The minimum value of bn−b0 is 1.
For convenience, we call a sequence a0,a1,…,an satisfying the conditions of the problem a "k-good sequence".
First, we prove that the k-good sequence is unique. To do this, we strengthen the statement, proving simultaneously that the k-good sequence satisfies the following condition:
If (a,b)=1,a+b≥k+1 and 1≤a,b≤k, then there exists a unique positive integer i satisfying ai=a,ai+1=b. ...... (*)
We induct on k. When k=1, if n≥2, then 1≤1≤n−1, so by (1) we know 2≤ai≤1, a contradiction. Hence n=1, so this sequence can only be 1,1, and is therefore unique.
Next we prove (*). If (a,b)=1,a+b≥k+1 and a,b≤k, then since k=1 we know a=b=1. Also a0=a1=1, so (*) holds.
Suppose the statement holds for k=t−1 (t≥2). Then for k=t, if ai=t, then since a0=an=1=t, we have 1≤i≤n−1. By (3) we know ai−1,ai+1=t (otherwise they would not be coprime), that is, t cannot be adjacent to itself. So ai−1,ai+1<t. By (3) we know ai∣ai−1+ai+1, but 0<ai−1+ai+1<2t, so combined with ai=t we know ai−1+ai+1=t.
Now remove all entries equal to t from a0,…,an, forming a new sequence A0,…,AN. Since
t=1, we have A0=AN=1 and N>0. We now prove: A0∼AN is a (t−1)-good sequence.
Let f:[0,N]→[0,n] denote the original position of Ai in the sequence a0,…,an. Since the original a0,…,an satisfies conditions (1)(2), and all t's have been removed from it, A0,…,AN satisfies conditions (1)(2). It remains to prove (3).
Set 0≤i≤N−1. If f(i+1)=f(i)+1, then (Ai,Ai+1)=(af(i),af(i)+1)=1. Otherwise, between af(i) and af(i+1) a t has been removed. Since t's are not adjacent, only one t is removed there, that is, af(i)=Ai,af(i)+1=t,af(i)+2=Ai+1.
By what was proved above, Ai+Ai+1=t=af(i)+1, and since a satisfies condition (3), (Ai,Ai+1)=(Ai,Ai+Ai+1)=(af(i),af(i)+1)=1. In summary, in either case (Ai,Ai+1)=1, which also means (Ai−1,Ai)=(Ai,Ai+1)=1∀1≤i≤N−1.
Next we prove Ai∣Ai−1+Ai+1∀1≤i≤N−1. Since a satisfies condition (3), it suffices to prove af(i)−1≡Ai−1modAi,af(i)+1≡Ai+1modAi. If f(i+1)=f(i)+1, this is clear. Otherwise, as discussed before, af(i)=Ai,af(i)+2=Ai+1 and Ai+Ai+1=af(i)+1, so af(i)+1=Ai+Ai+1≡Ai+1modAi.
So in either case, af(i)+1≡Ai+1modAi, and similarly af(i)−1≡Ai−1modAi, combined with the fact that a satisfies condition (3), we know Ai∣Ai−1+Ai+1. Thus, we have proved that A satisfies condition (3), and therefore A is a (t−1)-good sequence. By the induction hypothesis, A is unique and satisfies (*). Next we prove a is unique. Note that given the choice of A, it suffices to prove that the way of inserting the ϕ(t) copies of t into A0,…,AN is unique. However, by what was proved above, two t's cannot appear simultaneously between some Ai and Ai+1, and if t is between Ai,Ai+1, then Ai+Ai+1=t. Also (Ai,Ai+1)=1 and Ai,Ai+1≤t, combined with (*) we know that if t can be inserted both between Ai,Ai+1 and between Aj,Aj+1 (i=j), then Ai=Aj (otherwise Ai+1=Aj+1, contradicting the uniqueness in (*)). Also (Ai,t)=1 and Ai≤t, so Ai can take at most ϕ(t) values, that is, there are at most ϕ(t) positions where t can be inserted. Therefore the way of inserting t is unique, that is, a is unique. Also note that if Ai+Ai+1=t, then a t must be inserted between Ai,Ai+1 (otherwise there would not be enough positions).
