Numerical integration

Podobne dokumenty
Chapter 1: Review Exercises

Helena Boguta, klasa 8W, rok szkolny 2018/2019

Weronika Mysliwiec, klasa 8W, rok szkolny 2018/2019

Hard-Margin Support Vector Machines

CS 6170: Computational Topology, Spring 2019 Lecture 09

Zakopane, plan miasta: Skala ok. 1: = City map (Polish Edition)

Gradient Coding using the Stochastic Block Model

Wojewodztwo Koszalinskie: Obiekty i walory krajoznawcze (Inwentaryzacja krajoznawcza Polski) (Polish Edition)

Linear Classification and Logistic Regression. Pascal Fua IC-CVLab

Machine Learning for Data Science (CS4786) Lecture 11. Spectral Embedding + Clustering

Karpacz, plan miasta 1:10 000: Panorama Karkonoszy, mapa szlakow turystycznych (Polish Edition)

Revenue Maximization. Sept. 25, 2018

Logo pole ochronne. 1/2 a. 1/4 a

OpenPoland.net API Documentation

Wojewodztwo Koszalinskie: Obiekty i walory krajoznawcze (Inwentaryzacja krajoznawcza Polski) (Polish Edition)

y = The Chain Rule Show all work. No calculator unless otherwise stated. If asked to Explain your answer, write in complete sentences.

MaPlan Sp. z O.O. Click here if your download doesn"t start automatically

ARNOLD. EDUKACJA KULTURYSTY (POLSKA WERSJA JEZYKOWA) BY DOUGLAS KENT HALL

Machine Learning for Data Science (CS4786) Lecture11. Random Projections & Canonical Correlation Analysis

Surname. Other Names. For Examiner s Use Centre Number. Candidate Number. Candidate Signature

Installation of EuroCert software for qualified electronic signature

Gamma3. Notations. Primary definition. Specific values. General characteristics. Traditional name. Traditional notation

Tychy, plan miasta: Skala 1: (Polish Edition)

General Certificate of Education Ordinary Level ADDITIONAL MATHEMATICS 4037/12

Katowice, plan miasta: Skala 1: = City map = Stadtplan (Polish Edition)

TTIC 31210: Advanced Natural Language Processing. Kevin Gimpel Spring Lecture 8: Structured PredicCon 2

SSW1.1, HFW Fry #20, Zeno #25 Benchmark: Qtr.1. Fry #65, Zeno #67. like

METHOD 2 -DIAGNOSTIC OUTSIDE

Wojewodztwo Koszalinskie: Obiekty i walory krajoznawcze (Inwentaryzacja krajoznawcza Polski) (Polish Edition)

Stargard Szczecinski i okolice (Polish Edition)

Proposal of thesis topic for mgr in. (MSE) programme in Telecommunications and Computer Science

PLSH1 (JUN14PLSH101) General Certificate of Education Advanced Subsidiary Examination June Reading and Writing TOTAL

Leba, Rowy, Ustka, Slowinski Park Narodowy, plany miast, mapa turystyczna =: Tourist map = Touristenkarte (Polish Edition)

Counting quadrant walks via Tutte s invariant method

Pielgrzymka do Ojczyzny: Przemowienia i homilie Ojca Swietego Jana Pawla II (Jan Pawel II-- pierwszy Polak na Stolicy Piotrowej) (Polish Edition)

Wybrzeze Baltyku, mapa turystyczna 1: (Polish Edition)

Aerodynamics I Compressible flow past an airfoil

Karpacz, plan miasta 1:10 000: Panorama Karkonoszy, mapa szlakow turystycznych (Polish Edition)

Previously on CSCI 4622

Dolny Slask 1: , mapa turystycznosamochodowa: Plan Wroclawia (Polish Edition)

Poland) Wydawnictwo "Gea" (Warsaw. Click here if your download doesn"t start automatically

Rozpoznawanie twarzy metodą PCA Michał Bereta 1. Testowanie statystycznej istotności różnic między jakością klasyfikatorów

Rachunek lambda, zima

tum.de/fall2018/ in2357

Analysis of Movie Profitability STAT 469 IN CLASS ANALYSIS #2

Jazz EB207S is a slim, compact and outstanding looking SATA to USB 2.0 HDD enclosure. The case is

Egzamin maturalny z języka angielskiego na poziomie dwujęzycznym Rozmowa wstępna (wyłącznie dla egzaminującego)

Miedzy legenda a historia: Szlakiem piastowskim z Poznania do Gniezna (Biblioteka Kroniki Wielkopolski) (Polish Edition)

Jak zasada Pareto może pomóc Ci w nauce języków obcych?

