Why must the domain of sine be restricted before we can define arcsin, and which restriction is chosen?
Think about your answer, then reveal below.
Model answer: Sine is periodic and therefore not one-to-one — every horizontal line between −1 and 1 intersects the sine curve infinitely many times. An inverse function can only exist for a one-to-one (injective) function. We restrict sine to [−π/2, π/2], one arc from the bottom to the top of the curve, where it is strictly increasing and passes the horizontal line test. This interval is chosen by convention because it includes 0 and is centered at the origin.
The restriction is not arbitrary geometry — it is a logical necessity. Without restricting the domain, 'arcsin(1/2)' would have infinitely many valid answers (π/6, 5π/6, π/6 ± 2π, etc.), making it not a function. The standard restriction [−π/2, π/2] is simply the most natural and convenient single branch; arccos uses [0, π] and arctan uses (−π/2, π/2) for analogous reasons.