Skip to content

Reciprocal and inverse trigonometric functions

Reciprocal and inverse trigonometric functions are different ideas. This distinction is essential:

secx=1cosxbutcos1x=arccosx.\boxed{\sec x=\frac1{\cos x}} \qquad\text{but}\qquad \boxed{\cos^{-1}x=\arccos x}.

The first takes the reciprocal of a trigonometric value. The second returns an angle whose cosine is xx. In particular,

cos1x1cosx.\cos^{-1}x\ne \frac1{\cos x}.

You should be able to:

The three reciprocal functions are defined by

secx=1cosx,cscx=1sinx,cotx=1tanx=cosxsinx.\boxed{\sec x=\frac1{\cos x}}, \qquad \boxed{\csc x=\frac1{\sin x}}, \qquad \boxed{\cot x=\frac1{\tan x}=\frac{\cos x}{\sin x}}.

The notation cosecx\operatorname{cosec}x is also used for cscx\csc x. Calculator buttons for these functions are not needed: evaluate the corresponding sine, cosine or tangent, then take the reciprocal.

Reciprocating an exact value gives an exact reciprocal value. Rationalise denominators where appropriate.

Worked example 1: evaluate reciprocal functions exactly

Section titled “Worked example 1: evaluate reciprocal functions exactly”

Find sec60\sec60^\circ, csc45\csc45^\circ and cot(5π/6)\cot(5\pi/6).

Since cos60=1/2\cos60^\circ=1/2,

sec60=11/2=2.\sec60^\circ=\frac1{1/2}=2.

Since sin45=2/2\sin45^\circ=\sqrt2/2,

csc45=12/2=22=2.\csc45^\circ =\frac1{\sqrt2/2} =\frac2{\sqrt2} =\sqrt2.

The angle 5π/65\pi/6 is in quadrant II with reference angle π/6\pi/6, so

tan(5π6)=13.\tan\left(\frac{5\pi}{6}\right)=-\frac1{\sqrt3}.

Therefore

cot(5π6)=3.\cot\left(\frac{5\pi}{6}\right)=-\sqrt3.

Hence

sec60=2,csc45=2,cot(5π/6)=3.\boxed{\sec60^\circ=2,\quad \csc45^\circ=\sqrt2,\quad \cot(5\pi/6)=-\sqrt3}.

A reciprocal is undefined when its denominator is zero:

functionundefined whenradian valuessecxcosx=0x=π2+nπcscxsinx=0x=nπcotxsinx=0x=nπnZ.\begin{array}{c|c|c} \text{function} & \text{undefined when} & \text{radian values} \\ \hline \sec x & \cos x=0 & x=\frac\pi2+n\pi \\ \csc x & \sin x=0 & x=n\pi \\ \cot x & \sin x=0 & x=n\pi \end{array} \qquad n\in\mathbb Z.

For example, sec(π/2)\sec(\pi/2) is undefined because it would require division by zero. It is not equal to zero or infinity.

Graphs of secx\sec x, cscx\csc x and cotx\cot x

Section titled “Graphs of sec⁡x\sec xsecx, csc⁡x\csc xcscx and cot⁡x\cot xcotx”

The reciprocal graphs follow directly from the familiar trigonometric graphs.

For any non-zero number uu:

  • if u=1u=1 or u=1u=-1, then 1/u=u1/u=u;
  • if 0<u<10<|u|<1, then 1/u>1|1/u|>1;
  • as uu approaches 00, the magnitude of 1/u1/u increases without bound;
  • uu and 1/u1/u always have the same sign.

Since 1sinx1-1\leq\sin x\leq1 and 1cosx1-1\leq\cos x\leq1, reciprocal values can never lie strictly between 1-1 and 11. Thus

range of secx=(,1][1,)\boxed{\text{range of }\sec x=(-\infty,-1]\cup[1,\infty)}

and

range of cscx=(,1][1,).\boxed{\text{range of }\csc x=(-\infty,-1]\cup[1,\infty)}.

The zeros of cosine become vertical asymptotes of secant. The zeros of sine become vertical asymptotes of cosecant. Neither secx\sec x nor cscx\csc x has a zero, because a fraction with numerator 11 cannot equal zero.

Key graph facts are:

FunctionPeriodVertical asymptotesRange
secx\sec x2π2\pix=π/2+nπx=\pi/2+n\piy1y\leq-1 or y1y\geq1
cscx\csc x2π2\pix=nπx=n\piy1y\leq-1 or y1y\geq1
cotx\cot xπ\pix=nπx=n\piall real numbers

Between consecutive asymptotes, cotx\cot x decreases from positive values to negative values and crosses the xx axis where cosx=0\cos x=0, namely at x=π/2+nπx=\pi/2+n\pi.

