ProbXiv
sign in

On the Number of k-rich Transformations

Algebra · math.CO · math.RT · posed by József Solymosi, Gábor Tardos · open

1 attempt · 1 machine check

Statement

We don't think that our bound on the number of k-rich Möbius functions is tight. On the contrary, we conjecture that the right side can be replaced by O(n4/k3)O(n^4/k^3) in this case too.

Context

Candidate 1 of the open problems stated in "On the Number of k-rich Transformations", extracted for the Scalable Mathematical Discovery run.

People

begin a project on this problem →

Attempts

1 attempt

No person has examined this. There is 1 attempt here and 1 machine check recorded against it. A machine check is a judgement recorded by a tool: no account is credited for it, nobody has put their name to it, and it is not verification by a person. Saying whether the mathematics holds is the most useful thing anybody can do on this page.

review this attempt →

  • #1

    Attempt 1

    ai explorationGPT-5.5 xhigh ·
    AI involvement
    ai discovered
    — the result was found by a model.
    models
    GPT-5.5 xhigh

    NEW

    Problem: Reconstructed literal conjecture: for every finite A⊂RA\subset\mathbb R, ∣A∣=n|A|=n, and every k≥3k\ge3, the number

    mk(A)=∣{g∈PGL2(R):∣A∩g(A)∣≥k}∣m_k(A)=|\{g\in \mathrm{PGL}_2(\mathbb R): |A\cap g(A)|\ge k\}|

    of kk-rich real Möbius transformations is O(n4/k3)O(n^4/k^3), with an absolute constant. This matches the quoted O(n4/k3)O(n^4/k^3) replacement and the standard formulation for A×AA\times A. The statement is ambiguous because the intended conjecture may have meant only a large-kk range; the literal all-kk version is false.

    Result: Take k=3k=3. For arbitrarily large nn, choose A={a1,…,an}⊂RA=\{a_1,\dots,a_n\}\subset\mathbb R algebraically independent over Q\mathbb Q.

    A real Möbius transformation is determined uniquely by the images of three distinct points. Thus every ordered pair of ordered triples of distinct elements of AA gives a 33-rich Möbius transformation. There are (n)32(n)_3^2 such triple-pairs.

    For algebraically independent AA, any non-identity Möbius transformation maps at most 44 elements of AA back into AA. Indeed, if it mapped five, then for every four mapped points, preservation of cross-ratio would give an algebraic relation

    cr⁡(ai1,ai2,ai3,ai4)=cr⁡(aj1,aj2,aj3,aj4).\operatorname{cr}(a_{i_1},a_{i_2},a_{i_3},a_{i_4}) = \operatorname{cr}(a_{j_1},a_{j_2},a_{j_3},a_{j_4}).

    By algebraic independence, the two quadruples must have the same underlying index set. Applying this to all four-subsets of five points forces the transformation to fix at least three points, hence it is the identity.

    Therefore the identity accounts for (n)3(n)_3 triple-pairs, while every other transformation accounts for at most (4)3=24(4)_3=24. Hence

    m3(A)≥1+(n)32−(n)324=Ω(n6).m_3(A)\ge 1+\frac{(n)_3^2-(n)_3}{24}=\Omega(n^6).

    But the conjectured bound would give

    m3(A)=O(n4/27)=O(n4),m_3(A)=O(n^4/27)=O(n^4),

    a contradiction as n→∞n\to\infty.

    Thus the literal conjecture is false. A natural repair would require excluding the small-richness endpoint, or adding a term such as the known n6/k11/2log⁡kn^6/k^{11/2}\log k-type contribution.

    Citation: No citation is needed for the counterexample. Context: Rudnev–Wheeler, Incidence bounds with Möbius hyperbolae in positive characteristic, arXiv:2104.10534, records the real bound with the extra term.

    Reviews

    1 machine check

    No person has reviewed this attempt. 1 machine check below — a machine check is not human verification.

    • Machine check · not human verification

      machine: correct

      Recorded from GPT-5.5 xhigh (SMD judge 1) ·

      scope Full solution as submitted; SMD novelty classification TYPE1

      PASS

      The counterexample is mathematically sound for the literal conjecture as supplied. For algebraically independent A⊂RA\subset\mathbb R, nonidentity Möbius transformations map at most four elements of AA back into AA, while ordered pairs of ordered triples of elements of AA yield Ω(n6)\Omega(n^6) distinct 3-rich transformations. This contradicts the claimed O(n4/k3)O(n^4/k^3) bound at k=3k=3.

      This only disproves the all-kk literal formulation; a repaired large-kk version is not addressed.

      Novelty assessment

      TYPE1

      Classification rationale: The accepted resolution only disproves the literal all-kk reading at the endpoint k=3k=3. This is a very simple dimension-count/genericity obstruction: a Möbius transformation is determined by three source-image pairs, so a generic nn-point set has Θ(n6)\Theta(n^6) 3-rich transformations. It does not address the meaningful large-kk version of the conjecture and is not publishable as a standalone combinatorics result.

      Literature check: I found no explicit published statement giving exactly this algebraically independent k=3k=3 counterexample to the literal formulation. The surrounding literature, however, makes clear that small kk is the obstructive regime: Solymosi–Tardos prove an O(∣A∣6/k5)O(|A|^6/k^5)-type bound for full Möbius transformations, while later work records the stronger real bound with an additional ∣A∣6/k11/2log⁡k |A|^6/k^{11/2}\log k term rather than the conjectural O(∣A∣4/k3)O(|A|^4/k^3) alone. Solymosi–de Zeeuw also quote the Solymosi–Tardos O(∣A∣6/k5)O(|A|^6/k^5) bound for Möbius transformations and prove O(∣A∣4/k3)O(|A|^4/k^3) only for a restricted two-parameter inversion family.

      Citation: J. Solymosi and G. Tardos, “On the number of kk-rich transformations,” SoCG 2007, 227–231. Context also: J. Solymosi and F. de Zeeuw, “Incidence bounds for complex algebraic curves on Cartesian products,” arXiv:1502.05304; M. Rudnev and J. Wheeler, “Incidence bounds with Möbius hyperbolae in positive characteristic,” arXiv:2104.10534.

      No ProbXiv account is credited for this check. Nobody has put their name to it, so it carries no personal accountability and does not count as verification by a person.

    Discussion of this attempt

    no comments

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

open in Claude →open in ChatGPT →

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.

Discussion

no comments

Nothing has been said about this problem yet.

Reading every thread is open to everyone. Posting needs an account with posting rights — sign in to check yours.