Why the Choice of Substitution Matters
A substitution can turn an algebraic radical into a rational function and still leave an unnecessarily complicated integral. In a mixed integral, the rational factors already present often indicate which substitution will simplify the calculation most. The goal is not merely to rationalize the radical; it is to rationalize the whole integrand in a useful form.
For a concrete comparison, consider the standard quadratic radical \(R=\sqrt{x^2+1}\). The two quantities \(x+R\) and \(R-x\) are positive for every real \(x\), and they are reciprocals. Choosing one or the other as the parameter gives two valid ways to express \(x\), \(R\), and \(dx\) rationally. The best choice depends on which factors appear in the integral.
Deriving the Two Reciprocal Forms
Set \(t=x+R\). Multiplying \(x+R\) by \(R-x\) gives \(R^2-x^2=1\), so \(R-x=1/t\). Adding and subtracting these expressions gives formulas for \(x\) and \(R\). Differentiating the formula for \(x\) gives the differential.
Because \(R>|x|\), \(t=x+R>0\). As \(x\) ranges over the real numbers, \(t\) ranges over the positive real numbers. In particular, the inverse formula gives the positive square root: \((t^2+1)/(2t)>0\).
Alternatively, set \(u=R-x\). Then \(x+R=1/u\), and adding and subtracting again gives the corresponding formulas. Here \(u>0\), and \(u=1/t\).
The negative sign in \(dx\) is essential: as \(x\) increases, \(t\) increases but \(u=1/t\) decreases. Both substitutions are reversible on the real line, but they can expose different cancellations in an integrand.
A Selection Principle
Before substituting, inspect the factors involving \(x\) and \(R\). If \(x+R\) occurs repeatedly, setting \(t=x+R\) makes that factor exactly \(t\). If the conjugate \(R-x\) is more useful, setting \(u=R-x\) makes it exactly \(u\). A factor such as \(x+2R\) can also become especially simple when expressed using \(u\) or \(t\). The differential must be included in this comparison: a simple transformed denominator can be offset by a more complicated \(dx\).
Worked Example: A Conjugate Factor Favors \(u=R-x\)
Worked Example: Simplifying a Reciprocal Radical Factor
Evaluate \(\displaystyle \int \frac{dx}{R(x+R)}\), where \(R=\sqrt{x^2+1}\).
The factor \(x+R\) is the reciprocal of \(u=R-x\), so use \(u\). The identities give \(x+R=1/u\), \(R=(1+u^2)/(2u)\), and \(dx=-(1+u^2)/(2u^2)\,du\). Substitution yields
Since \(u=R-x\), the result is \(x-R+C\). To check it, differentiate: \(\frac{d}{dx}(x-R)=1-x/R=(R-x)/R=u/R\). Also, \((R-x)(R+x)=1\), so \(1/[R(x+R)]=(R-x)/R=u/R\). The derivative agrees with the integrand.
If instead we had used \(t=x+R\), the same integral would become \(\int t^{-2}\,dt\): indeed \(R=(t^2+1)/(2t)\) and \(dx=(t^2+1)/(2t^2)\,dt\). That is still simple, but \(u\) makes the integrand a constant immediately. This example illustrates a useful, rather than absolute, preference.
Worked Example: A Repeated \(x+R\) Factor Favors \(t=x+R\)
Worked Example: A Cubed Conjugate-Sum Factor
Evaluate \(\displaystyle \int \frac{dx}{R(x+R)^3}\), where \(R=\sqrt{x^2+1}\).
The repeated factor \(x+R\) suggests setting \(t=x+R\). Then \(x+R=t\), \(R=(t^2+1)/(2t)\), and \(dx=(t^2+1)/(2t^2)\,dt\). Therefore
Returning to \(x\), an antiderivative is \(-1/[3(x+R)^3]\). To verify it, first note that \(dt/dx=1+x/R=(x+R)/R=t/R\). Thus
This confirms the result. The reciprocal parameter would also work, but it would express the repeated factor as \(1/u\); the direct parameter makes the factor transparent from the start.