Worked example 2: features of a transformed reciprocal graph

Section titled “Worked example 2: features of a transformed reciprocal graph”

State the period, vertical asymptotes and range of

y=2sec(3x)1.y=2\sec(3x)-1.

Secant has period 2π2\pi. Replacing xx by 3x3x divides the period by 33:

period=2π3.\boxed{\text{period}=\frac{2\pi}{3}}.

Asymptotes occur where cos(3x)=0\cos(3x)=0:

3x=π2+nπ,3x=\frac\pi2+n\pi,

so

x=π6+nπ3,nZ.\boxed{x=\frac\pi6+\frac{n\pi}{3}},\qquad n\in\mathbb Z.

For u=sec(3x)u=\sec(3x), either u1u\leq-1 or u1u\geq1. Multiplying by 22 and subtracting 11 gives

2u13or2u11.2u-1\leq-3 \qquad\text{or}\qquad 2u-1\geq1.

Therefore

range=(,3][1,).\boxed{\text{range}=(-\infty,-3]\cup[1,\infty)}.
  1. Evaluate sec(2π/3)\sec(2\pi/3), csc(3π/2)\csc(3\pi/2) and cot(π/4)\cot(\pi/4) exactly.
  2. Where is csc(2x)\csc(2x) undefined?
  3. State the period and range of y=3cscx+2y=3\csc x+2.
  4. Explain why the equation secx=1/2\sec x=1/2 has no real solutions.
Answers
  1. sec(2π/3)=2\sec(2\pi/3)=-2, csc(3π/2)=1\csc(3\pi/2)=-1 and cot(π/4)=1\cot(\pi/4)=1.
  2. It is undefined when sin(2x)=0\sin(2x)=0, so 2x=nπ2x=n\pi and x=nπ/2x=n\pi/2, where nZn\in\mathbb Z.
  3. The period is 2π2\pi. Since cscx1\csc x\leq-1 or cscx1\csc x\geq1, the range is y1y\leq-1 or y5y\geq5.
  4. Every real secant value satisfies secx1|\sec x|\geq1, so 1/21/2 is outside its range.

Dividing the identity sin2x+cos2x1\sin^2x+\cos^2x\equiv1 by cos2x\cos^2x gives

tan2x+1sec2x.\tan^2x+1\equiv\sec^2x.

Dividing instead by sin2x\sin^2x gives

1+cot2xcsc2x.1+\cot^2x\equiv\csc^2x.

Thus the three Pythagorean identities are

sin2x+cos2x1,\boxed{\sin^2x+\cos^2x\equiv1}, 1+tan2xsec2x,1+cot2xcsc2x.\boxed{1+\tan^2x\equiv\sec^2x}, \qquad \boxed{1+\cot^2x\equiv\csc^2x}.

The notation sec2x\sec^2x means (secx)2(\sec x)^2. It does not mean sec(secx)\sec(\sec x).

Worked example 3: find reciprocal values from one ratio

Section titled “Worked example 3: find reciprocal values from one ratio”

Given that tanθ=3/4\tan\theta=-3/4 and θ\theta lies in quadrant II, find secθ\sec\theta exactly.

Use

1+tan2θ=sec2θ.1+\tan^2\theta=\sec^2\theta.

Then

sec2θ=1+(34)2=1+916=2516.\sec^2\theta =1+\left(-\frac34\right)^2 =1+\frac9{16} =\frac{25}{16}.

Therefore

secθ=±54.\sec\theta=\pm\frac54.

In quadrant II, cosine is negative, so its reciprocal secant is also negative:

secθ=54.\boxed{\sec\theta=-\frac54}.

The identity determines the magnitude but not the sign. The quadrant supplies the missing information.

An inverse trigonometric function reverses a trigonometric function:

sin1x=arcsinx,cos1x=arccosx,tan1x=arctanx.\sin^{-1}x=\arcsin x, \qquad \cos^{-1}x=\arccos x, \qquad \tan^{-1}x=\arctan x.

For example,

arcsin(12)=30\arcsin\left(\frac12\right)=30^\circ

because sin30=1/2\sin30^\circ=1/2.

However, sine and cosine repeat every 360360^\circ, and tangent repeats every 180180^\circ. There are infinitely many angles with the same trigonometric value. A function must produce exactly one output, so each original function is restricted before it is inverted.

