Since trig functions are periodic (not one-to-one), we restrict their domains to create invertible versions: arcsin on [-pi/2, pi/2], arccos on [0, pi], arctan on (-pi/2, pi/2). These inverse trig functions answer the question "what angle has this trig value?" Understanding their restricted ranges is critical for getting correct answers in equations and for the derivative formulas in calculus.
Start with why restriction is necessary (horizontal line test fails on unrestricted trig). Graph each inverse function as a reflection of the restricted parent. Practice evaluating arcsin(1/2), arccos(-sqrt(2)/2), etc. by thinking about the unit circle within the restricted range.
From the unit circle, you know that sin(π/6) = 1/2. But if someone asks "what angle has a sine of 1/2?", there is a problem: infinitely many angles do. Every full rotation brings you back to the same y-coordinate, so π/6, π/6 + 2π, π/6 − 2π, and also 5π/6, 5π/6 + 2π, and so on all have sine equal to 1/2. You know from inverse functions that a function must be one-to-one (pass the horizontal line test) to have an inverse. Sine fails that test badly — every horizontal line between −1 and 1 hits the sine curve infinitely many times.
The solution is domain restriction: we agree to keep only a portion of the sine curve where it is one-to-one, then define the inverse on that restricted piece. For sine, we use [−π/2, π/2] — one arc from the bottom to the top. This gives arcsin (also written sin^−1), which takes a value in [−1, 1] and returns the unique angle in [−π/2, π/2] with that sine. So arcsin(1/2) = π/6, not 5π/6. Both are valid angles with sine 1/2, but only π/6 falls in the agreed-upon range. The restriction is a convention — we had to pick one branch, and these are the standard choices.
For arccos, the restriction is [0, π] — the upper semicircle of the unit circle. For arctan, the restriction is (−π/2, π/2) — one period of the tangent function (open because tangent has vertical asymptotes at the endpoints). These choices make each inverse function produce a single, predictable output. The notation sin^−1(x) means the arcsine, *not* 1/sin(x). If you mean the reciprocal, write csc(x). This distinction matters constantly.
The most common mistake is forgetting the range when composing a trig function with its inverse. arcsin(sin(θ)) = θ is *only* true when θ is already in [−π/2, π/2]. If θ = 5π/4, then sin(5π/4) = −√2/2, and arcsin(−√2/2) = −π/4, not 5π/4. The function "undo" only works within the restricted range. Think of it like taking a square root: √(x²) = |x|, not always x, because squaring destroys the sign information. Similarly, applying sin then arcsin "forgets" which branch you were on if you started outside the principal range. Keeping the restricted ranges firmly in mind — [−π/2, π/2] for arcsin, [0, π] for arccos, (−π/2, π/2) for arctan — prevents this entire family of errors.