← All problems

Erdős Problem #274 — The Herzog–Schönheim conjecture

49 years old · traced to

The standard precise form is the Herzog–Schönheim question. Let GG be any group, let k≥2k\geq 2 be finite, and let H1,…,HkH_1,\ldots,H_k be proper subgroups of finite index in GG. Choose x1,…,xk∈Gx_1,\ldots,x_k\in G and suppose the right cosets partition GG:

G=⋃i=1kHixi,Hixi∩Hjxj=∅(i≠j).G=\bigcup_{i=1}^{k}H_i x_i, \qquad H_i x_i\cap H_j x_j=\varnothing\quad(i\ne j).

Must two of the indices be equal; that is, must there exist i≠ji\ne j such that [G:Hi]=[G:Hj][G:H_i]=[G:H_j]? Equivalently, can such a finite exact coset partition have pairwise distinct subgroup indices? The Herzog-Schönheim conjecture for simple and symmetric groups Herzog and Schönheim (1974), Research problem no. 9

For finite GG, a coset HixiH_i x_i has size ∣G∣/[G:Hi]|G|/[G:H_i], so distinct coset sizes are equivalent to distinct indices. For infinite GG, every finite-index coset has cardinality ∣G∣|G|; the standard question is therefore about indices, not cardinalities. Finite exact partitions involve only finite-index subgroups. The Herzog-Schönheim conjecture for simple and symmetric groups Akman and Sissokho, Transversal Coset Partitions of Groups

Background

Erdős’s earlier version concerned abelian groups; Herzog and Schönheim posed the broader group-theoretic problem in 1974. The general formulation allows arbitrary groups and finitely many cosets, rather than only the abelian setting of the earlier question. Erdős Problem #274 Herzog and Schönheim (1974), Research problem no. 9

A basic arithmetic constraint comes from counting the cosets at each index. If ndn_d cosets in the partition have index dd, then their relative sizes give the necessary identity

∑dndd=1.\sum_d \frac{n_d}{d}=1.

This unit-fraction condition can exclude candidate index lists, but is not by itself a proof that two indices must coincide. Akman and Sissokho, Transversal Coset Partitions of Groups

References

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.