Integrals & Estimation Tricks (IPhO Math Tools)

Many IPhO integrals look scary until you do one smart move: a substitution that exposes the scale, a symmetry argument, or a recognition that you only need an estimate. This lesson focuses on the moves that most often convert “hard” integrals into one-line results.

1. Definitions (Must Know)

A. Definite vs indefinite integrals

• Indefinite integral: ∫ f(x) dx = F(x) + C (an antiderivative).
• Definite integral: ∫ₐ^b f(x) dx (a number, often an area or accumulated quantity).

B. Substitution (change of variables)

If u = u(x), then

Always change limits for definite integrals.

C. Integration by parts

From the product rule:

Good for: polynomials times exponentials/trig, and for turning an integral into something simpler or bounded.

D. Improper integrals

Integrals with infinite limits or singularities are defined by limits, for example:

E. Estimation language

If an integrand has typical size A over a width w, then a first estimate is

This becomes accurate when the integrand is sharply peaked (or nearly constant) in the region that contributes most of the area.

2. Key Ideas (What Earns Marks)

• Use a scaling substitution to pull parameters outside the integral.
• Look for symmetry (even and odd functions, periodicity, geometric symmetry).
• If you see a product, consider integration by parts.
• If the integral depends on a parameter, consider differentiating with respect to that parameter.
• When exact evaluation is unnecessary, bound and estimate:
  • identify where the integrand is large,
  • approximate locally (often by a Gaussian),
  • state the expected error scale.

3. Detailed Explanations

A. Scaling substitution (parameter extraction)

If an integral contains a parameter a only through x/a, try x = a u.

Example pattern:

The physics interpretation: the integrand decays on a length scale set by a.

B. Symmetry (even, odd, and geometry)

If f(-x) = -f(x) (odd), then

If f(-x) = f(x) (even), then

These are time-saving when integrals arise from center-of-mass, field, or average-value calculations.

C. Parts as a “reduction move”

Integration by parts is often used to:

• reduce powers (for example, ∫ xⁿ e^(-x) dx),
• move derivatives onto an easier factor,
• generate recursion relations (useful for ∫₀^(π/2) sinⁿ θ dθ).

D. Differentiate under the integral sign (Feynman trick)

If

and the integral behaves well, then

You differentiate to get a simpler integral, then integrate back in α using an easy boundary value.

E. Laplace method (sharp peak estimates)

If f(x) has a maximum at x₀ and the integral contains e^(n f(x)) with large n, the main contribution comes from near x₀.

Expand:

and the integral often becomes approximately Gaussian.

You rarely need full rigor in IPhO scripts, just a clear dominant-region argument.

4. Common Mistakes

• Doing a substitution but forgetting the dx Jacobian or the new limits.
• Losing a minus sign when reversing limits.
• Integrating by parts in a loop (you end up with the same integral again without progress).
• Treating an improper integral as convergent without checking decay at infinity or behaviour at a singular point.
• Estimating without stating where the integral “gets its area”.

5. Exam Tips

• Decide early: exact integral, or estimate. If the question asks for scaling, do not over-integrate.
• If you introduce a substitution, write it and the transformed limits explicitly (one line each).
• Keep a mini mental list of standard results:
  • Gaussian integrals
  • ∫₀^∞ e^(-ax) dx
  • ∫ dx/(a² + x²) = 1/a arctan(x/a)
• For estimates, write “dominant region” in words. This often earns method credit.

6. Worked Examples

A. A Gaussian normalization integral

Evaluate ∫_(-∞)^∞ exp(-x²/2σ²) dx.

B. A rational integral that looks harder than it is

Evaluate ∫₀^Rr²/(a² + r²) dr for a > 0.

C. A standard improper integral with a fast complex method

For a > 0, evaluate ∫₀^∞e^(-ax) cos(bx) dx.

7. Mind Stretchers

A. Estimating ∫₀^(π/2) sinⁿ θ dθ for large n

Let

Estimate Iₙ for n≫ 1.

8. Practice