Comparing Both Choices on One Integral
Sometimes the right choice is not obvious from a single factor. The following integral shows how to compare both forms explicitly and choose the one that gives the shorter integration step.
Worked Example: Choosing Between Two Rationalized Forms
Evaluate \(\displaystyle \int \frac{dx}{R(x+R)(x+2R)}\), with \(R=\sqrt{x^2+1}\).
First use \(u=R-x\). Then \(x+R=1/u\), \(x+2R=(3+u^2)/(2u)\), \(R=(1+u^2)/(2u)\), and \(dx=-(1+u^2)/(2u^2)\,du\). These substitutions give
The logarithm is defined because \(u>0\), so \(u^2+3>0\). In terms of \(x\), one antiderivative is \(-\ln((R-x)^2+3)\). For a comparison, use \(t=x+R\) instead. Then \(x+R=t\), \(x+2R=(3t^2+1)/(2t)\), and substitution gives
The decomposition follows because \(2/t-6t/(3t^2+1)=2/[t(3t^2+1)]\). Since \(u=1/t\), the two antiderivatives differ only by a constant: \(2\ln t-\ln(3t^2+1)=-\ln(3+u^2)\). The \(u\)-form is shorter to integrate directly.
Check the chosen result by differentiation. Since \(u=R-x\), \(du/dx=x/R-1=-u/R\). Therefore
Using \(x+R=1/u\) and \(x+2R=(3+u^2)/(2u)\), the original integrand is also \(2u^2/[R(u^2+3)]\). The derivative therefore checks exactly.
Why the Formulas Work
The rationalization is not a guess. It follows from the difference of squares. For \(R^2=x^2+1\), the conjugate product satisfies \((R-x)(R+x)=1\). Thus knowing either \(x+R\) or \(R-x\) determines its partner as a reciprocal. Adding the pair recovers \(R\), while subtracting them recovers \(x\). This proves the inverse formulas; differentiating those formulas then supplies the corresponding differential. In particular, the substitution is reversible because the parameter determines \(x\) uniquely on its positive domain.
For a different quadratic radical, the exact formulas change. The same selection idea can still guide the choice among applicable rationalizing substitutions: identify the factors the substitution will turn into the parameter, its reciprocal, or a low-degree polynomial, and compare the complete transformed integrands. The calculations here rely specifically on \(\sqrt{x^2+1}\); do not assume these particular formulas apply unchanged to a different radicand.
A Common Pitfall: Losing the Sign in \(dx\)
When \(u=R-x\), the differential is negative. Omitting that sign changes the answer, even though the denominator has been rationalized correctly. For instance, in the first example, dropping the minus sign would lead to \(\int 1\,du=u+C\). But \(d(R-x)/dx=x/R-1\), not \(1-x/R\). The correct derivative of \(u\) is \(-u/R\), so the derivative of \(u\) itself does not match the positive integrand; the derivative of \(-u\) does.
How to Make the Choice
Look for \(x+R\), \(R-x\), or combinations such as \(x+2R\) that may simplify under one parameter.
Include the formulas for \(x\), \(R\), and \(dx\); a denominator-only comparison can be misleading.
Compare the resulting rational functions and favor the one with fewer factors or a direct elementary substitution.
Convert back to \(x\) and differentiate on the domain where the original integrand is defined.
The broader lesson is that a rationalizing substitution is a tool for organizing the whole integrand, not just removing a square root. A good choice can reveal cancellations that are difficult to see in the original variable. When two substitutions are both valid, comparing their rational forms is a practical way to decide which route is more efficient.
Check Your Understanding
Use the identities and selection principle from this tutorial to answer the following questions.
- For \(R=\sqrt{x^2+1}\), show directly that \((x+R)(R-x)=1\).
- Write \(x\), \(R\), and \(dx\) in terms of \(u=R-x\), and state the sign of \(dx/du\).
- Which parameter would you test first for an integrand containing \((x+R)^4\)? Explain your choice.
- In the comparison example, why does the \(u\)-substitution produce a direct logarithmic integral?
- What sign error can occur when using \(u=R-x\), and how can differentiating the inverse formula prevent it?
