Why Decompose an Integrand?
A function does not have to be entirely even or entirely odd for symmetry to help evaluate its integral. It may contain both kinds of behavior at once. For example, \(x^2+3x\) is neither even nor odd, but its \(x^2\) term is even and its \(3x\) term is odd. On an interval symmetric about zero, the odd part cancels while the even part remains.
The useful idea is to separate these contributions systematically. Every function defined on a domain symmetric about zero has a natural even component and a natural odd component. Once separated, the integral over a centered interval depends only on the even component.
Definitions and the Decomposition
A domain \(D\) is symmetric about zero if \(x\in D\) implies \(-x\in D\). This condition is essential: both \(f(x)\) and \(f(-x)\) must be defined for each \(x\) under consideration.
A function \(g\) is even if \(g(-x)=g(x)\) wherever both sides are defined. A function \(h\) is odd if \(h(-x)=-h(x)\). These definitions concern how function values change under reflection; an odd function need not be a power of \(x\), and an even function need not be a power of \(x^2\).
The two formulas give the original function back when added:
They also have the required parity. Replacing \(x\) by \(-x\) in the formula for \(f_e\) gives \(f_e(-x)=(f(-x)+f(x))/2=f_e(x)\). For the odd component, \(f_o(-x)=(f(-x)-f(x))/2=-f_o(x)\).
To see uniqueness, write \(f(x)=g(x)+h(x)\) and \(f(-x)=g(x)-h(x)\), using the parity of \(g\) and \(h\). Adding these equations gives \(f(x)+f(-x)=2g(x)\), so \(g(x)=f_e(x)\). Subtracting the second equation from the first gives \(f(x)-f(-x)=2h(x)\), so \(h(x)=f_o(x)\). Thus no other even–odd splitting of the same function is possible.
What the Decomposition Does to a Definite Integral
Suppose \(a>0\) and \(f\) is continuous on \([-a,a]\). Its even and odd components are then continuous there as well, so their definite integrals exist. The odd component contributes zero on this interval, and the integral of the even component can be written using only the positive half of the interval.
Here is the cancellation argument for the odd part. In \(\int_{-a}^{a} f_o(x)\,dx\), substitute \(x=-u\), so \(dx=-du\). When \(x=-a\), \(u=a\); when \(x=a\), \(u=-a\). Therefore
The integral equals its own negative, so it is zero. By linearity of the definite integral and \(f=f_e+f_o\), this proves that \(\int_{-a}^{a}f=\int_{-a}^{a}f_e\).
For the even part, split the interval at zero. In the integral from \(-a\) to \(0\), substitute \(x=-u\). Evenness gives \(f_e(-u)=f_e(u)\), so that half of the integral equals the integral from \(0\) to \(a\). Adding the two halves proves the factor of \(2\). This is a reduction, not a claim that the integral must be easy: the remaining integral of \(f_e\) may still require an antiderivative or another method.
Worked Examples
Worked Example: Separate a Polynomial Integrand
Find the even and odd components of \(f(x)=x^5+3x^2-4x+7\), then evaluate \(\int_{-2}^{2}f(x)\,dx\).
First calculate the reflected value. Odd powers change sign under \(x\mapsto -x\), while even powers do not:
Using the defining formulas gives
As a check, their sum is \(3x^2+7+x^5-4x=x^5+3x^2-4x+7=f(x)\). The odd component integrates to zero, so
Check by using an antiderivative of the full original function: \(F(x)=x^6/6+x^3-2x^2+7x\). Then \(F(2)=64/6+8-8+14=74/3\) and \(F(-2)=64/6-8-8-14=-58/3\). Thus \(F(2)-F(-2)=132/3=44\), as obtained by decomposition.
Worked Example: Separate Two Rational Terms
Evaluate \(\int_{-1}^{1}\left(\frac{x}{1+x^2}+\frac{1}{1+x^2}\right)\,dx\) by decomposing the integrand.
