We first show a solution for each n∈{2,3,4}. We will later show the impossibility of finding such a solution for n⩾5. For n=2, take for example (a1,a2)=(1,3) and r=2. For n=3, take the root r>1 of x2−x−1=0 (the golden ratio) and set (a1,a2,a3)=(0,r,r+r2). Then
(a2−a1,a3−a2,a3−a1)=(r,r2,r+r2=r3)
For n=4, take the root r∈(1,2) of x3−x−1=0 (such a root exists because 13−1−1<0 and 23−2−1>0) and set (a1,a2,a3,a4)=(0,r,r+r2,r+r2+r3). Then
(a2−a1,a3−a2,a4−a3,a3−a1,a4−a2,a4−a1)=(r,r2,r3,r4,r5,r6)
For n⩾5, we will proceed by contradiction. Suppose there exist numbers a1<a2<⋯<an and r>1 satisfying the conditions of the problem. We start with a lemma:
Lemma. We have rn−1>2.
Proof. There are only n−1 differences aj−ai with j=i+1, so there exists an exponent e⩽n and a difference aj−ai with j⩾i+2 such that aj−ai=re. This implies
rn⩾re=aj−ai=(aj−aj−1)+(aj−1−ai)>r+r=2r,
thus rn−1>2 as desired.
To illustrate the general approach, we first briefly sketch the idea behind the argument in the special case n=5. In this case, we clearly have a5−a1=r10. Note that there are 3 ways to rewrite a5−a1 as a sum of two differences, namely
(a5−a4)+(a4−a1),(a5−a3)+(a3−a1),(a5−a2)+(a2−a1).
Using the lemma above and convexity of the function f(n)=rn, we argue that those three ways must be r10=r9+r1=r8+r4=r7+r6. That is, the "large" exponents keep dropping by 1, while the "small" exponents keep increasing by n−2,n−3,…,2. Comparing any two such equations, we then get a contradiction unless n⩽4.
Now we go back to the full proof for any n⩾5. Denote b=21n(n−1). Clearly, we have an−a1=rb. Consider the n−2 equations of the form:
an−a1=(an−ai)+(ai−a1) for i∈{2,…,n−1}
In each equation, one of the two terms on the right-hand side must be at least 21(an−a1). But from the lemma we have rb−(n−1)=rb/rn−1<21rb, so the other term must be at least 21(an−a1). Therefore, we can write
an−a1=rb=rαi+rβi
for some αi and βi with αi⩾b−(n−1) and βi⩽b−(n−1). Since r>1 and f(r)=rn is convex, we have
rb−1−rb−2>rb−2−rb−3>…>rb−(n−3)−rb−(n−2)
implying
rα2−rα1>rα3−rα2>…>rαn−2−rαn−3.
Convexity of f(r)=rn further implies
α2−α1>α3−α2>…>αn−2−αn−3.
Note that αn−2−αn−3⩾2: Otherwise we would have αn−2−αn−3=1 and thus
rαn−3⋅(r−1)=rαn−2−rαn−3=rb−(n−3)−rb−(n−2)=rb−(n−2)⋅(r−1),
implying that αn−3=b−(n−2), a contradiction. Therefore, we have
αn−2−α1=(αn−2−αn−3)+⋯+(α2−α1)⩾2+3+⋯+(n−2)=21(n−2)(n−1)−1=21n(n−3)
On the other hand, from αn−2⩽b−(n−1) and α1⩾1 we get
αn−2−α1⩽b−n=21n(n−1)−n=21n(n−3),
implying that equalities must occur everywhere and the claim about the small terms follows. Now, assuming n−2⩾2, we have the two different equations:
rb=rb−(n−2)+rb−(n−2)−1 and rb=rb−(n−3)+rb−(n−2)−3,
which can be rewritten as
rn−1=r+1 and rn+1=r4+1
Simple algebra now gives
r4+1=rn+1=rn−1⋅r2=r3+r2⟹(r−1)(r3−r−1)=0.
Since r=1, using Equation (1) we conclude r3=r+1=rn−1, thus n=4, which gives a contradiction.