This question is inspired by the Collatz-like "Binary Complement Sequences," as discussed here and here and here on seq-fan.
For a given exponent $e$, let $f$ be defined as $f(x) = -1-3x \pmod{2^e}$. Now calculate $f(0)$, $f^2(0) = f(f(0))$, $f^3(0)$, $f^4(0)$...
For example, when $e=3$, the sequence is $$0, 7, 2, 1, 4, 3, 6, 5, 0$$ (because e.g. $f^2(0)=f(7)=-1-21\pmod{8}=2$).
Prove that this sequence is always a cyclic permutation of the integers from 0 to $2^e$. That is, prove that $f^n(0) = 0$ iff $n \equiv 0 \pmod{2^e}$.
(I don't know that this is always true, but it seems to be true, and so I bet it's easy to prove for someone who knows math. :) If it's true, my next question will be whether it's possible to efficiently compute discrete logarithms in this group — like, "given $e=64$, find me the $x$ such that $f^x(0) = 42$." I know this is a well-studied hard problem for functions of the form $f(x)=px\pmod{2^e}$, but I'm not sure what effect our extra "minus 1" might have.)
(Credit for this answer really goes to Junekey Jeon, who had the key insight, but asked me to write it up.)
The trick is to see that iterating $f(x)=-1-3x\pmod{2^e}$ gives us exactly a linear congruential generator! Using the terminology from that Wikipedia page, we have a modulus of $m=2^e$, a multiplier of $a=2^e-3$, an increment of $c=2^e-1$, and a seed of $X_0=0$.
By the Hull–Dobell Theorem, an LCG with $c\neq 0$ will have period $m$ if and only if all three of these conditions hold:
So, all three conditions hold, and so this LCG does have period $m=2^e$.