1

We need to solve this problem without using trigonometry.

KLMN square is inside ABCD square. Prove that midpoints of the segments AK, BL, CM, and DN are vertices of a square. KLMN can be situated anywhere inside ABCD. I think that we are going to use congruent triangles.

I tried to draw so that the intersection point of diagonals of ABCD and KLMN coincide, I also tried the opposite to consider the general case, because KLMN can take an arbitrary position.

So, I would like to see approaches, how to prove.

3 Answers3

1

Hints towards a solution. If you're stuck, explain what you've tried.

  • Show that the angles between the corresponding sides of the squares $KLMN$ and $ABCD$ are the same. Let the midpoints of the segments be $W,X,Y, Z$.
  • Show that $WX$ is the median of a triangle with edges $AB$ and $KL$, and the angle between them is the above angle.
    • As an example, see this problem for the situation when the angle between $AB$ and $KL$ is a right angle.
    • Hence, by congruency (no need for trigo), $WXYZ$ is a rhombus.
  • Show that $WX$ and $XY$ are perpendicular.
    • Think about what it means to shift $KL$, $KN$ to start at $X$, and do some angle chasing (no need for trigo).
    • Hence, $WXYZ$ is a square.
Calvin Lin
  • 68,864
  • The property is easy to verify when the two squares have parallel sides. Then one could try to show that this property is rotation invariant. – Piquito May 23 '23 at 09:57
1

According to the way I read the problem, $KLMN$ can be rotated any amount when it is placed inside $ABCD$. For example, $K$ could be near $C$, $L$ near $D$, and so forth so that the square formed by the midpoints of $AK$ and the other corresponding segments is inside $KLMN$ and small.

But the orientation of $KLMN$ must be the same as $ABCD$, for example if $ABCD$ is a list of vertices in counterclockwise order then $KLMN$ must also be in counterclockwise order. It is easy to come up with counterexamples if the squares are allowed to have opposite orientation.

Given $KLMN$ with the same orientation as $ABCD$, and quadrilateral $WXYZ$ with vertices at the midpoints of $AK$, $BL$, $CM$, and $DN$, you might show that translating $KLMN$ without moving $ABCD$ leads to the quadrilateral $WXYZ$ being translated half as much without any change in shape or size.

You could translate $KLMN$ to $K'L'M'N'$ so that $K'$ coincides with $A$. In most cases this would cause part of $KLMN$ (or maybe all of it) to fall outside $ABCD$. But now you can easily use congruent triangles to prove that the midpoints of $AK'$, $BL'$, and $DN'$ form two equal segments at right angles.

Translate back to the original $KLMN$ and now you know that $ZW$ and $WX$ are equal segments at right angles.

But there is nothing special about $K$ and $A$ that you can't do with $L$ and $B$, $M$ and $C$, or $N$ and $D$. The same method shows that $WX$ and $XY$ are equal segments at right angles, so are $XY$ and $YZ$, and so are $YZ$ and $ZW$.

Therefore all sides of $WXYZ$ are of equal length and all angles are right angles, so it is a square.

David K
  • 98,388
  • Thanks. This makes much more sense now. $\quad$ In fact, your "translate $K$ to $A$" is essentially what I do with my "Shift $AB$ and $KL$ to $W$". – Calvin Lin May 22 '23 at 23:03
-1

COMMENT.-Let $S$ and $s$ be the sides of both squares.

$(1)$ Inside square $ABCD$ take a point $N$.

$(2)$ Draw the circle $\Gamma$ centered at $N$ having radius $s$.

$(3)$ Take a point $K$ in $\Gamma$.

$(4)$ The line $NM$ perpendicular to side $NK$, cuts $\Gamma$ in $M$.

$(5)$ The two lines, parallel to $NM$ and $NK$ passing by $K$ and $M$ respectively interset at point $L$ which clearly is the fourth vertex of the small square $KLMN$.

$(6)$ What remains to finish is taking the midpoints of segments $AK,BL,CM$ and $DN$ and verify that them are the vertices of an square which is easy and straightforward.

enter image description here

Piquito
  • 29,594
  • What's up with the mindset of these downvoters? Bad conscience or imbecility? I want it to be the first. – Piquito May 23 '23 at 09:44