Lec11 - Numerical Integration(Quadrature Another...

Info iconThis preview shows pages 1–7. Sign up to view the full content.

View Full Document Right Arrow Icon
Numerical Integration (Quadrature) Another application for our interpolation tools!
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Lecture 11 2 Integration: Area under a curve Curve = data or function z Integrating data Finite number of data points—spacing specified Data may be noisy Must interpolate or regress a smooth between points z Integrating functions Non-integrable function (usually) As many points as you need Spacing arbitrary Today we will discuss methods that work on both of these problems () b a f xd x Area = Area Area x x ab y y
Background image of page 2
Lecture 11 3 Numerical Integration: The Big Picture z Virtually all numerical integration methods rely on the following procedure: Start from N+1 data points ( x i ,f i ), i = 0,…, N , or sample a specified function f ( x ) at N + 1x i values to generate the data set Fit the data set to a polynomial, either locally (piecewise) or globally Analytically integrate the polynomial to deduce an integration formula of the general form: z Numerical integration schemes are further categorized as either: Closed – the x i data points include the end points a and b of the interval Open – the x i data points are interior to the interval w i are the “weights” x i are the “abscissas”, “points”, or “nodes”
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Lecture 11 4 Further classification of numerical integration schemes z Newton-Cotes Formulas Use equally spaced abscissas Fit data to local order N polynomial approximants Examples: Trapezoidal rule, N=1 Simpson’s 1/3 rule, N=2 Errors are algebraic in the spacing h between points z Clenshaw-Curtis Quadrature Uses the Chebyshev abscissas Fit data to global order N polynomial approximants Errors can be spectral, ~exp(-N) ~ exp (-1/h), for smooth functions z Gaussian Quadrature Unequally spaced abscissas determined optimally Fit data to global order N polynomial approximants Errors can be spectral, and smaller than Clenshaw-Curtis
Background image of page 4
Lecture 11 5 Trapezoid rule: 1 interval, 2 points () b a f xd x Area = x f ( x ) ab x f ( x ) [] 1 ()( ) 2 f af bb a + Area = average height 1 2 f b + f b f a Approximate the function by a linear interpolant between the two end points, then integrate that degree-1 polynomial
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Lecture 11 6 Trapezoid rule: 2 intervals, 3 points () b a f xd x Area = x f ( x ) x 0 x 2 x f ( x ) ab [] 01 12 2 22 ()2 2 hh f xf x f x h fx ++ + =+ + Area = x 1 We should be able to improve accuracy by increasing the number of intervals: this leads to “composite” integration formulas
Background image of page 6
Image of page 7
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page1 / 25

Lec11 - Numerical Integration(Quadrature Another...

This preview shows document pages 1 - 7. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online