Kvant Math Problem 1126

Verified: no
Verdicts: SKIP + SKIP
Solve time: 8m51s
Source on kvant.digital

Problem

In the trapezoid $ABCD$ (with bases $BC$ and $AD$), points $K$ and $M$ are chosen on the sides $AB$ and $CD$. Prove that if $\angle{BAM}=\angle{CDK}$, then $\angle{BMA}=\angle{CKD}$.

A. S. Merkuriev

Leningrad City Mathematical Olympiad (1988)

Exploration

The statement resembles a converse of a familiar fact about equal angles subtending the same segment. The desired conclusion says that the triangles determined by the pairs of segments $AM,AB$ and $DK,DC$ have not only one equal angle, namely

$$\angle BAM=\angle CDK,$$

but also the second angle

$$\angle BMA=\angle CKD.$$

Thus the goal is to prove that the triangles $ABM$ and $DCK$ are similar.

Since $ABCD$ is a trapezoid with bases $AD$ and $BC$, the nonparallel sides are $AB$ and $CD$. The condition already involves exactly these sides. This suggests expressing the given angles through the angles that the lines $AM$ and $DK$ make with the base $AD$.

Let

$$\alpha=\angle BAM=\angle CDK.$$

Because $AD\parallel BC$, the angles made by $AB$ and $CD$ with the base are naturally related to the angles of the trapezoid. Write

$$\beta=\angle DAB,\qquad \gamma=\angle CDA.$$

Then

$$\angle MAD=\beta-\alpha,$$

since $AM$ lies inside the angle between $AB$ and $AD$.

At $D$,

$$\angle KDA=\gamma-\alpha,$$

because $DK$ lies inside the angle between $DC$ and $DA$.

A useful observation is

$$\beta+\gamma=180^\circ,$$

since the interior angles on the same side of transversal $AD$ between the parallel lines $AB$ and $CD$ sum to $180^\circ$.

Hence

$$(\beta-\alpha)+(\gamma-\alpha)=180^\circ-2\alpha.$$

Now consider the quadrilateral formed by the lines $AM$, $DK$, and the side lines $AB$, $CD$. The relation above suggests that $AM$ and $DK$ may be symmetric with respect to the angle structure of the trapezoid. Computing the remaining angles of triangles $ABM$ and $DCK$ seems promising.

Indeed,

$$\angle ABM=\angle ABC=\beta',$$

and because $AD\parallel BC$,

$$\angle ABC=180^\circ-\beta.$$

Similarly,

$$\angle DCK=\angle DCB=180^\circ-\gamma.$$

Since $\beta+\gamma=180^\circ$,

$$180^\circ-\beta=\gamma,\qquad 180^\circ-\gamma=\beta.$$

Thus

$$\angle ABM=\gamma,\qquad \angle DCK=\beta.$$

Together with $\angle BAM=\alpha$ and $\angle CDK=\alpha$, the angle sums in the two triangles give

$$\angle BMA=180^\circ-\alpha-\gamma,$$

and

$$\angle CKD=180^\circ-\alpha-\beta.$$

Using $\beta+\gamma=180^\circ$,

$$180^\circ-\alpha-\gamma=\beta-\alpha,$$

and

$$180^\circ-\alpha-\beta=\gamma-\alpha.$$

These are not immediately equal. The comparison should instead be made through similarity. In triangles $ABM$ and $DCK$ we have

$$\angle BAM=\angle CDK=\alpha,$$

and

$$\angle ABM=\gamma=\angle CDK+\ ?$$

This does not match.

A better calculation is

$$\angle DCK=\angle DCB=180^\circ-\gamma=\beta.$$

Also

$$\angle MAB=\alpha.$$

Since $\beta+\gamma=180^\circ$, we get

$$\angle ABM=\gamma,\qquad \angle DCK=\beta.$$

Hence

$$\angle ABM+\angle DCK=180^\circ.$$

This does not directly yield similarity.

The crucial point is to compare the angles at $M$ and $K$ using the base $AD$:

$$\angle BMA=\angle(AB,AM),$$

which equals the difference between the angles that $AB$ and $AM$ make with $AD$,

$$\angle BMA=\beta-(\beta-\alpha)=\alpha$$

only if the lines were on the same side, which is incorrect. Careful oriented-angle computation is needed.

Let the direction of $AD$ be $0^\circ$. Then $AB$ has direction $\beta$, while $CD$ has direction $180^\circ-\gamma=\beta$. Thus $AB$ and $CD$ are parallel in direction only when extended, a useful coordinate insight.

The cleanest route is analytic. Put

$$A=(0,0),\quad D=(1,0),$$

and let the side lines $AB$ and $DC$ meet at $P$. Since $AD\parallel BC$, the trapezoid is the section of the angle at $P$ cut by the parallel lines $AD$ and $BC$.

The condition $\angle BAM=\angle CDK$ means that the line $AM$ and the line $DK$ cut equal angles from the sides of the same angle $APD$. In an angle, points on opposite sides connected by cevians making equal deviations from the sides are related by a spiral similarity centered at $P$. Then triangles $PAM$ and $PDK$ are similar, yielding

$$\angle PMA=\angle PKD.$$

Since $B,K,P$ are collinear and $C,M,P$ are collinear, this becomes exactly

$$\angle BMA=\angle CKD.$$

This appears to be the intended synthetic idea.

