Відповідь - так. Припустимо , що ми маємо розкладання на множники Q = A ⋅ BQ=A⋅B .
Одне легке зауваження - це те, що AA і BB повинні бути неперервними (оскільки для w ∈ A ∩ Bw∈A∩B ми отримаємо w 2 ∈ Qw2∈Q ). Зокрема, лише один з A , BA,B може містити ϵϵ . Можна вважати , без втрати спільності (так як в іншому випадку цілком симетрично) , що е ∈ Bϵ∈B . Тоді , так як і б не може бути врахований в непусті чинники, ми повинні мати , б ∈ A .aba,b∈A
Далі отримаємо, що a m b n ∈ Aambn∈A (і, абсолютно аналогічно, b m a n ∈ Abman∈A ) для всіх m , n > 0m,n>0 шляхом індукції на mm :
При т = 1m=1 , так як в б п ∈ Qabn∈Q , ми повинні мати в б п = U Vabn=uv з U ∈ A , про ∈ Bu∈A,v∈B . Оскільки u ≠ ϵu≠ϵ , vv має бути b kbk для деякого k ≤ nk≤n . Але якщо k > 0k>0 , то оскільки b ∈ Ab∈A отримуємо b 1 + k ∈ Qb1+k∈Q , протиріччя. Такv = εv=ϵ , а б п ∈ . abn∈A
For the inductive step, since am+1bn∈Qam+1bn∈Q we have am+1bn=uvam+1bn=uv with u∈A,v∈Bu∈A,v∈B. Since again u≠ϵu≠ϵ, we have either v=akbnv=akbn for some 0<k<m+10<k<m+1, or v=bkv=bk for some k<nk<n. But in the former case, vv is already in AA by the induction hypothesis, so v2∈Qv2∈Q, contradiction. In the latter case, we must have k=0k=0 (i.e. v=ϵv=ϵ) since from b∈Ab∈A we get b1+k∈Qb1+k∈Q. So u=am+1bn∈Au=am+1bn∈A.
Now consider the general case of primitive words with rr alternations between aa and bb, i.e. ww is either am1bn1…amsbnsam1bn1…amsbns, bm1an1…bmsansbm1an1…bmsans (for r=2s−1r=2s−1), am1bn1…ams+1am1bn1…ams+1, or bm1an1…bms+1bm1an1…bms+1 (for r=2sr=2s); we can show that they are all in AA using induction on rr. What we did so far covered the base cases r=0r=0 and r=1r=1.
For r>1r>1, we use another induction on m1m1, which works very much the same way as the one for r=1r=1 above:
If m1=1, then w=uv with u∈A,v∈B, and since u≠ϵ, v has fewer than r alternations. So v (or its root in case v itself is not primitive) is in A by the induction hypothesis on r for a contradiction as above unless v=ϵ. So w=u∈A.
If m1>1, in any factorization w=uv with u≠ϵ, v either has fewer alternations (and its root is in A unless v=ϵ by the induction hypothesis on r), or a shorter first block (and its root is in A unless v=ϵ by the induction hypothesis on m1). In either case we get that we must have v=ϵ, i.e. w=u∈A.
The case of Q′:=Q∪{ϵ} is rather more complicated. The obvious things to note are that in any decomposition Q=A⋅B, both A and B must be subsets of Q′ with A∩B={ϵ}. Also, a,b must be contained in A∪B.
With a bit of extra work, one can show that a and b must be in the same subset. Otherwise, assume wlog that a∈A and b∈B. Let us say that w∈Q′ has a proper factorization if w=uv with u∈A∖{ϵ} and v∈B∖{ϵ}. We have two (symmetric) subcases depending on where ba goes (it must be in A or B since it has no proper factorization).
- If ba∈A, then aba has no proper factorization since ba,a∉B. Since aba∈A would imply abab∈A⋅B, we get aba∈B. As a consequence, bab is neither in A (which would imply bababa∈A⋅B) nor in B (which would imply abab∈A⋅B). Now consider the word babab. It has no proper factorization since bab∉A∪B and abab,baba are not primitive. If babab∈A, then since aba∈B we get (ba)4∈A⋅B; if babab∈B, then since a∈A we get (ab)3∈A⋅B. So there is no way to have babab∈A⋅B, contradiction.
- The case ba∈B is completely symmetric. In a nutshell: bab has no proper factorization and cannot be in B, so it must be in A; therefore aba cannot be in A or B; therefore ababa has no proper factorization but also cannot be in either A or B, contradiction.
I am currently not sure how to proceed beyond this point; it would be interesting to see if the above argument can be systematically generalized.