Wroclaw, plan nowy: Nowe ulice, 1:22500, sygnalizacja swietlna, wysokosc wiaduktow : Debica = City plan (Polish Edition)

DODATKOWE ĆWICZENIA EGZAMINACYJNE

Estimation and planing. Marek Majchrzak, Andrzej Bednarz Wroclaw,

Rev Źródło:

Stability of Tikhonov Regularization Class 07, March 2003 Alex Rakhlin

Extraclass. Football Men. Season 2009/10 - Autumn round

POLITYKA PRYWATNOŚCI / PRIVACY POLICY

Miedzy legenda a historia: Szlakiem piastowskim z Poznania do Gniezna (Biblioteka Kroniki Wielkopolski) (Polish Edition)

Convolution semigroups with linear Jacobi parameters

Has the heat wave frequency or intensity changed in Poland since 1950?

Polski Krok Po Kroku: Tablice Gramatyczne (Polish Edition) By Anna Stelmach

How to translate Polygons

POL1. General Certificate of Education June 2006 Advanced Subsidiary Examination. Responsive Writing. Time allowed: 3 hours. Instructions.

Testy jednostkowe - zastosowanie oprogramowania JUNIT 4.0 Zofia Kruczkiewicz

Wojewodztwo Koszalinskie: Obiekty i walory krajoznawcze (Inwentaryzacja krajoznawcza Polski) (Polish Edition)

Zwora Yale US06. Yale seria US kg. Zastosowanie. Właściwości. Parametry techniczne

ANKIETA ŚWIAT BAJEK MOJEGO DZIECKA

TTIC 31210: Advanced Natural Language Processing. Kevin Gimpel Spring Lecture 9: Inference in Structured Prediction

All Saints Day. Chants of the Proper of the Mass for. Adapted to English words and Edited by. Bruce E. Ford

Midterm Review (2) Trigonometry L R2V0C1K9U XKhuptOaT GSUoefUtowraqrleQ aldlscv.a X VA[ltlB LrmizgNhStesM krgeoseefrzvte_di.

Few-fermion thermometry

Blow-Up: Photographs in the Time of Tumult; Black and White Photography Festival Zakopane Warszawa 2002 / Powiekszenie: Fotografie w czasach zgielku


Mixed-integer Convex Representability

Machine Learning for Data Science (CS4786) Lecture 24. Differential Privacy and Re-useable Holdout


Test sprawdzający znajomość języka angielskiego

Nazwa projektu: Kreatywni i innowacyjni uczniowie konkurencyjni na rynku pracy


EXAMPLES OF CABRI GEOMETRE II APPLICATION IN GEOMETRIC SCIENTIFIC RESEARCH

GRY EDUKACYJNE I ICH MOŻLIWOŚCI DZIĘKI INTERNETOWI DZIŚ I JUTRO. Internet Rzeczy w wyobraźni gracza komputerowego

Formularz recenzji magazynu. Journal of Corporate Responsibility and Leadership Review Form

Ogólnopolski Próbny Egzamin Ósmoklasisty z OPERONEM. Język angielski Kartoteka testu. Wymagania szczegółowe Uczeń: Poprawna odpowiedź 1.1.

USB firmware changing guide. Zmiana oprogramowania za przy użyciu połączenia USB. Changelog / Lista Zmian

Working Tax Credit Child Tax Credit Jobseeker s Allowance

Lecture 18 Review for Exam 1

Instrukcja konfiguracji usługi Wirtualnej Sieci Prywatnej w systemie Mac OSX

Before Adam starts work he needs to know where everything is. Maria shows him around the restaurant.