Next we prove that a0,…,an satisfies condition (*). Let (a,b)=1,a+b≥t+1 and a,b≤t. If a,b=t, then by the fact that A satisfies condition (*) and a+b=t, it is easy to see that there exists a unique positive integer i satisfying ai=a,ai+1=b. If a=t, then since (t−b,b)=(t,b)=1,(t−b)+b=t≥t and t−b,b≤t−1 (note that t's are not adjacent, so b=t), combined with the fact that A satisfies condition (*), we know there exists a positive integer i satisfying Ai=t−b,Ai+1=b. Since Ai+Ai+1=t, by what was proved above, af(i)+1=t,af(i)+2=Ai+1, and the existence part of (*) is proved. It remains only to prove uniqueness. If ai=t,ai+1=b, then ai−1=t−b and there exists a unique I satisfying f(I)=i−1,f(I+1)=i+1. Hence AI=t−b,AI+1=b. Also since A satisfies (*), there exists a unique I satisfying AI=t−b,AI+1=b, from which we know i is also unique. Uniqueness is proved. Hence a0,…,an also satisfies (*).
By mathematical induction, the k-good sequence is unique and satisfies (*).
Let the numerators in order be b0,…,bn, and the denominators in order be c0,…,cn. For convenience, we call b,c respectively the k-numerator sequence and k-denominator sequence. We now prove: c0,…,cn is a k-good sequence and
cibi−1−ci−1bi=1∀1≤i≤n. This, combined with the uniqueness of the k-good sequence, shows that the numerators b0,…,bn satisfy the conditions of the problem. Since b0=0,bn=1, we get that the minimum value of bn−b0 is 1.
It is easy to see that c0,…,cn satisfies conditions (1)(2). Next, we use mathematical induction on k to prove that c also satisfies (3) and cibi−1−ci−1bi=1∀1≤i≤n.
First, when k=1 this is obviously true. If it holds for k=t−1, then for k=t, let B0,…,BN and C0,…,CN be respectively the (t−1)-numerator sequence and denominator sequence. Consider the fraction tx in lowest terms in [0,1], and suppose it falls in the interval (CiBi,Ci+1Bi+1). By the induction hypothesis, it is easy to see that it suffices to prove (t,Ci)=(t,Ci+1)=1,Ci≡t(modCi+1),Ci+1≡t(modCi),t∤Ci+Ci+1,xCi−Bit=Bi+1t−xCi+1=1. ... (Δ) Since (x,t)=1, let q<t and qx≡1(modt),p=⌊tqx⌋. Then qp<tx. By the definition of B, C and q<t, we know qp≤CiBi. Also xq−pt=qx−t⌊tqx⌋=qx divided by t has remainder 1. Similarly, xCi−Bit=Cix divided by t has remainder (call it r) (note that here we use the fact that CiBi is the closest fraction to tx). Since rqx≡r≡Cix(modt), so Ci≡rq(modt). Also Ci≤t and rq>0, so Ci≤rq. From qp≤CiBi we know tx−CiBi≤tx−qp. However tx−qp=qtxq−pt=qt1 and tx−CiBi=CitxCi−Bit≥rqtr (since Ci≤rq) =qt1=tx−qp, so equality must hold, that is, Bi=p,Ci=q. Hence (t,Ci)=(t,q)=1,xCi≡1(modt) and xCi−Bit=xq−pt=1. Similarly we can prove (t,Ci+1)=1,xCi+1≡−1(modt) and Bi+1t−xCi+1=1. Hence x(Ci+Ci+1)≡0(modt), that is, t∤Ci+Ci+1. (Δ) is fully proved.
In summary, by mathematical induction we know c=a and cibi−1−ci−1bi=1∀1≤i≤n, and the proof of the whole problem is complete.