Inverse Trigonometry: Find the Angle, Choose the Branch

Why reversing a function needs a choice

A ratio answers an angle-to-number question. For example, $\sin(\pi/6)=1/2$. Reversing that statement is less straightforward: $\sin(5\pi/6)$ is also $1/2$, and adding full turns produces further answers. An inverse function must return one output for each allowed input. We therefore choose a part of the original function on which its values do not repeat.

For sine, the standard choice is $[-\pi/2,\pi/2]$. On this interval sine increases from $-1$ to $1$, passing through each value exactly once. Thus $\sin^{-1}(1/2)=\pi/6$, although $5\pi/6$ solves the equation $\sin\theta=1/2$. The first question asks for a principal value (मुख्य मान); the second, unless restricted, may have many solutions.

The notation needs care. In $\sin^{-1}x$, the superscript means the inverse sine function, also written $\arcsin x$. It does not mean $1/\sin x$, which is $\operatorname{cosec}x$. Similarly, $(\sin x)^2$ and $\sin(x^2)$ describe different operations.

Inverting also swaps input and output roles. Sine takes an angle in its selected interval and returns a number between $-1$ and $1$. Inverse sine takes that number and returns the selected angle. On a graph this exchange reflects the selected curve in the line $y=x$. The two axes now represent different roles; read their labels before interpreting a point.

Read sine and cosine backwards

Inverse sine and inverse cosine accept the same inputs, yet they choose different output intervals. This makes their negative-input values especially useful for testing your understanding.

On the sine interval, negative inputs give negative angles. On the cosine interval, negative inputs give angles in the second quadrant. Consequently, $\sin^{-1}(-1/2)=-\pi/6$, whereas $\cos^{-1}(-1/2)=2\pi/3$. Finding a familiar reference angle is only the first step; the principal range chooses the actual answer.

The graphs make the choice visible. The inverse sine graph rises from $(-1,-\pi/2)$ to $(1,\pi/2)$ and passes through the origin. The inverse cosine graph falls from $(-1,\pi)$ to $(1,0)$, passing through $(0,\pi/2)$. All four endpoint points are included. These are graphs of angles against ratios, rather than repeating waves.

Move the input in the explorer and predict the sign and size of the output before reading it. At an input of $1$, sine’s selected angle is $\pi/2$, but cosine’s is $0$. At an input outside $[-1,1]$, neither has a real output. A calculator error there is a domain issue, not a request for a larger angle. Keep exact values such as $\pi/3$ when the input is a standard ratio; use a decimal only when approximation is useful.

Two inverses with every real input

Tangent and cotangent can produce any real number, so both inverse functions have domain $\mathbb R$. Their principal output intervals, however, are different:

Inverse tangent is increasing and passes through $(0,0)$. As its input becomes very large positive or negative, its output approaches $\pi/2$ or $-\pi/2$ without reaching either. Those excluded endpoints are horizontal asymptotes of the inverse graph.

Inverse cotangent, using the school convention here, is decreasing. It passes through $(0,\pi/2)$, approaches $0$ for very large positive input and approaches $\pi$ for very large negative input. Neither endpoint belongs to its range. A negative cotangent input must therefore produce a second-quadrant angle, not a negative principal angle.

It is tempting to write $\cot^{-1}x=\tan^{-1}(1/x)$ for every nonzero $x$. That shortcut fails for negative inputs because the output intervals differ. For $x=-1$, the right-hand side gives $-\pi/4$, while $\cot^{-1}(-1)=3\pi/4$. A useful relation that respects the chosen branches is $$\cot^{-1}x=\frac\pi2-\tan^{-1}x\quad(x\in\mathbb R).$$ Both sides lie in $(0,\pi)$, and taking cotangent verifies the value. Before comparing two calculator outputs, make sure they use this same cotangent convention.

At $x=1$, both selected angles are $\pi/4$. At $x=0$, they separate: inverse tangent returns zero, while inverse cotangent returns $\pi/2$. Testing these two inputs is a quick way to distinguish the graphs.

Keep the gap in secant and cosecant

For real angles where they are defined, secant and cosecant cannot lie strictly between $-1$ and $1$. Their inverses therefore accept $x\leq-1$ or $x\geq1$, with a genuine gap between the two input pieces.

Read the missing output points as carefully as the interval endpoints. Secant is undefined at $\pi/2$; cosecant is undefined at $0$. Thus those angles cannot be inverse outputs. By contrast, $\sec^{-1}(1)=0$, $\sec^{-1}(-1)=\pi$, $\operatorname{cosec}^{-1}(1)=\pi/2$ and $\operatorname{cosec}^{-1}(-1)=-\pi/2$.

