Why Combine Substitution and Symmetry?
A useful change of variable can make a definite integral easier in two ways at once. It can simplify the formula for the integrand, and it can move the interval of integration to one where symmetry is visible. Once that symmetry appears, paired contributions may cancel or combine.
The order matters: first transform the integral correctly, including its differential and bounds; then test the transformed integrand for symmetry. Symmetry does not generally make every integral elementary, but it can reduce the work substantially or determine the value exactly.
The Symmetry Rules We Will Use
An interval is symmetric about \(c\) if it has the form \([c-r,c+r]\), where \(r>0\). A function \(H\) is even about \(c\) if \(H(c+t)=H(c-t)\) for every \(t\) in the relevant range. It is odd about \(c\) if \(H(c+t)=-H(c-t)\).
These facts are useful after a substitution has created a symmetric interval and a transformed integrand with the corresponding parity. Evenness means the contributions on the two sides match; oddness means they cancel.
How a Substitution Reveals Symmetry
Suppose \(f\) is continuous on \([a,b]\), and use a continuously differentiable, strictly increasing substitution \(u=\phi(x)\) with \(\phi'(x)>0\). Write \(\alpha=\phi(a)\) and \(\beta=\phi(b)\). If \(x=x(u)\) is the inverse substitution, then
The transformed integrand is the entire expression \(f(x(u))\frac{dx}{du}\), not just \(f(x(u))\). The bounds also change from \(a,b\) to \(\alpha,\beta\). The goal is to choose \(\phi\) so the new bounds are centered around some \(c\), and the transformed integrand has a simple relationship between \(c+t\) and \(c-t\).
Look for an inner expression whose endpoint values become equally distant from a center, or whose reflection simplifies the integrand.
Replace the integrand and differential, and compute both new bounds.
Compare the transformed integrand at \(c+t\) and \(c-t\), or pair original values under a reflection of the interval.
Use cancellation for oddness, doubling for evenness, or add reflected integrands when neither parity condition holds.
Why the Symmetry Formulas Hold
The formulas follow by splitting at the center and reflecting one half of the interval. On the left half, set \(u=c-t\); on the right half, use \(u=c+t\). Equivalently, the integral over the left half can be rewritten using the variable \(t\) running from \(0\) to \(r\). This gives
If \(H\) is even about \(c\), the two terms in parentheses are equal, so the right side is \(2\int_0^r H(c+t)\,dt=2\int_c^{c+r}H(u)\,du\). If \(H\) is odd about \(c\), those terms sum to zero for every \(t\), so the integral is zero. This derivation also explains a more general strategy: even without pure parity, simplify the sum of the two reflected values.
Worked Example: A Substitution Reveals Oddness
Worked Example: A Substitution Reveals Oddness
Evaluate \(\int_0^1 2x\,\frac{x^2-\frac12}{1+(x^2-\frac12)^2}\,dx\).
The composite expression \(x^2-\frac12\) suggests the substitution \(u=x^2-\frac12\). Its differential and endpoint values are
Thus the integral becomes
The function \(H(u)=\frac{u}{1+u^2}\) is odd because its denominator is unchanged by \(u\mapsto -u\), while its numerator changes sign. The bounds are symmetric about zero, so the integral is zero.
For a direct check, an antiderivative is \(\frac12\ln(1+u^2)\), since its derivative is \(\frac12\cdot\frac{2u}{1+u^2}=\frac{u}{1+u^2}\). Evaluating at the bounds gives
Worked Example: A Substitution Reveals Evenness
Worked Example: A Substitution Reveals Evenness
Evaluate \(\int_0^1 2x\bigl(1+(x^2-\frac12)^2\bigr)\,dx\).
Again set \(u=x^2-\frac12\). Then \(du=2x\,dx\), and the bounds are \(-\frac12\) and \(\frac12\). Therefore
The transformed integrand is even. Apply the even-function rule and evaluate on half the interval:
To check, an antiderivative in the original variable is \(x^2+\frac{(x^2-\frac12)^3}{3}\). Its derivative is \(2x+2x(x^2-\frac12)^2\), the original integrand. At \(x=1\), its value is \(1+\frac{1}{24}=\frac{25}{24}\); at \(x=0\), its value is \(-\frac{1}{24}\). Their difference is \(\frac{26}{24}=\frac{13}{12}\), as obtained by symmetry.
Pairing Reflected Values
Sometimes a substitution does not produce an integrand that is plainly even or odd. It can still be useful if it pairs values across the interval. For an integrable function \(f\) on \([a,b]\), the reflection \(x=a+b-t\) reverses the interval. The negative differential and reversed bounds cancel, giving
Adding this identity to the original integral shows that the integral can be computed using the average of the two reflected values. This is especially effective when their sum is much simpler than either value alone.
Worked Example: Reflected Values Add to a Constant
Worked Example: Reflected Values Add to a Constant
Evaluate \(\int_0^1 \frac{x^2}{x^2+(1-x)^2}\,dx\).
The denominator is positive on the interval: its two squared terms cannot both vanish. Let \(I\) denote the integral. Under the reflection \(x=1-t\), the bounds reverse and \(dx=-dt\); after reversing the bounds, this gives
Rename \(t\) as \(x\) and add this expression to the original one. The denominators match, and the numerators sum to that denominator:
Therefore \(I=\frac12\). The check is the paired-integrand identity itself: at each \(x\), the two reflected fractions sum to \(1\), so their integrals sum to the length of the interval, which is \(1\). Since the two integrals are equal, each is \(\frac12\).
A Common Pitfall: Losing the Differential Factor
A substitution changes both the expression in the integrand and the differential. Omitting the differential factor can produce a plausible-looking symmetry calculation with the wrong scale.
Worked Example: Correcting a Lost Factor
For the even-integrand example above, a wrong calculation might write \(u=x^2-\frac12\), then replace \(2x\,dx\) by \(2\,du\), obtaining
This is incorrect: differentiating \(u=x^2-\frac12\) gives \(du=2x\,dx\), so the factor \(2x\,dx\) is exactly \(du\), not \(2\,du\). The correct transformed integral is \(\int_{-1/2}^{1/2}(1+u^2)\,du=\frac{13}{12}\). The earlier endpoint and antiderivative check confirms this value.
When This Strategy Is Useful
Combining substitution and symmetry is most effective when the integrand contains a composite expression that naturally becomes a centered variable. Powers such as \(x^2\), shifts such as \(x^2-k\), or expressions whose endpoint values are reflected around a common center are possible clues. After the change of variable, check parity on the whole transformed interval; a formula that appears even or odd locally is not enough if the bounds are not symmetric.
For reflected-value pairing, look for an interval-preserving reflection such as \(x\mapsto a+b-x\). The transformed expression need not match the original one. The key opportunity is that their sum may simplify. In every case, verify the bounds, orientation, and differential before using symmetry; otherwise the symmetry argument may apply to a different integral from the one intended.
Check Your Understanding
Use the transformed integrand and the bounds to decide what symmetry information is available.
- What condition on the new bounds allows the odd-function cancellation rule to apply?
- Under \(u=x^2-\frac12\), what are \(du\) and the transformed bounds when \(x\) runs from \(0\) to \(1\)?
- For an even transformed integrand on \([c-r,c+r]\), which half-interval integral can replace the full integral, and with what factor?
- Why can adding an integral to its reflected version help even when neither integrand is even or odd?
- For the substitution \(u=x^2-\frac12\), explain why replacing \(2x\,dx\) by \(2\,du\) is incorrect.