The delicate step is proving the similarity of $PAM$ and $PDK$.

Problem Understanding

We are given a trapezoid $ABCD$ with bases $AD$ and $BC$. Points $K$ and $M$ lie on the nonparallel sides $AB$ and $CD$, respectively. The condition

$$\angle BAM=\angle CDK$$

is assumed.

We must prove that

$$\angle BMA=\angle CKD.$$

This is a Type B problem, a pure proof problem.

The core difficulty is to exploit the trapezoid structure. Extending the nonparallel sides until they meet converts the trapezoid into a configuration inside an angle. The given equality then becomes a relation between cevians in that angle, and the desired conclusion should follow from a similarity argument.

Proof Architecture

Let $P=AB\cap CD$.

Lemma 1. The points $A,K,B,P$ are collinear and the points $D,M,C,P$ are collinear; this follows from the definitions of $K$, $M$, and $P$.

Lemma 2. In the angle $\angle APD$, the triangles $PAM$ and $PDK$ are similar; both have the angle at $P$ equal to $\angle APD$, and the given condition translates into $\angle PAM=\angle PDK$.

Lemma 3. From the similarity of $PAM$ and $PDK$ we obtain $\angle PMA=\angle PKD$; corresponding angles in similar triangles are equal.

The hardest step is Lemma 2, because the given angle equality must be translated correctly into the language of the angle with vertex $P$.

Solution

Let

$$P=AB\cap CD.$$

Since $ABCD$ is a trapezoid with bases $AD$ and $BC$, the sides $AB$ and $CD$ are not parallel, so they meet at a point $P$.

Because $K\in AB$ and $M\in CD$, we have

$$A,K,B,P \text{ are collinear}, \qquad D,M,C,P \text{ are collinear}.$$

Consider the triangles $PAM$ and $PDK$.

Since $P,A,B$ are collinear and $P,D,C$ are collinear, the given equality

$$\angle BAM=\angle CDK$$

can be rewritten as

$$\angle PAM=\angle PDK.$$

Also, the angle at $P$ in both triangles is the same angle formed by the lines $AB$ and $CD$:

$$\angle APM=\angle APD,$$

because $M$ lies on $PD$,

and

$$\angle DPK=\angle DPA,$$

because $K$ lies on $PA$.

Hence

$$\angle APM=\angle DPK.$$

Therefore the triangles $PAM$ and $PDK$ have two equal angles:

$$\angle PAM=\angle PDK, \qquad \angle APM=\angle DPK.$$

The triangles are similar.

Corresponding angles of similar triangles are equal, so

$$\angle PMA=\angle PKD.$$

Finally, $P,M,C,D$ are collinear, hence the rays $MP$ and $MC$ coincide. Likewise, $P,K,A,B$ are collinear, hence the rays $KP$ and $KB$ coincide. Therefore

$$\angle PMA=\angle BMA, \qquad \angle PKD=\angle CKD.$$

Substituting these equalities into the previous relation yields

$$\angle BMA=\angle CKD.$$

This completes the proof.

∎

Verification of Key Steps

The first delicate step is the translation of the given condition. Since $P,A,B$ are collinear, replacing the side $AB$ by the ray $AP$ does not change the angle with the line $AM$:

$$\angle BAM=\angle PAM.$$

Similarly, since $P,D,C$ are collinear,

$$\angle CDK=\angle PDK.$$

Thus the hypothesis becomes exactly

$$\angle PAM=\angle PDK.$$

The second delicate step is the equality of the angles at $P$. Because $M$ lies on the line $PD$,

$$\angle APM=\angle APD.$$

Because $K$ lies on the line $PA$,

$$\angle DPK=\angle DPA.$$

Both are the angle formed by the lines $PA$ and $PD$, hence they are equal.

The final delicate step is replacing $PM$ by $MC$ and $PK$ by $KB$. Since $P,M,C,D$ are collinear, the rays $MP$ and $MC$ lie on the same line, so

$$\angle PMA=\angle CMA=\angle BMA.$$

Likewise $P,K,A,B$ are collinear, giving

$$\angle PKD=\angle BKD=\angle CKD.$$

No orientation issue arises because the angles involved are formed by the same geometric lines.

Alternative Approaches

A projective-geometric approach is possible. Extend $AB$ and $CD$ to meet at $P$. The condition $\angle BAM=\angle CDK$ may be interpreted as saying that the cevians $AM$ and $DK$ are isogonal in the angle $\angle APD$. The corresponding pairs of rays cut equal angles from the sides of that angle. Properties of isogonal cevians then imply that the angles they form with the opposite sides are equal, which is exactly

$$\angle BMA=\angle CKD.$$

The similarity proof is preferable because it uses only elementary geometry. After introducing the intersection point $P$, the statement becomes an immediate consequence of the similarity of two triangles determined by the equal-angle condition.