(a) Take any point
in the plane
. To reflect this point in the
-axis, you need to negate the
-component, so that

(b) This can be done easily by choosing

as

(c) Computing
(very easy since it's diagonal) returns the identity matrix
, which as a linear transformation returns whatever it is applied to. Geometrically, we're applying the same reflection twice, which returns the original vector to its starting value/position in the plane.
(d) If

then



Multiplying
by
on the left is equivalent to negating the first row of
, and on the right to negative the first column of
. The third product is equivalent to negating the antidiagonal.