Kvant Math Problem 938

Let the angular speed be $\dfrac{360^\circ}{n}$ per second.

Verified: no
Verdicts: SKIP + SKIP
Solve time: 1m22s
Source on kvant.digital

Problem

A radius of a circle centered at $O$ rotates uniformly, turning through an angle of $\dfrac{360^\circ}n$ in one second (where $n$ is a natural number greater than 3). Initially it occupied the position $OM_0$; after one second, the position $OM_1$; after a further 2 seconds, the position $OM_2$; after a further 3 seconds, the position $OM_3$; and so on; after a further $n-1$ seconds, the position $OM_{n-1}$.

  1. Prove that if $n$ is a power of 2, then the radii $OM_1$, $\ldots$, $OM_{n-1}$ divide the circle into $n$ equal sectors;
  2. Is this possible for other values of $n$?

V. V. Proizvolov

Tournament of Towns (Spring 1985)

Exploration

Let the angular speed be $\dfrac{360^\circ}{n}$ per second. The times at which the successive radii appear are

$$1,\quad 1+2,\quad 1+2+3,\quad \ldots,\quad 1+2+\cdots+k=\frac{k(k+1)}2.$$

Hence the radius $OM_k$ makes with the initial radius $OM_0$ the angle

$$\frac{360^\circ}{n}\cdot \frac{k(k+1)}2.$$

The problem asks when the set of residues

$$\frac{k(k+1)}2 \pmod n,\qquad k=0,1,\ldots,n-1,$$

contains every residue modulo $n$ exactly once. Indeed, equal sectors arise precisely when the $n$ radii occupy all $n$ directions

$$0,\frac{360^\circ}{n},\ldots,\frac{360^\circ(n-1)}n.$$

Testing small cases:

For $n=4$,

$$\frac{k(k+1)}2\equiv 0,1,3,2 \pmod 4,$$

all residues occur.

For $n=8$,

$$0,1,3,6,2,7,5,4,$$

again all residues occur.

For $n=5$,

$$0,1,3,1,0,$$

repetitions appear.

For $n=6$,

$$0,1,3,0,4,3,$$

repetitions again appear.

This suggests that powers of $2$ work and no other $n$ do.

The crucial point is to determine when the map

$$f(x)=\frac{x(x+1)}2 \pmod n$$

is a permutation of the residue classes modulo $n$.

Suppose

$$f(a)\equiv f(b)\pmod n.$$

Then

$$\frac{a(a+1)-b(b+1)}2 =\frac{(a-b)(a+b+1)}2.$$

For $n=2^m$, multiplication by $2$ is the only obstruction. One expects to prove injectivity by examining the exact power of $2$ dividing $a-b$.

For the converse, if $n$ has an odd divisor $d>1$, then

$$f(x+d)-f(x) =\frac{d(2x+d+1)}2.$$

Since $d$ is odd, choosing

$$2x+d+1\equiv 0\pmod{n/d}$$

should produce two distinct arguments with the same image modulo $n$. This would show that no $n$ having an odd factor can work.

Problem Understanding

We are given a rotating radius. After $1$ second it reaches $OM_1$, after a further $2$ seconds it reaches $OM_2$, and in general after a further $k$ seconds it reaches $OM_k$.

The direction of $OM_k$ is determined by the total elapsed time

$$\frac{k(k+1)}2.$$

The radii divide the circle into $n$ equal sectors exactly when the directions of

$$OM_0,OM_1,\ldots,OM_{n-1}$$

are the $n$ equally spaced directions of a regular $n$-division of the circle.

This is a Type B problem. We must prove the stated property for powers of $2$ and determine whether it can occur for any other $n$.

The core difficulty is to decide for which $n$ the residues

$$\frac{k(k+1)}2 \pmod n$$

form a complete residue system modulo $n$.

Proof Architecture

First, translate the geometric condition into an arithmetic one: the radii form $n$ equal sectors if and only if the residues $\frac{k(k+1)}2 \pmod n$ are all distinct.

Second, prove that for $n=2^m$ the map $x\mapsto \frac{x(x+1)}2$ is injective modulo $n$. The argument starts from

$

(a-b)(a+b+1)\equiv0\pmod{2^{m+1}}

$

and uses the exact power of $2$ dividing $a-b$.

Third, prove that injectivity implies the desired geometric conclusion, since there are exactly $n$ residues modulo $n$.

Fourth, prove the converse. If $n$ has an odd divisor $d>1$, write $n=dq$. Choose $x$ satisfying

$

2x+d+1\equiv0\pmod q

$.

Then

$

f(x+d)\equiv f(x)\pmod n

$,

so the map is not injective.

The hardest step is the injectivity proof for $n=2^m$, because one must control the $2$-adic divisibility of both factors carefully.

Solution

Let

$$f(k)=\frac{k(k+1)}2.$$

The total time elapsed when the radius reaches the position $OM_k$ is

$$1+2+\cdots+k=\frac{k(k+1)}2=f(k).$$

Since the radius turns through $\dfrac{360^\circ}{n}$ each second, the direction of $OM_k$ differs from that of $OM_0$ by

$$\frac{360^\circ}{n},f(k).$$

Thus two radii $OM_a$ and $OM_b$ coincide precisely when

$$f(a)\equiv f(b)\pmod n.$$

The radii

$$OM_0,OM_1,\ldots,OM_{n-1}$$

divide the circle into $n$ equal sectors if and only if their directions are all distinct. Since there are exactly $n$ residue classes modulo $n$, this is equivalent to the residues

$$f(0),f(1),\ldots,f(n-1)$$

forming a complete residue system modulo $n$.

