/- 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. -/ module public import FormalConjecturesUtil

Erdős Problem 910

References:

    erdosproblems.com/910

    [Er82e] Erdős, Paul, Some of my favourite problems which recently have been solved. (1982), 59--79. PDF

    [Er44] Erdős, P., Some remarks on connected sets. Bull. Amer. Math. Soc. 50 (1944), 442-446. PDF

    [Ru58] Rudin, M. E., A connected subset of the plane. Fund. Math. 46 (1958), 15-24.

@[expose] public sectionopen scoped Cardinalnamespace Erdos910

Does every connected set in $\mathbb{R}^n$ contain a connected subset which is not a point and not homeomorphic to the original set?

Asked by Erdős [Er44, p. 445], see also [Er82e, p. 77]. Following [Er44, pp. 443 and 445], a single point is not a connected set, so $X$ is required to have more than one point.

Erdős [Er82e, p. 77] reports that Rudin, assuming the continuum hypothesis, found counterexamples to both questions of this problem. The theorem of [Ru58] (erdos_910.variants.rudin_CH) does not say that Rudin's set is homeomorphic to its connected subsets with more than one point. A counterexample that assumes the continuum hypothesis would not disprove this statement in ZFC.

@[category research open, question, AMS 54] theorem erdos_910.parts.i : ∀ (n : ℕ) (X : Set (EuclideanSpace ℝ (Fin n))), IsConnected X → X.Nontrivial → ∃ Y ⊆ X, IsConnected Y ∧ Y.Nontrivial ∧ ¬ Nonempty (Y ≃ₜ X) := ⊢ ∀ (n : ℕ) (X : Set (EuclideanSpace ℝ (Fin n))), IsConnected X → X.Nontrivial → ∃ Y ⊆ X, IsConnected Y ∧ Y.Nontrivial ∧ ¬Nonempty (↑Y ≃ₜ ↑X) All goals completed! 🐙

If $n\geq 2$ does every connected set in $\mathbb{R}^n$ contain more than $2^{\aleph_0}$ many connected subsets?

Erdős [Er82e, p. 77] asks this for connected sets of dimension greater than one; [Er44, p. 446] asks, for these sets, for $2^{\mathfrak{c}}$ connected subsets. The condition $n\geq 2$ alone would make the answer trivially no: a segment in $\mathbb{R}^2$ has only $2^{\aleph_0}$ connected subsets, its intervals. So we require 1 < smallInductiveDimension X, which agrees with the covering dimension for subsets of $\mathbb{R}^n$. As in erdos_910.parts.i, single points are not counted as connected sets [Er44, p. 445].

Rudin's set $M$ [Ru58] has at most $2^{\aleph_0}$ connected subsets, since each one with more than one point is $M$ minus a countable set. But $M$ lies in Rudin's continuum $I=\bigcap_n\Gamma_n^*$, where each $\Gamma_n$ is a chain of $2$-cells whose links have diameter less than $1/n$ [Ru58, 1.1-1.2], so $I$, and hence $M$, has dimension at most one. Rudin's set therefore does not answer this question as Erdős states it.

@[category research open, question, AMS 54] theorem erdos_910.parts.ii : ∀ (n : ℕ) (X : Set (EuclideanSpace ℝ (Fin n))), IsConnected X → 1 < smallInductiveDimension X → 𝔠 < Cardinal.mk {Y : Set (EuclideanSpace ℝ (Fin n)) | Y ⊆ X ∧ IsConnected Y ∧ Y.Nontrivial} := ⊢ ∀ (n : ℕ) (X : Set (EuclideanSpace ℝ (Fin n))), IsConnected X → 1 < smallInductiveDimension ↑X → 𝔠 < #↑{Y | Y ⊆ X ∧ IsConnected Y ∧ Y.Nontrivial} All goals completed! 🐙

Assume the continuum hypothesis. Rudin [Ru58, p. 15] constructed a connected subset $M$ of the plane, with more than one point, such that $M\setminus N$ is at most countable for every connected subset $N\subseteq M$ with more than one point.

This disproves a conjecture of Erdős [Er44, p. 443] that every connected set $C$ contains a connected subset $C'$ such that $C\setminus C'$ has cardinality $2^{\aleph_0}$. Erdős [Er82e, p. 77] reports that Rudin's examples answer both questions of this problem in the negative; see the docstrings of erdos_910.parts.i and erdos_910.parts.ii.

@[category research solved, AMS 54] theorem erdos_910.variants.rudin_CH : (ℵ₁ = 𝔠) → ∃ M : Set (EuclideanSpace ℝ (Fin 2)), IsConnected M ∧ M.Nontrivial ∧ ∀ N ⊆ M, IsConnected N → N.Nontrivial → (M \ N).Countable := ⊢ ℵ_ 1 = 𝔠 → ∃ M, IsConnected M ∧ M.Nontrivial ∧ ∀ N ⊆ M, IsConnected N → N.Nontrivial → (M \ N).Countable All goals completed! 🐙end Erdos910