/-
Copyright 2026 The Formal Conjectures Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-/modulepublicimportFormalConjecturesUtil
Bugeaud Collection of Conjectures and Open Questions: Rapidly Increasing Sequences Dense Modulo One
Problem 10.6 asks for a "very rapidly increasing" sequence $(m_n){n \ge 1}$ of positive
integers with $({\xi m_n}){n \ge 1}$ dense modulo one for every irrational $\xi$. The
informal phrase admits several readings, and they are not equivalent:
a lower bound on the size of the terms, $m_n \ge f(n)$;
an upper bound on the counting function of ${m_n}$;
a lower bound on the consecutive ratios, $m_{n+1}/m_n \ge 1 + c/\log n$.
The first two readings are vacuous, and exists_dense_of_growth_and_sparsity below says so: a
sequence built from runs of consecutive integers is dense modulo one for every irrational $\xi$
by an elementary argument, and the runs may be placed so that its terms dominate any prescribed
$f$ while its counting function stays below $(\log\log N)^2$. In the subexponential range the
growth half is a special case of [Bos83, Thm. 1.5], which produces such a sequence in any
prescribed subexponential corridor and in the stronger uniform-distribution form; the sparsity
half sits below the $\log N$ floor that every sublacunary sequence obeys, and so below the
counting function of [Kat16, Cor. 4.10]. Consequently problem_10_6_variant_2, which asks only
for intermediate growth, is not open.
The third reading is IsGenuinelySublacunary, and it is the only one with content: it implies
the first (hasIntermediateGrowth_of_isGenuinelySublacunary), and it is not answered by runs,
since inside a run $m_{n+1} - m_n = 1$. That reading is problem_10_6_variant_1, which remains
open; [Kat16, Cor. 4.9] gives a universally densifying sequence of multiplicative shape
${2^n 3^e}$ whose ratio behaviour reduces the question to a Diophantine statement about
$\log_2 3$.
The three statements that need a long proof carry a formal_proof link to the version of this
file that contains it: the construction by runs, its counting function, and the growth estimate
behind hasIntermediateGrowth_of_isGenuinelySublacunary.
References:
[Bos83] Boshernitzan, Michael D. "Homogeneously distributed sequences and Poincaré sequences
of integers of sublacunary growth." Monatshefte für Mathematik 96.3 (1983): 173-181.
[Bos94] Boshernitzan, Michael D. "Density modulo 1 of dilations of sublacunary sequences."
Advances in Mathematics 108.1 (1994): 104-117.
[Bug12] Bugeaud, Yann. "Distribution modulo one and Diophantine approximation."
Vol. 193. Cambridge University Press, 2012. Chapter 10.
[Fur67] Furstenberg, H. "Disjointness in ergodic theory, minimal sets, and a problem
in diophantine approximation". Math. Systems Theory 1, 1–49 (1967).
[Kat16] Katz, Asaf. "Generalizations of Furstenberg's Diophantine result."
arXiv:1607.00670 (2016).
[Mat80] de Mathan, Bernard. "Numbers contravening a condition in density modulo 1."
Acta Mathematica Hungarica 36.3-4 (1980): 237-241.
[Pol79] Pollington, Andrew Douglas. "On the density of sequence ${n_ {k}\xi} $."
Illinois Journal of Mathematics 23.4 (1979): 511-515.
The Pollington–de Mathan theorem [Pol79][Mat80]. For every lacunary sequence
$(m_n){n \ge 1}$ of positive integers, the set of real numbers $\xi$ for which
$({\xi m_n}){n \ge 1}$ is not dense modulo one has full Hausdorff dimension.
Furstenberg's theorem [Fur67] (the $\times 2, \times 3$ case). For every irrational
number $\xi$, the two-parameter family $({\xi , 2^m 3^n})_{m, n \ge 1}$ is dense modulo
one.
Boshernitzan's theorem [Bos94]. Given a real sublacunary sequence $r$, the set of
real numbers $\xi$ for which $({\xi r_n})_{n \ge 1}$ is not dense modulo one has
Hausdorff dimension zero.