For a permitted input, invert the reciprocal value using the matching sine or cosine branch: $$\sec^{-1}x=\cos^{-1}(1/x),\qquad \operatorname{cosec}^{-1}x=\sin^{-1}(1/x).$$ These are relations between different inverse functions; they do not confuse an inverse with a reciprocal. For $x=-2$, the intermediate ratio is $-1/2$. The cosine branch selects $2\pi/3$, while the sine branch selects $-\pi/6$.

The graphs have two separated pieces. A line must not be drawn through inputs between $-1$ and $1$. When you move across that gap in the explorer, expect an undefined interval, not a connecting segment. Function graphs communicate their domain through these breaks as well as through a printed formula.

The gap is an input restriction, not a choice to hide inconvenient values. For example, no real angle can have secant $1/2$, because that would require cosine $2$. Test an impossible input against the original ratio before trying to assign an angle.

Why the symbols do not simply cancel

The order of two functions matters. If $x\in[-1,1]$, inverse sine first selects an angle whose sine is $x$. Therefore $\sin(\sin^{-1}x)=x$. In the opposite order, sine first loses information about the original turn and quadrant. Inverse sine then returns only the angle in its principal interval.

For example, $\sin(5\pi/6)=1/2$, so $$\sin^{-1}(\sin(5\pi/6))=\pi/6,$$ not $5\pi/6$. Evaluate the inside, then select a permitted outside output. You may use $\sin^{-1}(\sin\theta)=\theta$ directly only when $\theta\in[-\pi/2,\pi/2]$. Similarly, $\cos^{-1}(\cos\theta)=\theta$ directly on $[0,\pi]$, and $\tan^{-1}(\tan\theta)=\theta$ directly on $(-\pi/2,\pi/2)$.

A useful way to reason is to find a familiar angle with the same ratio and then test its range. To simplify $\cos^{-1}(\cos(7\pi/6))$, use cosine’s symmetry: the inner value is $-\sqrt3/2$. The chosen output is $5\pi/6$. Both angles give the same cosine, but only one lies in $[0,\pi]$.

Domain checks come before branch checks. The expression $\tan^{-1}(\tan(\pi/2))$ is undefined because its inner tangent is undefined. No inverse function repairs that missing input. Likewise, expressions involving square roots need the correct sign: $\sqrt{\cos^2\theta}=|\cos\theta|$, not automatically $\cos\theta$. A principal branch sometimes tells you that the sign is positive; establish that fact before removing an absolute value.

When an angle is unfamiliar, subtract complete turns first and use a reference-angle sketch. This preserves the inner ratio. Then compare your candidate output with the selected inverse interval; the sketch alone does not make that final choice.

Turn an inverse angle into useful ratios

An inverse expression can often be simplified without calculating a decimal angle. Let $\theta=\tan^{-1}x$. Then $\tan\theta=x$, and $\theta\in(-\pi/2,\pi/2)$. The branch tells you $\cos\theta>0$, even when $x<0$.

From $1+\tan^2\theta=\sec^2\theta$, choose the positive cosine value: $$\cos\theta=\frac1{\sqrt{1+x^2}},\qquad \sin\theta=\frac{x}{\sqrt{1+x^2}}.$$ Consequently, $\sin(\tan^{-1}x)=x/\sqrt{1+x^2}$ for every real $x$. The denominator is always positive. For $x<0$, the sine is negative, just as the selected angle requires.

A right triangle with opposite length $|x|$, adjacent length $1$ and hypotenuse $\sqrt{1+x^2}$ can provide the magnitudes. It cannot, by itself, display a negative side length or a second-quadrant angle. Use the principal interval or a unit-circle sketch to restore the signs. A triangle gives magnitudes; the chosen angle gives signs.

The same habit handles other compositions. If $\phi=\cos^{-1}x$, then $x\in[-1,1]$ and $\phi\in[0,\pi]$. Since sine is nonnegative there, $\sin\phi=\sqrt{1-x^2}$. Hence $\tan(\cos^{-1}x)=\sqrt{1-x^2}/x$, provided $x\neq0$. At $x=0$, the angle is $\pi/2$, where tangent is undefined. Write that exclusion alongside the formula, rather than treating it as a small afterthought. These substitutions connect inverse functions to identities and prepare you for their derivatives.

The algebraic formula also handles $x=0$: it gives sine zero and cosine one without requiring a nondegenerate reference triangle. This is another reason to use the branch and identity as the argument, with a triangle as supporting visual reasoning.