The standard restrictions are:

Inverse functionInput domainPrincipal value range in radiansPrincipal value range in degrees
y=arcsinxy=\arcsin x1x1-1\leq x\leq1π/2yπ/2-\pi/2\leq y\leq\pi/290y90-90^\circ\leq y\leq90^\circ
y=arccosxy=\arccos x1x1-1\leq x\leq10yπ0\leq y\leq\pi0y1800^\circ\leq y\leq180^\circ
y=arctanxy=\arctan xall real xxπ/2<y<π/2-\pi/2<y<\pi/290<y<90-90^\circ<y<90^\circ

The endpoints ±π/2\pm\pi/2 are excluded from the range of arctangent because tangent is undefined there.

These ranges explain calculator outputs. Although sin150=1/2\sin150^\circ=1/2,

arcsin(1/2)=30,\arcsin(1/2)=30^\circ,

not 150150^\circ, because 150150^\circ is outside the principal range of arcsine.

The graph of an inverse is the reflection of the restricted original graph in the line y=xy=x. Consequently, the original range becomes the inverse domain, and the original restricted domain becomes the inverse range.

Worked example 4: find principal values exactly

Section titled “Worked example 4: find principal values exactly”

Evaluate

arcsin(32),arccos(22),arctan(1)\arcsin\left(-\frac{\sqrt3}{2}\right), \qquad \arccos\left(-\frac{\sqrt2}{2}\right), \qquad \arctan(-1)

in radians.

The reference angles are π/3\pi/3, π/4\pi/4 and π/4\pi/4 respectively.

Arcsine outputs an angle in [π/2,π/2][-\pi/2,\pi/2]. The required sine is negative, so

arcsin(32)=π3.\arcsin\left(-\frac{\sqrt3}{2}\right)=-\frac\pi3.

Arccosine outputs an angle in [0,π][0,\pi]. Cosine is negative in the relevant part of quadrant II, so

arccos(22)=3π4.\arccos\left(-\frac{\sqrt2}{2}\right)=\frac{3\pi}{4}.

Arctangent outputs an angle in (π/2,π/2)(-\pi/2,\pi/2). The required tangent is negative, so

arctan(1)=π4.\arctan(-1)=-\frac\pi4.

Therefore

π3,3π4,π4.\boxed{-\frac\pi3,\quad\frac{3\pi}{4},\quad-\frac\pi4}.

Worked example 5: use an inverse function in a model

Section titled “Worked example 5: use an inverse function in a model”

A straight path rises 7.47.4 m over a horizontal distance of 3232 m. Find its angle of elevation to the nearest tenth of a degree.

If the angle is θ\theta, then

tanθ=7.432.\tan\theta=\frac{7.4}{32}.

Apply arctangent to both sides:

θ=arctan(7.432).\theta=\arctan\left(\frac{7.4}{32}\right).

In degree mode,

θ13.018.\theta\approx13.018^\circ.

Hence

θ13.0.\boxed{\theta\approx13.0^\circ}.

Do not round 7.4/327.4/32 before applying arctangent. Keeping the full calculator value avoids avoidable error.

Within their stated domains,

sin(arcsinx)=x(1x1),\sin(\arcsin x)=x \qquad (-1\leq x\leq1), cos(arccosx)=x(1x1),\cos(\arccos x)=x \qquad (-1\leq x\leq1),

and

tan(arctanx)=x(xR).\tan(\arctan x)=x \qquad (x\in\mathbb R).

In the opposite order, the answer is the original angle only when that angle lies in the inverse function’s principal range.

Worked example 6: why cancellation can fail

Section titled “Worked example 6: why cancellation can fail”

Evaluate arcsin(sin(5π/6))\arcsin(\sin(5\pi/6)).

First evaluate the inner function:

sin(5π6)=12.\sin\left(\frac{5\pi}{6}\right)=\frac12.

Then apply arcsine:

arcsin(12)=π6.\arcsin\left(\frac12\right)=\frac\pi6.

Therefore

arcsin(sin(5π/6))=π/6,\boxed{\arcsin(\sin(5\pi/6))=\pi/6},

not 5π/65\pi/6. Arcsine must return its principal value in [π/2,π/2][-\pi/2,\pi/2].

Worked example 7: exact composition without finding the angle

Section titled “Worked example 7: exact composition without finding the angle”

Find

cos(arcsin35)\cos\left(\arcsin\frac35\right)

exactly.