The counting function $\pi_A(N) = #(A \cap [0, N])$ of the range $A$ of m. The outer
Finset.range (N + 1) is harmless for a strictly increasing m, since then $n \le m_n$.
A lower bound on the size of the terms cannot make Problem 10.6 nontrivial. For every
growth demand f there is a strictly increasing sequence dominating f at every index which
is dense modulo one for every irrational $\xi$.
In the subexponential range this is a special case of [Bos83, Thm. 1.5], which produces such a
sequence inside any prescribed corridor $[n_k, m_k)$ with $m_k - n_k \to \infty$ and
$m_k / n_k \to 1$, and in the stronger form of uniform distribution; the elementary
construction of the linked proof is uniform in f and has no growth ceiling.
Neither can an upper bound on the counting function, even together with the growth
demand. The same sequence dominates an arbitrary fand has counting function below
$(\log\log N)^2$ — an exponential below the $\log N$ floor that every sublacunary sequence
obeys, and so below the counting function of [Kat16, Cor. 4.10]. So a reading of "very rapidly
increasing" that constrains only the size of the terms or the sparsity of their range leaves
Problem 10.6 with no content.
A ratio floor is weaker than lacunarity: a lacunary sequence has ratios bounded below by a
constant $> 1$, which dominates $1 + c/\log n$. So IsGenuinelySublacunary is a floor under
the ratios, not a denial of lacunarity.
The sequence $m$ eventually grows at least as fast as $\exp(n^{\alpha})$, i.e., super-exponential
growth when $\alpha > 1$, and stretched-exponential when $0 < \alpha < 1$.
A ratio floor implies intermediate growth, at every exponent $\alpha < 1$. So the
ratio reading of Problem 10.6 is strictly stronger than the growth reading, and by
exists_dense_of_growth it is the reading that carries the content of the problem.
mSeq has intermediate (subexponential but super-polynomial) growth: for every
0 < α < 1 its terms eventually dominate $\exp(n^\alpha)$. This is the ratio floor of
example_isGenuineSublacunary fed into
hasIntermediateGrowth_of_isGenuinelySublacunary.
A sequence that is dense modulo one for every irrational $\xi$ cannot be lacunary: its
exceptional set is contained in $\mathbb{Q}$, hence countable and of Hausdorff dimension zero,
whereas pollington_de_mathan gives dimension one for a lacunary sequence. So an answer to
Problem 10.6 under the ratio reading is pinched: the ratios must return to $1$ along a
subsequence, and yet stay above $1 + c/\log n$ eventually. The two are compatible only because
$c/\log n \to 0$.
Problem 10.6. Find a very rapidly increasing sequence $(m_n){n \ge 1}$ of positive
integers such that $({\xi m_n}){n \ge 1}$ is dense modulo one for every irrational
number $\xi$. Note: Furstenberg's $2^m3^n$ is sublacunary but requires two parameters.
This is the reading of "very rapidly increasing" as a lower bound on the consecutive ratios,
and it is the one that remains open. It is not answered by runs of consecutive integers, since
inside a run $m_{n+1} - m_n = 1$; and by not_isLacunary_of_dense any answer is pinched
between a decaying ratio floor and the failure of lacunarity. [Kat16, Cor. 4.9] gives a
universally densifying sequence of the multiplicative shape ${2^n 3^e}$, for which the
ratio condition reduces to a Diophantine statement about the differences of the exponents
$e$ times $\log_2 3$ modulo one.
Problem 10.6, intermediate-growth variant. This asks only for a lower bound on the size of
the terms, and under that reading the problem is not open: runs of consecutive integers answer
it, by exists_dense_of_growth applied to $f(n) = \lceil e^n \rceil$. The witness below is
exhibited at $\alpha = 1/2$; the same sequence works for every $\alpha < 1$, since
$\exp(n^\alpha) \le \exp(n) \le f(n)$ there.
The result is not new: [Bos83, Thm. 1.5] gives a sequence of any prescribed subexponential
growth that is uniformly distributed — not merely dense — modulo one for every irrational
dilate.