Why a Quadratic Radical Can Signal Trigonometric Substitution
A square root involving a constant and a squared variable can hide a Pythagorean identity. Recognizing that structure can suggest a substitution that replaces the radical with a trigonometric expression. The aim here is to identify the candidate substitution and check that it fits the domain; we will not carry out the resulting integration.
This is a different signal from an integrand built mainly from powers or products of trigonometric functions. In that setting, the trigonometric factors themselves guide the method. Here, the structure inside the square root is the clue.
The Three Standard Forms
The substitutions follow from the identities \(\sin^2\theta+\cos^2\theta=1\) and \(\sec^2\theta-\tan^2\theta=1\). The angle range matters: a square root is nonnegative, so a trigonometric expression replacing it must also be nonnegative on the chosen range.
| Radical form | Candidate substitution | Useful range and simplification |
|---|---|---|
| \(\sqrt{a^2-x^2}\) | \(x=a\sin\theta\) | \(-\frac{\pi}{2}\leq\theta\leq\frac{\pi}{2}\), so \(\sqrt{a^2-x^2}=a\cos\theta\) |
| \(\sqrt{a^2+x^2}\) | \(x=a\tan\theta\) | \(-\frac{\pi}{2}<\theta<\frac{\pi}{2}\), so \(\sqrt{a^2+x^2}=a\sec\theta\) |
| \(\sqrt{x^2-a^2}\), with \(x\geq a\) | \(x=a\sec\theta\) | \(0\leq\theta<\frac{\pi}{2}\), so \(\sqrt{x^2-a^2}=a\tan\theta\) |
In each case, \(a\) is positive. It sets the scale of the expression and determines where the radical is defined. The third form also has a negative branch, \(x\leq-a\). On that branch one can use \(x=-a\sec\theta\) with \(0\leq\theta<\frac{\pi}{2}\); this keeps the radical equal to \(a\tan\theta\). For an integral over both branches, the domain should be handled in separate intervals.
Why the Substitutions Work
For the first form, the domain condition is \(-a\leq x\leq a\). Put \(x=a\sin\theta\), with \(-\frac{\pi}{2}\leq\theta\leq\frac{\pi}{2}\). Then \(dx=a\cos\theta\,d\theta\), and \(\cos\theta\geq0\) on this interval. Therefore:
The last equality uses both \(a>0\) and \(\cos\theta\geq0\). Without the range condition, the square root would be \(a|\cos\theta|\), not necessarily \(a\cos\theta\).
For the second form, \(x=a\tan\theta\) gives \(dx=a\sec^2\theta\,d\theta\). On \((-\frac{\pi}{2},\frac{\pi}{2})\), \(\sec\theta>0\), so:
For the third form on \(x\geq a\), use \(x=a\sec\theta\) with \(0\leq\theta<\frac{\pi}{2}\). Then \(dx=a\sec\theta\tan\theta\,d\theta\), and both \(\sec\theta\) and \(\tan\theta\) are nonnegative. Thus:
Normalize the Coefficients Before Choosing
The variable does not need to appear with coefficient \(1\) inside the square. A useful way to recognize the form is to factor out the positive constant and rewrite the expression as a constant times \(1\) plus or minus a squared scaled variable.
For \(A>0\) and \(B>0\), the three patterns can be written as:
Thus the scale in the standard forms is \(a=\sqrt{A/B}\). For the first two forms, \(x=a\) times a trigonometric function; for the third form, the same scale appears in \(x=a\sec\theta\). Normalization is a recognition step, not a reason to force every radical into one of these forms. The expression must have the correct signs and a real domain.
Worked Examples: Identifying the Candidate Substitution
Worked Example: A Difference of Squares Under the Radical
Consider the setup \(\int \frac{dx}{\sqrt{25-4x^2}}\). Identify the candidate substitution and check the transformed integrand, without evaluating it.
First factor the radicand into the standard pattern:
This is the form \(a^2-u^2\), where \(a=5\) and \(u=2x\). Equivalently, the variable scale is \(x=\frac{5}{2}\sin\theta\). The original radical is real when \(25-4x^2\geq0\), or \(-\frac{5}{2}\leq x\leq\frac{5}{2}\). Choose \(-\frac{\pi}{2}\leq\theta\leq\frac{\pi}{2}\), and set:
Substitute into the radical and use \(\cos\theta\geq0\) on the selected range:
Therefore the transformed setup is:
Check the substitution: \(4x^2=25\sin^2\theta\), so \(25-4x^2=25\cos^2\theta\), and its nonnegative square root on this range is \(5\cos\theta\). The transformed integrand follows from that radical and the displayed differential. The setup is checked; no antiderivative has been found here.
