AP Calculus AB and BC glossary

Numerical integration

Also called: Approximate integration

Numerical integration means approximating a definite integral with sums of areas, using left, right, midpoint, or trapezoid estimates. You need it when the integrand has no elementary antiderivative, or when the function is given only as a table of values.

Two situations force it. Some integrands, ex2e^{-x^2} and sinxx\frac{\sin x}{x} among them, have no antiderivative expressible in elementary functions. More often on the exam there is no formula at all, only a table of measurements, and a sum is the only thing you can build. A graph is different: when it is made of line segments and circular arcs, geometry gives the integral exactly, and you only approximate when the curve has no geometric shape.

The four standard estimates all approach the same number as the number of subintervals grows, but not at the same speed. Doubling nn roughly halves the error of a left or right sum, and roughly quarters the error of a midpoint or trapezoid estimate, so the second pair is worth using when the data allows it.

On a calculator active question the machine returns the value in one step, but the credit is in the setup. Write the definite integral you evaluated, then the number, rounded to three decimal places or better.

The mistake

Reporting the estimate with an equals sign and stopping there. These are approximations, and questions usually want the direction too. Monotonicity settles left and right sums, so a left sum of an increasing rate is an underestimate, while concavity settles midpoint and trapezoid.

Appears in: Unit 6: Integration and Accumulation