Let \(f(x)=\frac{x}{1+x^2}+\frac{1}{1+x^2}\). Since \(1+(-x)^2=1+x^2\),
Consequently, averaging \(f(x)\) and \(f(-x)\) leaves the second term, while taking half their difference leaves the first:
The sum is the original integrand, which checks the decomposition. The odd part has zero integral on \([-1,1]\), and an antiderivative of \(1/(1+x^2)\) is \(\arctan x\): its derivative is \(1/(1+x^2)\). Therefore
For a direct check, the original integrand has antiderivative \(\frac12\ln(1+x^2)+\arctan x\). Evaluating at \(1\) gives \(\frac12\ln 2+\frac{\pi}{4}\), and at \(-1\) gives \(\frac12\ln 2-\frac{\pi}{4}\). Their difference is \(\pi/2\).
Worked Example: Decompose a Quotient
Find the even and odd components of \(f(x)=\frac{1+x}{1-x}\) on \([-1/2,1/2]\), and evaluate its integral on that interval.
The denominator is nonzero throughout this interval. Reflecting \(x\) gives \(f(-x)=\frac{1-x}{1+x}\). Put the two fractions over the common denominator \(1-x^2\):
Indeed, \((1+x^2)/(1-x^2)+2x/(1-x^2)=(1+x)^2/(1-x^2)=(1+x)/(1-x)\), as required. It remains to integrate the even part. Rewrite its numerator as \(1+x^2=-(1-x^2)+2\), so \(\frac{1+x^2}{1-x^2}=-1+\frac{2}{1-x^2}\). An antiderivative is \(-x+\ln\left(\frac{1+x}{1-x}\right)\): differentiating the logarithm gives \(1/(1+x)+1/(1-x)=2/(1-x^2)\). Thus
The antiderivative of the even part differentiates to \((1+x^2)/(1-x^2)\), and its value at zero is zero, confirming the endpoint evaluation. The displayed result is exact.
A Common Mistake: Testing the Whole Function Only
A frequent mistake is to check whether the entire integrand is even or odd, and then conclude that symmetry is useless if it is neither. For example, \(x^2+x\) is neither: reflecting it gives \(x^2-x\), which is not the same function or its negative. It would be incorrect to conclude that its integral on \([-1,1]\) is zero just because the interval is symmetric.
Instead, decompose it: \(f_e(x)=x^2\) and \(f_o(x)=x\). The odd part cancels, but the even part remains:
The concrete wrong calculation would be \(\int_{-1}^{1}(x^2+x)\,dx=0\), based on treating the mixed integrand as odd. The correction is to test or construct the components separately: only the odd component has zero integral. A reliable check is to add the proposed components and confirm that their sum is exactly the original function.
When This Method Is Useful
Use the decomposition when the interval is \([-a,a]\) and the integrand is mixed or its parity is not immediately apparent. The formulas can expose cancellation even when the expression is written in a complicated form. In practice, the calculation follows a short sequence:
Confirm that the integrand is defined at both \(x\) and \(-x\) throughout the interval.
Calculate \(f(-x)\), then form \((f(x)+f(-x))/2\) and \((f(x)-f(-x))/2\).
Verify the component parities and add the components to recover \(f(x)\).
Discard the odd component in the definite integral and evaluate the even component, often as twice its integral from zero to \(a\).
The method depends on a symmetric interval centered at zero for the cancellation stated here. It does not say that the odd part vanishes on an arbitrary interval, nor does it eliminate the need to check that the function is defined and integrable. The decomposition itself is pointwise algebra; the integral shortcut additionally requires an interval and integrability conditions that allow the definite integral to be taken.
Check Your Understanding
Use the definitions and conditions in this tutorial to answer the following.
- Find the even and odd components of \(f(x)=x^4-2x^3+5x\), and check that they sum to \(f\).
- Why must the domain of a function be symmetric about zero to define both components using \(f(x)\) and \(f(-x)\)?
- If \(f\) is continuous on \([-a,a]\), which component determines \(\int_{-a}^{a}f(x)\,dx\), and why?
- Evaluate \(\int_{-2}^{2}(x^2+4x)\,dx\) using a decomposition.
- Does the odd component necessarily have an antiderivative that is even? Explain what parity of an antiderivative would be consistent with differentiating it.
