Kvant Math Problem 1580

Verified: no
Verdicts: UNKNOWN + UNKNOWN
Solve time: 12m23s
Source on kvant.digital

Problem

Is it possible, using several line segments and arcs, to cut a circle into pieces from which an equal-area square can be assembled?

A. Ya. Kanel-Belov

Tournament of Towns (Autumn 1996)

Exploration

The problem asks whether a circle can be dissected into finitely many pieces whose boundaries consist of line segments and circular arcs, and then reassembled into a square of the same area.

At first glance this resembles the classical circle-squaring problem by dissection. If arbitrary measurable pieces are allowed, area alone is enough for an equidecomposition by deep results, but here the pieces must arise from finitely many cuts made by segments and arcs. The boundary structure is highly restricted.

A natural invariant is the total curvature of boundaries. A circle has boundary consisting entirely of one arc with total turning $2\pi$. A square has boundary made of four segments and has no curved portions. When pieces are cut apart and reassembled, interior boundary portions cancel in pairs. Any additive quantity attached to circular arcs and invariant under rigid motions might survive.

Consider assigning to every positively oriented circular arc its angle, with sign according to orientation. A full circle contributes $2\pi$. A line segment contributes $0$. If two copies of the same cut edge are traversed in opposite directions, their contributions cancel. Hence the sum of these signed arc-angles over all piece boundaries should be preserved under cutting and reassembly.

For the original circle the value is $2\pi$. For a square it is $0$, because its boundary contains no circular arcs. This suggests impossibility.

The delicate point is proving rigorously that the signed sum of arc-angles is indeed additive and cancels on internal boundaries. One must formulate it so that every circular arc contributes its central angle with a sign determined by the orientation induced from the piece boundary.

Problem Understanding

We are given a circle. We may cut it into finitely many pieces using line segments and circular arcs. The pieces may then be moved by rigid motions and assembled into another figure. The question is whether they can be assembled into a square having the same area as the original circle.

This is a Type B problem, a pure proof.

The core difficulty is finding an invariant preserved by such dissections and reassemblies. Area is not an obstruction because the square is required to have the same area. The relevant invariant must distinguish boundaries containing circular arcs from boundaries consisting only of line segments.

Proof Architecture

Define for every piece an invariant $I$ equal to the sum of the signed central angles of all circular arcs occurring on its boundary.

Show that $I$ is unchanged under rigid motions, because central angles are preserved.

Show that $I$ is additive under gluing: when two boundary arcs are identified, they are traversed in opposite directions in the two pieces, hence their contributions cancel.

Compute $I$ for the original circle. Its boundary is one positively oriented full circle, so $I=2\pi$.

Compute $I$ for a square. Its boundary contains only line segments, so $I=0$.

Assume a dissection and reassembly into a square exists. Additivity gives preservation of $I$, yielding $2\pi=0$, a contradiction.

The most delicate lemma is the additivity under gluing, because one must verify the exact cancellation of contributions from every internal cut arc.

Solution

Orient the boundary of every piece positively, that is, so that the piece lies on the left as the boundary is traversed.

For a boundary component consisting of a circular arc, define its contribution to be the signed central angle of that arc. The sign is positive when the arc is traversed counterclockwise around its center and negative when traversed clockwise. A line segment contributes $0$.

For a piece $P$, let

$$I(P)=\sum \alpha_i,$$

where the sum runs over all circular arcs occurring on the boundary of $P$, and $\alpha_i$ denotes the corresponding signed central angle.

The quantity $I(P)$ is preserved by rigid motions, since rigid motions preserve circles, orientations of arcs on the boundary of the moved piece, and central angles.

We next prove additivity. Suppose two pieces $P_1$ and $P_2$ are glued along a common boundary arc. Along that arc, the positive boundary orientation of $P_1$ induces one direction of traversal, while the positive boundary orientation of $P_2$ induces the opposite direction, because each piece lies on the left of its own boundary. Hence the common arc contributes $\alpha$ to one piece and $-\alpha$ to the other. Their contributions cancel.

The same argument applies to every boundary portion that becomes internal after gluing. Therefore, when pieces are assembled into a figure $F$,

$$I(F)=\sum I(P_k),$$

because all contributions coming from internal cut curves cancel pairwise.

Now consider the initial circle $C$. Its positively oriented boundary is one complete counterclockwise circle. The corresponding signed central angle equals

$$2\pi.$$

Hence

$$I(C)=2\pi.$$

Suppose that the circle can be cut into finitely many pieces by segments and arcs and reassembled into a square $Q$.

By the additivity proved above,

$$I(Q)=I(C)=2\pi.$$

On the other hand, the boundary of a square consists entirely of line segments. Since line segments contribute $0$,

$$I(Q)=0.$$

Thus

$$2\pi=0,$$

which is impossible.

The assumed dissection and reassembly cannot exist. This completes the proof.

∎

Verification of Key Steps

The first delicate point is the cancellation on a common arc. Let an arc subtend angle $\theta$. When it appears as part of the boundary of one piece, the piece lies on the left. For the adjacent piece, the same geometric arc is traversed in the opposite direction, because that second piece must also lie on the left of its own boundary. Thus the signed contributions are $\theta$ and $-\theta$, whose sum is $0$. Forgetting the boundary orientation would destroy the cancellation argument.

The second delicate point is the computation for the original circle. The boundary is a single full circle traversed counterclockwise. A full revolution subtends angle $2\pi$, not $0$. The invariant records the total signed angular measure of circular boundary arcs, not merely the net change of tangent direction.

A third possible error is to think that introducing new circular arcs during cutting might alter the total invariant. Additivity shows exactly why this does not happen: every newly created cut arc appears twice in the collection of piece boundaries, once with each orientation, so its net contribution is zero.

Alternative Approaches

One may interpret the same invariant as the integral of curvature over the circular parts of the boundary. A circular arc of radius $r$ and central angle $\theta$ contributes

$$\int k,ds=\frac1r\cdot r\theta=\theta,$$

with sign determined by orientation. Internal boundaries cancel because they are traversed in opposite directions after gluing. The original circle has total contribution $2\pi$, whereas a polygon has contribution $0$ from its sides. This yields the same contradiction.

The arc-angle formulation is preferable because it avoids introducing curvature integrals and uses only elementary geometry of circles and orientations.