2017 R. Robert Gajewski: Mathcad Prime 4. Solution of examples Rozwiązania przykładów

Emilka szuka swojej gwiazdy / Emily Climbs (Emily, #2)

Title: On the curl of singular completely continous vector fields in Banach spaces

Zmiany techniczne wprowadzone w wersji Comarch ERP Altum

Bardzo formalny, odbiorca posiada specjalny tytuł, który jest używany zamiast nazwiska

ENGLISH GRAMMAR. reported speech stylistic inversion both, either, neither have & have got

USB firmware changing guide. Zmiana oprogramowania za przy użyciu połączenia USB. Changelog / Lista Zmian

Zdecyduj: Czy to jest rzeczywiście prześladowanie? Czasem coś WYDAJE SIĘ złośliwe, ale wcale takie nie jest.

USB firmware changing guide. Zmiana oprogramowania za przy użyciu połączenia USB. Changelog / Lista Zmian

Wydział Fizyki, Astronomii i Informatyki Stosowanej Uniwersytet Mikołaja Kopernika w Toruniu

FASCINATING RHYTHM. for S.A.B. voices and piano with optional SoundPax and SoundTrax CD* Preview Only. Got mp. E m/g B 7sus/F E m A 9 E 5 E m7/b

Angielski Biznes Ciekawie

First-order logic. Usage. Tautologies, using rst-order logic, relations to natural language

Transkrypt:

Numericl integrtion Recll tht Lgrnge interpoltion of f b f (x) = n f (x i )L n,i (x) i=0 }{{} Lgrnge polnomil P n (x) + f (n+1) (ξ(x)) (n + 1)! n (x x i ) i=0 So we cn tke integrl on both sides: f (x) dx = = n f (x i ) L n,i (x) dx + i=0 n i f (x i ) + E(f ) f (n+1) (ξ(x)) (n + 1)! n (x x i ) dx i=0 i=0 where for i = 0,..., n, i = L n,i (x) dx nd E(f ) = 1 (n + 1)! f (n+1) (ξ(x)) (n + 1)! n (x x i ) dx i=0 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 150

Trpezoidl rule Suppose we know f t x 0 = nd x 1 = b, then P 1 (x) = (x x 1) (x 0 x 1 ) f (x 0) + (x x 0) (x 1 x 0 ) f (x 1) Then tking integrl of f ields f (x) dx = x1 x 0 + 1 [ (x x1 ) (x 0 x 1 ) f (x 0) + (x x ] 0) (x 1 x 0 ) f (x 1) dx x1 x 0 f (ξ(x)) (x x 0 ) (x x 1 ) dx Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 151

Trpezoidl rule Integrl of the first term on the right is strightforwrd. Note tht the second term on the right is x1 x 0 = f (ξ) = f (ξ) f (ξ(x)) (x x 0 ) (x x 1 ) dx x1 [ x 0 x 3 = h3 6 f (ξ) (x x 0 ) (x x 1 ) dx 3 (x 1 + x 0 ) where ξ (x 0, x 1 ) b MVT for integrls nd x + x 0 x 1 x ] x1 x 0 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 15

Trpezoidl rule Therefore, we obtin [ b (x x 1 ) f (x)dx = (x 0 x 1 ) f (x 0) + (x x 0) (x 1 x 0 ) f (x 1) = (x 1 x 0 ) Trpezoidl rule: [ f (x0 ) + f (x 1 ) ] h3 1 f (ξ) ] x1 x 0 h3 1 f (ξ) f (x) dx = h [ f (x0 ) + f (x 1 ) ] h3 1 f (ξ) Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 153

Trpezoidl rule 0 1 his is clled the Trpezoidl rule becuse when f is function with positive vlues b f(x) dx is pproximted b the re in trpezoid, s shown in Figure 4.3. Illustrtion of Trpezoidl rule: f (x) P 1 (x) 1 x 0 x 1 b x ed. M not be copied, scnned, or duplicted, in whole or in prt. Due to electronic rights, some third prt content m be suppressed from the ebook nd/or echpter(s). oes not mterill ffect the overll lerning experience. Cengge Lerning reserves the right to remove dditionl content t n time if subsequent rights restrictions require Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 154

