Why the Choice of Substitution Matters
A substitution is useful when it makes the structure of an integrand easier to integrate. The choice is often suggested by a familiar pattern: a function of an inner expression appears together with the derivative of that inner expression. Replacing the inner expression with a new variable can then turn a complicated-looking integrand into a direct antiderivative.
There may be more than one possible choice. A good choice reduces the work immediately; another may leave the integral just as complicated, or introduce extra algebra. The aim here is to compare candidates and choose one that exposes the simplest structure. The central test is not whether an expression can be called \(u\), but whether the resulting integral is easier to evaluate.
The Chain-Rule Pattern Behind Substitution
The reason this strategy works is the chain rule. If \(H\) is an antiderivative of a function \(h\), and \(g\) is differentiable, then
Thus an integrand of the form \(h(g(x))g'(x)\) has antiderivative \(H(g(x))+C\). Substitution names the inner expression \(u=g(x)\), so its derivative is \(du=g'(x)\,dx\). The new integral is \(\int h(u)\,du\), which can be integrated using an antiderivative of \(h\).
This also explains why a constant multiple of the inner derivative is enough. If the integrand contains \(k g'(x)\,dx\), where \(k\) is a nonzero constant, then \(g'(x)\,dx=du\) and the factor \(k\) remains outside the new integral. Keeping track of that factor is part of choosing and carrying out the substitution correctly.
A Practical Way to Compare Candidates
When an integrand has several layers, more than one expression may look like a possible inner function. Compare candidates by asking what the entire integral becomes after substitution. A promising choice usually has three features:
- It replaces a repeated or complicated expression with one variable.
- Its derivative is already present, up to a manageable constant or sign.
- The transformed integrand has a standard, simpler antiderivative.
For instance, if the integrand contains a power of \(3x^2+2\) multiplied by \(x\), choosing \(u=3x^2+2\) turns the power into a power of \(u\). Choosing \(u=x\) changes no essential structure. The first choice is useful because it simplifies the composition, not simply because it is an available expression.
It is also possible for two choices to work. In that case, choose the one that makes the transformed integral most direct. An alternative is not automatically wrong just because it takes more steps; the relevant comparison is the amount of simplification it provides.
Worked Example: An Algebraic Inner Expression
Worked Example: An Algebraic Inner Expression
Evaluate \(\int 6x(3x^2+2)^4\,dx\). The expression \(3x^2+2\) occurs inside a fourth power, and its derivative is \(6x\), exactly the factor multiplying that power. Choose \(u=3x^2+2\).
Substitute both the inner expression and its differential:
Return to \(x\):
Check by differentiating the proposed antiderivative:
This is the original integrand, so the antiderivative checks. The candidate \(u=x\) would leave the factor \((3x^2+2)^4\) in terms of \(x\) and would not expose the simple power rule as directly.
Worked Example: An Exponential Expression
Worked Example: An Exponential Expression
Evaluate \(\int \frac{e^{2x}}{3+e^{2x}}\,dx\). The denominator is a natural candidate because its variable part has derivative \(2e^{2x}\), which is twice the numerator. Let \(u=3+e^{2x}\).
Replacing the denominator and numerator factor gives
Since \(3+e^{2x}>0\), the absolute value can be omitted after substituting back:
Check by differentiation:
One could instead start with \(v=2x\), which changes the integral to \(\frac{1}{2}\int \frac{e^v}{3+e^v}\,dv\). That route is valid, but it still requires recognizing the derivative of the denominator. Taking the denominator itself as the substitution makes the logarithm pattern explicit at once.
Worked Example: A Trigonometric Expression and Its Sign
Worked Example: A Trigonometric Expression and Its Sign
Evaluate \(\int \frac{\sin x}{2+\cos x}\,dx\). The denominator contains \(\cos x\), whose derivative is \(-\sin x\). This makes \(u=2+\cos x\) a useful choice, but the negative sign must be included.
Therefore,
Because \(2+\cos x\geq 1\), it is positive for every real \(x\), so the answer is
Differentiate to check:
The candidate \(u=\sin x\) does not simplify the denominator directly: expressing \(\cos x\) in terms of \(\sin x\) would introduce a square root and a sign issue. The denominator choice instead uses its derivative and produces a logarithm immediately.
Why These Choices Work
In each example, the selected expression organizes the integrand into a composition and its derivative. The first example has a power of an inner polynomial; the second has a reciprocal of an expression built from an exponential; the third has a reciprocal of a trigonometric expression. In every case, differentiation of the chosen inner expression accounts for the remaining factor in the numerator.
This is more than a visual trick. If the substitution produces \(\int h(u)\,du\), and \(H'=h\), then the chain rule gives
The final expression is exactly the original integrand when \(u=g(x)\) and the differential factor is present. This establishes why the computed antiderivative is valid. Differentiating the answer afterward is a practical check that the factors and signs were carried correctly.
A Common Pitfall: Losing the Sign
In the trigonometric example, a tempting but incorrect line is
The error is replacing \(\sin x\,dx\) by \(du\) instead of by \(-du\). The proposed answer fails its check:
which is the negative of the integrand. The correction is to retain the minus sign from \(du=-\sin x\,dx\), giving \(-\ln(2+\cos x)+C\). Checking by differentiation is especially useful when the inner expression has a derivative with a negative sign.
Choosing Without Overcomplicating
A substitution does not need to remove every \(x\) by itself or make the integral look shorter at the first glance. Its purpose is to expose a tractable structure. Before committing to a candidate, write its derivative and inspect the whole integrand. If the transformed expression still contains an awkward mixture of the old variable and the new one, the choice may not be helping.
Look for a repeated inner quantity, such as a polynomial inside a power, an exponential inside a denominator, or a trigonometric expression inside a logarithm-like pattern.
Check whether a candidate's derivative appears among the remaining factors, allowing for a constant multiple or a negative sign.
Prefer the candidate that reduces the integrand to a familiar, simpler form with the fewest extra steps.
Differentiate the proposed antiderivative and confirm that it reproduces the original integrand.
When two choices produce similarly simple integrals, either may be suitable. Choosing the expression that appears as the whole denominator or repeated inner quantity is often a convenient way to make the composition visible. But the decisive evidence is the transformed integral, not the appearance of the candidate alone.
Check Your Understanding
For each question, focus on how the candidate changes the structure of the integral.
- For \(\int 4x(x^2+5)^3\,dx\), what inner expression is a useful substitution, and what is its derivative?
- In \(\int \frac{e^x}{4+e^x}\,dx\), compare \(u=e^x\) with \(u=4+e^x\). Which makes the logarithm pattern most direct?
- What sign must be included when \(u=1-\cos x\), and why?
- Give one reason a candidate substitution may be valid but not useful.
- What calculation checks whether a proposed antiderivative is correct?