We now determine for which $n$ this happens.

Assume first that

$$n=2^m.$$

Suppose

$$f(a)\equiv f(b)\pmod{2^m}.$$

Then

$$\frac{a(a+1)-b(b+1)}2\equiv0\pmod{2^m},$$

hence

$$(a-b)(a+b+1)\equiv0\pmod{2^{m+1}}.$$

Let

$$a-b=2^t u,$$

where $u$ is odd and $t\ge0$.

If $t<m$, then $a-b$ is divisible by $2^t$ but not by $2^{t+1}$. Since

$$(a+b+1)-(a-b)=2b+1,$$

and $2b+1$ is odd, the numbers $a+b+1$ and $a-b$ have opposite parity modulo $2^{t+1}$. Consequently $a+b+1$ is divisible by exactly $2^t$ and by no higher power of $2$.

Therefore the highest power of $2$ dividing

$$(a-b)(a+b+1)$$

is $2^{2t}$.

Because $t<m$, we have

$$2t\le 2m-2<m+1+t.$$

In particular,

$$2^{m+1}\nmid (a-b)(a+b+1),$$

contrary to the congruence above.

Hence $t\ge m$, which means

$$2^m\mid(a-b).$$

Thus

$$a\equiv b\pmod{2^m}.$$

The map $f$ is injective modulo $2^m$. Since there are exactly $2^m$ residue classes modulo $2^m$, it is a permutation of those residue classes. Therefore the directions of

$$OM_0,OM_1,\ldots,OM_{n-1}$$

are exactly the $n$ equally spaced directions, and the radii divide the circle into $n$ equal sectors.

Now suppose that $n$ is not a power of $2$. Then $n$ has an odd divisor $d>1$. Write

$$n=dq.$$

Because $d$ is odd, $2$ is invertible modulo $q$. Choose an integer $x$ satisfying

$$2x+d+1\equiv0\pmod q.$$

Then

$$f(x+d)-f(x) =\frac{(x+d)(x+d+1)-x(x+1)}2 =\frac{d(2x+d+1)}2.$$

The factor $2x+d+1$ is divisible by $q$, so

$$f(x+d)-f(x)$$

is divisible by

$$dq=n.$$

Hence

$$f(x+d)\equiv f(x)\pmod n.$$

Since $d>1$, the numbers $x$ and $x+d$ represent distinct residue classes modulo $n$. Thus $f$ is not injective modulo $n$.

Therefore the residues $f(0),\ldots,f(n-1)$ do not form a complete residue system modulo $n$, and the corresponding radii cannot divide the circle into $n$ equal sectors.

We have proved that the required division into $n$ equal sectors occurs precisely when $n$ is a power of $2$.

This completes the proof.

Verification of Key Steps

The first delicate point is the passage from geometry to arithmetic. The direction of $OM_k$ is determined by the angle

$$\frac{360^\circ}{n}f(k).$$

Two directions coincide exactly when their difference is an integer multiple of $360^\circ$, which is equivalent to

$$f(a)-f(b)\equiv0\pmod n.$$

No geometric information is lost in this translation.

The second delicate point is the injectivity proof for $n=2^m$. From

$$(a-b)(a+b+1)\equiv0\pmod{2^{m+1}}$$

one cannot conclude that either factor is divisible by $2^{m+1}$. The correct argument uses the exact power of $2$ dividing $a-b$. Writing

$$a-b=2^t u,$$

with $u$ odd, the identity

$$(a+b+1)-(a-b)=2b+1$$

shows that $a+b+1$ has exactly the same $2$-adic valuation $t$. Hence the product has valuation $2t$. If $t<m$, then $2t<m+1+t$, so the product cannot be divisible by $2^{m+1}$. This forces $t\ge m$.

The third delicate point is the construction for the converse. Since $n=dq$ with $d$ odd, the congruence

$$2x+d+1\equiv0\pmod q$$

always has a solution because $2$ is invertible modulo $q$. Substituting into

$$f(x+d)-f(x)=\frac{d(2x+d+1)}2$$

gives divisibility by $dq=n$, producing two distinct residue classes with the same image.

Alternative Approaches

A different approach uses permutation polynomials modulo powers of $2$. The map

$$f(x)=\frac{x(x+1)}2$$

is the classical triangular-number polynomial. One may prove directly that it induces a permutation of the residue classes modulo $2^m$ by studying its action on even and odd residues and proceeding recursively from modulus $2^m$ to modulus $2^{m+1}$. This yields a short proof of the first part.

For the converse, one may invoke the Chinese remainder theorem. If $n$ has an odd factor, write

$$n=2^m r,$$

with $r$ odd and $r>1$. The congruence

$$f(x+r)\equiv f(x)\pmod r$$

shows that $f$ is not injective modulo $r$. A permutation modulo $n$ would induce a permutation modulo every divisor of $n$, which is impossible. The direct construction given in the main solution avoids introducing this additional machinery.