𝐷𝑅 be the selected exponents in {2,…,𝑅}, and let
𝑠𝑅=𝑇2𝑅−∑𝑑∈𝐷𝑅𝑞2𝑅(𝑑).
The
alignment in F21 refers
to the full integer word through 2𝑅
agreeing with the real greedy word through 2𝑅, for all sufficiently large 𝑅. Its other clauses are a finite fatal
gap 𝑟𝑛(1/21) >𝑅𝑛 with all later
exponents selected, and a selected exponent in every sufficiently late
block (𝐾,2𝐾]. On this conjunction,
the eventual recurrence is
𝑠𝑅+1=4𝑠𝑅+⌊4(4𝑅mod21)21⌋−2𝑐𝐷𝑅(2𝑅+1)−𝑐𝐷𝑅(2𝑅+2)−(2𝑅+1+1),𝐷𝑅+1=𝐷𝑅∪{𝑅+1},𝑠𝑅>2𝑅.(15)
Proof. Put 𝑥 =1/21, let
𝐺 =𝐺1/21, and write 𝛿 =𝑥 −𝑋𝐺(2) ≥0. If 𝑥 ∉A, then 𝛿 >0 and 𝑟𝑛 ↓𝛿. Since 𝑤𝑛 →0, every sufficiently large
exponent is selected. Choose 𝑛
beyond that point with 𝑅𝑛 <𝛿. This proves the first
clause of F21, and
cofinite support gives its third clause.
We verify the integer comparisons as well. Set 𝐵 =22𝑅. When the integer and real rules
agree through 𝑑 −1, their remainders
after normalisation by 𝐵 differ by
at most 𝑑/𝐵, because each floor
changes its argument by less than one. The normalised integer weight
differs from 𝑤𝑑 by less than 1/𝐵. Each fixed real comparison is
strict: equality would give a finite representation of 𝑥, and hence 𝑥 ∈A. Choose 𝑑0 such that 𝑤𝑑 <𝛿/2 for 𝑑 >𝑑0. For large 𝑅 the finitely many comparisons through
𝑑0 agree, and 2𝑅/𝐵 <𝛿/2. At every subsequent
step 𝑑 ≤2𝑅, the normalised integer
remainder is at least 𝛿 −𝑑/𝐵 >𝛿/2 >𝑤𝑑, so that
both rules select 𝑑. Induction
proves the second clause. Conversely, a represented target is recovered
by the real greedy rule, and its remainder always satisfies 𝑟𝑛 ≤𝑅𝑛. Thus F21 is impossible for a
represented target, proving the membership equivalence.
For any 𝑅, rounding the target
and the selected weights gives
|
|
|
|𝑥−𝑋𝐷𝑅(2)−𝑠𝑅4𝑅|
|
|
|≤𝑅+14𝑅.(16)
If 𝑠𝑅 ≤2𝑅 on an unbounded sequence, the
finite subseries values 𝑋𝐷𝑅(2)
tend to 𝑥. The achievement set is
closed, so 𝑥 ∈A. This
proves the closed-return implication. A finite representation is
excluded by the finite-denominator argument already given.
On F21, the integer
comparison and cofinite support give 𝐷𝑅 =𝐺 ∩{2,…,𝑅} and 𝐷𝑅+1 =𝐷𝑅 ∪{𝑅 +1} for all large
𝑅. Equation (16)
yields 𝑠𝑅/4𝑅 →𝛿 >0, and
hence 𝑠𝑅 >2𝑅 eventually. For
𝑑 ≥2, the finite geometric
expansion of 𝑞𝑀(𝑑) gives
𝑞𝑀+2(𝑑)=4𝑞𝑀(𝑑)+2𝟏𝑑∣𝑀+1+𝟏𝑑∣𝑀+2.
Also
𝑇2𝑅+2=4𝑇2𝑅+⌊4(22𝑅mod21)21⌋,𝑞2𝑅+2(𝑅+1)=2𝑅+1+1.
Subtract the contributions of
𝐷𝑅 and of the new exponent 𝑅 +1 to obtain (15). ◻
The remaining question is whether this eventual recurrence can occur
for the actual greedy word, or whether closed returns occur at unbounded
depths. Neither assertion is established here. Since no finite support
represents 1/21, membership would
give an infinite-support rational counterexample to the universal
problem.