Why Reduce a Secant Power?
An integral such as \(\int \sec^6 x\,dx\) can be difficult to evaluate directly. A reduction recurrence replaces an integral with a high exponent by an expression involving a lower exponent. For secant powers, the exponent drops by two at each step, so repeated use eventually reaches one of two simple base cases.
The recurrence is useful because it gives a systematic route from a general positive integer exponent to explicit antiderivatives. Its derivation also shows why integration by parts and the identity \(\tan^2 x=\sec^2 x-1\) work together particularly well here.
The Secant-Power Recurrence
For an integer \(n\ge 2\), the reduction formula is
In integral notation, this says that an antiderivative of \(\sec^n x\) can be written as the displayed boundary term plus \(\frac{n-2}{n-1}\) times an antiderivative of \(\sec^{n-2}x\). Any constant of integration is included in the final answer.
The condition \(n\ge 2\) matters. The formula is obtained by dividing by \(n-1\), and the reduction is designed for integer exponents that can be lowered by two. When \(n=2\), the coefficient of \(I_0\) is zero, so the formula gives the antiderivative directly.
Deriving the Formula
On an interval where \(\cos x\ne 0\), we can use \(\frac{d}{dx}\tan x=\sec^2 x\). Split off a factor of \(\sec^2 x\) and apply integration by parts, taking \(u=\sec^{n-2}x\) and \(dv=\sec^2 x\,dx\). Then \(v=\tan x\), and the chain rule gives \(du=(n-2)\sec^{n-2}x\tan x\,dx\). Therefore,
Now use \(\tan^2 x=\sec^2 x-1\). The integral on the right becomes a difference of two secant-power integrals:
Substitute this expression into the integration-by-parts equation, then collect the terms involving \(I_n\):
Since \(n\ge2\), \(n-1\ne0\). Dividing by \(n-1\) proves the recurrence. This derivation explains the reduction: integration by parts produces another integral at the original exponent, but the tangent identity expresses it as a difference that allows the original integral to be isolated.
Base Cases and Termination
Repeatedly applying the recurrence lowers the exponent by two. If the original exponent is even, the reductions end at \(n=2\), and then at exponent zero as needed. If it is odd, they end at \(n=3\), followed by exponent one. The two base antiderivatives are
For the second identity, differentiating the logarithm gives \(\sec x\) wherever \(\cos x\ne0\). The quantity \(\sec x+\tan x\) does not vanish on this domain: \((\sec x+\tan x)(\sec x-\tan x)=1\). Thus the logarithm is defined with the absolute value. For exponent two, the recurrence itself gives \(I_2=\tan x\), since its \(I_0\) coefficient is zero.
Worked Reductions
Worked Example: Integrating \(\sec^4 x\)
Apply the recurrence with \(n=4\). The exponent drops to two:
To check the result, differentiate its nonconstant terms. Using \(\frac{d}{dx}\sec x=\sec x\tan x\) and \(\tan^2 x=\sec^2 x-1\),
The derivative matches the integrand, so the antiderivative is verified.
Worked Example: Integrating \(\sec^5 x\)
For an odd exponent, reduce until the exponent is one. First apply the recurrence at \(n=5\), then at \(n=3\):
Substituting the last two lines into the first gives
For a check, differentiate each term and combine the results:
Adding these derivatives cancels the \(\sec^3 x\) and \(\sec x\) terms, leaving \(\sec^5 x\), as required.
Worked Example: Integrating \(\sec^6 x\)
Apply the recurrence once to reduce the exponent from six to four, then use the result for \(I_4\):
Differentiate the three nonconstant terms:
The \(\sec^4 x\) and \(\sec^2 x\) terms cancel when these are added, leaving \(\sec^6 x\). This confirms the result.
A Common Sign Error
A frequent mistake is to use \(\tan^2 x=1-\sec^2 x\), reversing the terms in the identity. That sign error changes the recurrence. For example, it could lead to the incorrect expression \(I_4=\frac{1}{3}\sec^2 x\tan x-\frac{2}{3}\tan x\). Differentiating shows that it is not an antiderivative of \(\sec^4 x\):
The correction is to keep the identity in its correct form, \(\tan^2 x=\sec^2 x-1\). Then \(\int\sec^{n-2}x\tan^2 x\,dx=I_n-I_{n-2}\), which leads to a positive coefficient \(\frac{n-2}{n-1}\) on the lower-power integral.
Using the Recurrence Efficiently
Before applying the formula, check that the exponent is an integer at least two and that the interval does not cross a point where \(\cos x=0\). Then lower the exponent by two until reaching the appropriate base case. Keep the recurrence's coefficients attached to each lower-power integral, and include one constant of integration in the final answer.
This method matters because it turns a family of integrals into a predictable sequence of simpler problems. The parity of the exponent tells you which base case will finish the calculation: an even power reaches \(I_0\), while an odd power reaches \(I_1\). Differentiation provides a direct check that the assembled terms reproduce the original secant power.
Check Your Understanding
Answer each question using the recurrence and its conditions.
- State the secant-power reduction formula and the range of integer exponents for which it applies.
- Which base case is reached when reducing \(\sec^7 x\), and which when reducing \(\sec^8 x\)?
- Use one recurrence step to express \(I_7\) in terms of \(\sec^5 x\tan x\) and \(I_5\).
- Which identity for \(\tan^2 x\) is used in the derivation, and how does it produce \(I_n-I_{n-2}\)?
- Why must the interval of integration avoid points where \(\cos x=0\)?