IMO 1986 · Problem 3

To each vertex of a regular pentagon an integer is assigned in such a way that the sum of all five numbers is positive. If three consecutive vertices are assigned the numbers \(x, y, z\) respectively and \(y < 0\) then the following operation is allowed: the numbers \(x, y, z\) are replaced by \(x+y, -y, z+y\) respectively. Such an operation is performed repeatedly as long as at least one of the five numbers is negative. Determine whether this procedure necessarily comes to and end after a finite number of steps.

Official text of the problem. Official problem paper (PDF)

Answer

Yes. The procedure terminates after finitely many steps, regardless of which negative number is chosen at each step.

Solution

Let the numbers, in cyclic order, be \(a,b,c,d,e\), and let \[ S=a+b+c+d+e>0. \] Every allowed operation preserves \(S\), since \[ (x+y)+(-y)+(z+y)=x+y+z. \] The numbers also remain integers.

Consider the nonnegative integer \[ \Phi=(a-c)^2+(b-d)^2+(c-e)^2+(d-a)^2+(e-b)^2. \] We prove that every allowed operation strictly decreases \(\Phi\).

The definition of \(\Phi\) is unchanged by cyclic relabeling, so we may label the chosen negative number \(b\). The operation replaces \[ (a,b,c,d,e)\quad\text{by}\quad(a+b,-b,c+b,d,e). \] The new value of \(\Phi\) is therefore \[ \Phi'=(a-c)^2+(b+d)^2+(c+b-e)^2 +(d-a-b)^2+(e+b)^2. \] Subtracting and expanding gives \[ \begin{aligned} \Phi'-\Phi &=4bd+\bigl(2b(c-e)+b^2\bigr) +\bigl(-2b(d-a)+b^2\bigr)+4be\\ &=2b(a+b+c+d+e)\\ &=2bS. \end{aligned} \] Since \(b\) is a negative integer, \(b\le -1\). Thus every operation decreases \(\Phi\) by at least \(2S>0\).

If \(\Phi_0\) is the initial value, then after \(k\) operations, \[ 0\le \Phi_k\le \Phi_0-2Sk. \] Consequently, there can be at most \[ \left\lfloor\frac{\Phi_0}{2S}\right\rfloor \] operations. The procedure must therefore terminate. At termination no negative number can remain, because any negative number would permit another operation. This argument applies to every possible choice of operations.

The solution was checked against John Scholes' IMO solution archive (Kalva).