Simpson s rule If we hve vlues of f t x 0 =, x 1 = +b, nd x = b. Then f (x) dx = x x 0 [ (x x1 ) (x x ) (x 0 x 1 ) (x 0 x ) f (x 0) + (x x 0) (x x ) (x 1 x 0 ) (x 1 x ) f (x 1) + (x x 0) (x x 1 ) (x x 0 ) (x x 1 ) f (x )] dx + x (x x 0 ) (x x 1 ) (x x ) x 0 6 f (3) (ξ(x)) dx Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 155

Simpson s Rule Simpson s rule Simpson s rule results from integrting over [, b] the second Lgrnge polnomil with equll-spced With similr nodeside, x 0 = we, cn x = derive b, nd the x 1 Simpson s = + h, where rule: h = (b )/. (See Figure 4.4.) x f (x)dx = h [ f (x0 ) + 4f (x 1 ) + f (x ) ] h5 x 0 3 90 f (4) (ξ) f (x) P (x) x 0 x 1 x b x Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 156

Exmple Exmple (Trpezoidl nd Simpson s rules for integrtion) Compre Trpezoidl nd Simpson s rules on 0 () x (b) x 4 (c) (x + 1) 1 (d) 1 + x (e) sin x (f) e x f (x) dx where f is Solution. Appl the the formuls respectivel to get: Problem () (b) (c) (d) (e) (f) f (x) x x 4 (x + 1) 1 1 + x sin x e x Exct vlue.667 6.400 1.099.958 1.416 6.389 Trpezoidl 4.000 16.000 1.333 3.36 0.909 8.389 Simpson s.667 6.667 1.111.964 1.45 6.41 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 157

Newton-Cotes formul We cn follow the sme ide to get higher-order pproximtions, clled the Netwon-Cotes formuls. For n = 3 where ξ (x 0, x 3 ): x3 x 0 f (x) dx = 3h 8 [ f (x0 ) + 3f (x 1 ) + 3f (x ) + f (x 3 ) ] 3h5 80 f (4) (ξ) For n = 4 where ξ (x 0, x 4 ): x4 x 0 f (x) dx = h 45 [ 7f (x0 ) + 3f (x 1 ) + 1f (x ) + 3f (x 3 ) + 7f (x 4 ) ] 8h7 945 f (6) (ξ) Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 158

Closed Newton-Cotes Formuls Composite numericl integrtion Figure 4.5 The (n+ 1)-point closed Newton-Cotes formuluses nodes x i = x 0 + ih, for i = 0, 1,..., n, where x 0 =, x n = b nd h = (b )/n. (See Figure 4.5.) It is clled closed becuse the endpoints of the closed intervl [, b] re included s nodes. Problem with Newton-Cotes rule for high degree is oscilltions. = f (x) = P n (x) x 0 x 1 x x n 1 x n b x Insted, we cn pproximte the integrl piecewisel. The formul ssumes the form f(x) dx n i f(x i ), Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte i= 0 Universit 159

Composite midpoint rule Let x 1 =, x 0, x 1 x, 0.. x 1., x n, x n+1 = x j 1 b bex j uniform x n 1 b prtition x n xof [, b] with h = b n+. Then we obtin the composite midpoint rule: Figure 4.9 n/ For the Composite Midpoint rule, n must gin be even. (See Figure 4.9.) f (x) dx = h j=0 f ( ) b x j + 6 h f (µ) f (x) x 1 x 0 x 1 x j 1 x j x j 1 x n 1 x n b x n 1 x Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 160

Composite trpezoidl rule Figure 4.8 Let x 0 =, x 1,..., x n = b be uniform prtition of [, b] with h = b n. Then we obtin the composite Trpezoidl rule: f (x) dx = h n 1 f () + f ( ) x j + f (b) b 1 h f (µ) j=1 4.4 Composite Numericl Integrtion 07 f (x) x 0 x 1 x j 1 x j x n 1 b x n x Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 161