Worked Example: A Sum of Squares Under the Radical
Consider \(\int \sqrt{9+4x^2}\,dx\). The plus sign identifies the \(a^2+x^2\) pattern after scaling.
Factor and compare with the standard form:
Here \(a=3\) and the scaled variable is \(\frac{2x}{3}\). The radical is real for every real \(x\). Choose \(-\frac{\pi}{2}<\theta<\frac{\pi}{2}\), and set:
Since \(\sec\theta>0\) on this interval:
Putting both substitutions into the integral gives:
Check the radical by squaring its proposed replacement: \((3\sec\theta)^2=9\sec^2\theta=9+9\tan^2\theta=9+4x^2\). Because \(3\sec\theta>0\) on the chosen range, it equals the nonnegative square root. The substitution and transformed setup are therefore consistent; the remaining trigonometric integral is not evaluated here.
Worked Example: A Difference in the Opposite Order
Consider the setup \(\int \frac{dx}{\sqrt{4x^2-49}}\), restricted to \(x\geq\frac{7}{2}\). The radicand is nonnegative on this branch, and the form is \(x^2-a^2\) after scaling.
The scale is \(a=\frac{7}{2}\), so choose \(x=\frac{7}{2}\sec\theta\). The restriction \(x\geq\frac{7}{2}\) corresponds to \(0\leq\theta<\frac{\pi}{2}\), where \(\sec\theta\geq1\) and \(\tan\theta\geq0\). The differential is:
Now simplify the radical on this range:
Substitution into the setup gives:
Check: \(4x^2-49=49(\sec^2\theta-1)=49\tan^2\theta\), and \(7\tan\theta\geq0\) on the stated range. Thus the radical replacement has the correct sign, and the differential gives the displayed transformed integrand. This verifies the candidate without completing the integration.
Common Mistake: Choosing Sine for a Plus Sign
A tempting but incorrect move is to treat \(\sqrt{a^2+x^2}\) as though it were \(\sqrt{a^2-x^2}\). For example, take \(\sqrt{9+4x^2}\) and try \(x=\frac{3}{2}\sin\theta\). Then the expression becomes:
not \(3\cos\theta\). The sine identity gives \(\cos^2\theta=1-\sin^2\theta\), so it cannot turn the plus sign into a cosine square. The appropriate choice is \(x=\frac{3}{2}\tan\theta\), because \(1+\tan^2\theta=\sec^2\theta\).
A related error is to drop the absolute value without checking the angle range. In general, \(\sqrt{a^2\cos^2\theta}=a|\cos\theta|\). It equals \(a\cos\theta\) only where \(\cos\theta\geq0\). State the range before using the simplified radical.
When to Use This Recognition
A matching radical makes trigonometric substitution a candidate, not an automatic choice. Before committing, check that the radicand is real on the interval of integration, that the coefficient scaling gives one of the three standard forms, and that the angle range makes the radical replacement valid. Also inspect the rest of the integrand: a simpler structural pattern may suggest another method.
If the expression does not already match one of these patterns, do not assume that it does. For instance, a shifted quadratic may require an algebraic rewrite before it can be compared with the standard forms. The important decision at this stage is whether the radical itself provides a clear sine, tangent, or secant signal—not to force a substitution or finish the integral.
Check Your Understanding
For each question, focus on identifying and checking the substitution rather than evaluating an integral.
- For \(\sqrt{16-9x^2}\), what is the candidate substitution, and what interval of \(x\) makes the radical real?
- For \(\sqrt{5+20x^2}\), rewrite the radicand to identify the scale \(a\), then state a suitable substitution.
- For \(\sqrt{9x^2-36}\) on \(x\geq2\), identify the substitution and an angle range that makes the radical simplification nonnegative.
- Why does the range of \(\theta\) matter when replacing \(\sqrt{a^2\cos^2\theta}\) by \(a\cos\theta\)?
- For which of the three standard forms is the original radical defined only on a bounded interval of \(x\)?