Let

θ=arcsin35.\theta=\arcsin\frac35.

Then sinθ=3/5\sin\theta=3/5 and π/2θπ/2-\pi/2\leq\theta\leq\pi/2. Since 3/5>03/5>0, θ\theta lies in quadrant I, where cosine is positive.

Using sin2θ+cos2θ=1\sin^2\theta+\cos^2\theta=1,

cos2θ=1925=1625.\cos^2\theta=1-\frac9{25}=\frac{16}{25}.

Therefore cosθ=4/5\cos\theta=4/5, and hence

cos(arcsin35)=45.\boxed{\cos\left(\arcsin\frac35\right)=\frac45}.

Before evaluating an inverse function, check the angle mode.

arctan(1)=45\arctan(1)=45^\circ

in degree mode, but

arctan(1)=π40.7854\arctan(1)=\frac\pi4\approx0.7854

in radian mode. These describe the same angle in different units.

Also check that an arcsine or arccosine input lies in [1,1][-1,1]. A calculator error for arccos(1.2)\arccos(1.2) reflects a mathematical fact: no real angle has cosine 1.21.2.

  • Confusing reciprocal and inverse notation: sin1x\sin^{-1}x conventionally means arcsinx\arcsin x, while (sinx)1(\sin x)^{-1} means 1/sinx=cscx1/\sin x=\csc x. Use arcsin\arcsin when ambiguity is possible.
  • Cancelling without checking a range: arccos(cosx)=x\arccos(\cos x)=x is guaranteed only for 0xπ0\leq x\leq\pi.
  • Treating a principal value as every solution: arcsin(1/2)=30\arcsin(1/2)=30^\circ is one chosen output. The equation sinx=1/2\sin x=1/2 has further solutions.
  • Ignoring undefined points: a reciprocal graph has an asymptote wherever its denominator is zero.
  • Forgetting the sign after taking a square root: identities such as sec2x=1+tan2x\sec^2x=1+\tan^2x produce two possible signs until a quadrant or interval is used.
  • Mixing degrees and radians: write a degree symbol when using degrees. A bare angle is normally interpreted in radians in advanced mathematics.
  1. Evaluate arcsin(1/2)\arcsin(-1/2), arccos(1/2)\arccos(-1/2) and arctan(3)\arctan(\sqrt3) exactly in radians.
  2. State the domain and range of y=arccosxy=\arccos x.
  3. Evaluate cos(arccos(0.3))\cos(\arccos(-0.3)).
  4. Evaluate arccos(cos(5π/3))\arccos(\cos(5\pi/3)).
  5. Find sin(arctan(3/4))\sin(\arctan(3/4)) exactly.
  6. Explain why arcsin2\arcsin 2 has no real value.
Answers
  1. π/6-\pi/6, 2π/32\pi/3 and π/3\pi/3 respectively.
  2. Domain [1,1][-1,1] and range [0,π][0,\pi].
  3. 0.3-0.3, because cos(arccosx)=x\cos(\arccos x)=x for every x[1,1]x\in[-1,1].
  4. cos(5π/3)=1/2\cos(5\pi/3)=1/2, so the principal value is arccos(1/2)=π/3\arccos(1/2)=\pi/3.
  5. Let θ=arctan(3/4)\theta=\arctan(3/4). Then tanθ=3/4\tan\theta=3/4 and θ\theta is in quadrant I. A right triangle has opposite side 33, adjacent side 44 and hypotenuse 55, so sinθ=3/5\sin\theta=3/5.
  6. The range of real sine is [1,1][-1,1], so no real angle has sine 22.
secx=1cosx,cscx=1sinx,cotx=1tanx.\sec x=\frac1{\cos x}, \qquad \csc x=\frac1{\sin x}, \qquad \cot x=\frac1{\tan x}.

Reciprocal functions inherit undefined points from zeros in their denominators. Secant and cosecant have range (,1][1,)(-\infty,-1]\cup[1,\infty).

Inverse trigonometric functions return principal angles:

arcsinx[π2,π2],arccosx[0,π],arctanx(π2,π2).\arcsin x\in\left[-\frac\pi2,\frac\pi2\right], \quad \arccos x\in[0,\pi], \quad \arctan x\in\left(-\frac\pi2,\frac\pi2\right).

These restrictions make the inverses single valued. They must be used when interpreting calculator outputs and simplifying compositions.

Next, develop reciprocal and Pythagorean relationships in trigonometric identities, then use principal values alongside symmetry and periodicity in trigonometric equations.