🧮 Brain Teaser

No Function with f(f(n))=n+2025f(f(n)) = n + 2025

Problem

Prove that there is no function f:ZZf: \mathbb{Z} \to \mathbb{Z} satisfying f(f(n))=n+2025for all nZ.f(f(n)) = n + 2025 \quad \text{for all } n \in \mathbb{Z}.


Field

Putnam / Combinatorics / Number Theory

Why It's Beautiful

The problem looks like it's about functions and iteration — but the real obstruction is purely combinatorial: a fixed-point-free involution on a finite set requires the set to have even size. The number 2025 is odd, and that's the entire reason no such ff can exist.

The proof has a satisfying three-act structure:

  1. Show ff induces a well-defined map on Z/2025Z\mathbb{Z}/2025\mathbb{Z}.
  2. Show that map is a fixed-point-free involution.
  3. Derive a contradiction from the odd size of Z/2025Z\mathbb{Z}/2025\mathbb{Z}.

The same argument kills f(f(n))=n+cf(f(n)) = n + c for any odd cc.

Key Idea / Trick

Derive the periodicity f(n+2025)=f(n)+2025f(n + 2025) = f(n) + 2025, which lets ff descend to a map fˉ\bar{f} on Z/2025Z\mathbb{Z}/2025\mathbb{Z}. Then show fˉ2=id\bar{f}^2 = \mathrm{id} (involution) and fˉ\bar{f} has no fixed points. A fixed-point-free involution pairs elements into 2-element orbits, requiring an even-sized set — but Z/2025Z=2025|\mathbb{Z}/2025\mathbb{Z}| = 2025 is odd.

Difficulty

3 / 5

Tags

Putnam, Functions, Involution, Modular arithmetic, Parity, Fixed points, Combinatorics

FunctionsInvolutionModular arithmeticFixed pointsParity

No Function with f(f(n))=n+2025f(f(n)) = n + 2025 — Answer

Let c=2025c = 2025 for convenience. Suppose for contradiction that f:ZZf: \mathbb{Z} \to \mathbb{Z} satisfies f(f(n))=n+cf(f(n)) = n + c for all nn.


Step 1: Periodicity — f(n+c)=f(n)+cf(n + c) = f(n) + c

Apply ff to both sides of f(f(n))=n+cf(f(n)) = n + c: f(f(f(n)))=f(n+c)f(f(f(n))) = f(n + c)

But also, applying the functional equation to f(n)f(n) in place of nn: f(f(f(n)))=f(n)+cf(f(f(n))) = f(n) + c

Combining: f(n+c)=f(n)+cf(n + c) = f(n) + c for all nZn \in \mathbb{Z}.


Step 2: ff descends to a map fˉ\bar{f} on Z/cZ\mathbb{Z}/c\mathbb{Z}

Since f(n+c)=f(n)+cf(n+c) = f(n) + c, the residue f(n)modcf(n) \bmod c depends only on nmodcn \bmod c. So define: fˉ:Z/cZZ/cZ,fˉ(nmodc)=f(n)modc\bar{f}: \mathbb{Z}/c\mathbb{Z} \to \mathbb{Z}/c\mathbb{Z}, \qquad \bar{f}(n \bmod c) = f(n) \bmod c

This is well-defined. From f(f(n))=n+cn(modc)f(f(n)) = n + c \equiv n \pmod{c}, we get: fˉ(fˉ(n))=nin Z/cZ\bar{f}(\bar{f}(n)) = n \quad \text{in } \mathbb{Z}/c\mathbb{Z}

So fˉ\bar{f} is an involution (fˉ2=id\bar{f}^2 = \mathrm{id}), hence a bijection.


Step 3: fˉ\bar{f} has no fixed points

Suppose fˉ(k)=k\bar{f}(k) = k for some kZ/cZk \in \mathbb{Z}/c\mathbb{Z}, i.e., f(n)n(modc)f(n) \equiv n \pmod{c} for some nn with nkn \equiv k.

Then f(n)=n+mcf(n) = n + mc for some integer mm. Applying ff again using Step 1: f(f(n))=f(n+mc)=f(n)+mc=n+2mcf(f(n)) = f(n + mc) = f(n) + mc = n + 2mc

But f(f(n))=n+cf(f(n)) = n + c, so 2mc=c2mc = c, giving m=12m = \tfrac{1}{2} — not an integer. Contradiction.

So fˉ\bar{f} is a fixed-point-free involution on Z/cZ\mathbb{Z}/c\mathbb{Z}.


Step 4: Contradiction

A fixed-point-free involution partitions every element into a 2-element orbit {k,fˉ(k)}\{k,\, \bar{f}(k)\} (since fˉ(k)k\bar{f}(k) \neq k and fˉ(fˉ(k))=k\bar{f}(\bar{f}(k)) = k). So the orbits partition Z/cZ\mathbb{Z}/c\mathbb{Z} into pairs, requiring Z/cZ=c|\mathbb{Z}/c\mathbb{Z}| = c to be even.

But c=2025=452c = 2025 = 45^2 is odd. Contradiction. \blacksquare


Why 2025 specifically?

The proof works for any odd cc. For even cc, no contradiction arises — and indeed f(f(n))=n+2f(f(n)) = n + 2 has a solution: f(n)=n+1f(n) = n + 1.

The key is not any special property of 2025 beyond its being odd.


Summary of the argument chain

f(f(n))=n+c    f(n+c)=f(n)+c    fˉ is a fixed-point-free involution on Z/cZ    c is evenf(f(n)) = n + c \implies f(n+c) = f(n)+c \implies \bar{f} \text{ is a fixed-point-free involution on } \mathbb{Z}/c\mathbb{Z} \implies c \text{ is even}

Type: PutnamEdit on GitHub ↗