Composite Simpson s 1 rule Let x 0, x 1,..., x n (n even) be uniform prtition of [, b]. Then The error for this pproximtion hs been reduced to 0.01807. ppl Simpson s rule on [x 0, x ], [x, x 4 ],..., totl of n such intervls. Then we obtin the composite Simpson s b rule: f (x)dx = h 3 + 6 e + 4e + e + 6 e + 4e + e ( = e 0 + 4e 1/ + e + 4e 3/ + e + 4e 5/ + e 3 + 4e 7/ + e 4) 6 = 53.616. To generlize this procedure for n rbitrr integrl integer n. Subdivide the intervl (n/) 1 [, b] into n subintervls, n/ nd ppl Simpson s rule on Figure 4.7 f () + ech consecutive pir of subintervls. (See Figure 4.7.) f (x j ) + 4 f (x j 1 ) + f (b) b 180 h4 f (4) (µ) j=1 j=1 f(x) dx, choose n even f (x) x 0 x x j x j 1 x j b x n x ngge Lerning. All Rights Numericl Reserved. M Anlsis not be copied, I scnned, Xiojing or duplicted, Ye, in Mth whole or in & prt. Stt, Due to Georgi electronic rights, Stte some Universit third prt content m be suppressed from the ebook nd/or echpter(s). 16

Guss qudrture f (x) f (x) f (x) 4.7 Gussin Qudrture 9 Previousl we chose points (nodes) with fixed gps. Wht if we re llowed to to Figure x 1 choose 4.15 x bpoints x x 0, x 1..., x n x nd b x evlute x 1 f there? x b x f (x) The Trpezoidl rule pproximtes the integrl of the function b integrting the liner f (x) function tht joins the endpoints off (x) the grph of the function. But this is not likel the best line for pproximting the integrl. Lines such s those shown in Figure 4.16 would likel give much better pproximtions in most cses. Figure 4.16 x 1 x b x x 1 x b x x 1 x b x f (x) The Trpezoidl rule pproximtes the integrl of the function f b (x) integrting the liner function tht joins the endpoints of f (x) the grph of the function. But this is not likel the best line for pproximting the integrl. Lines such s those shown in Figure 4.16 would likel give much better pproximtions in most cses. Figure x 1 4.16 x b x x 1 x b x x 1 b x x f (x) Gussin qudrture chooses the points for evlution in n optiml, rther thn equllspced, w. The nodes x 1, x,..., x n in the intervl [, b] nd coefficients f (x) c 1, c,..., c n, re Numericl Guss demonstrted Anlsis I his method Xiojingchosen Ye, Mth to minimize & Stt, the Georgi expected Stte error f (x) Universit obtined in the pproximtion 163

Guss qudrture Guss qudrture tries to determine x 1,..., x n nd c 1,..., c n s.t. f (x) dx n c i f (x i ) i=1 Conceptull, since we hve n prmeters, i.e., c i, x i for i = 1,..., n, we expect to get = if f (x) is polnomil of degree n 1. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 164

Guss qudrture Let s first tr the cse with intervl [ 1, 1] nd two points x 1, x [ 1, 1]. Then we need to find x 1, x, c 1, c such tht 1 1 f (x)dx c 1 f (x 1 ) + c f (x ) nd = holds for ll polnomils of degree 3. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 165

Guss qudrture We first note ( 0 + 1 x + x + 3 x 3) dx = 0 1 dx+ 1 x dx+ x dx+ 3 x 3 dx Then we need x 1, x, c 1, c s.t. 1 1 f (x) dx = c 1f (x 1 ) + c f (x ) for f (x) = 1, x, x, nd x 3 : c 1 1 + c 1 = c 1 x 1 + c x = c 1 x1 + c x = c 1 x1 3 + c x 3 = 1 1 1 1 1 1 1 1 1 dx =, x dx = 0 x dx = 3, x 3 dx = 0 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 166

Guss qudrture Solve the sstem of four equtions to obtin x 1, x, c 1, c : c 1 = 1, c = 1, x 1 = So the pproximtion is 1 1 f (x) dx f ( 3 3, nd x = 3 3 ) + f ( ) 3 3 3 3 which is exct for ll polnomils of degree 3. This point nd weight selection is clled Guss qudrture. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 167

Legendre polnomils To obtin Guss qudrture for lrger n, we need Legendre polnomils {P n : n = 0, 1,... }: 1. All P n re monic (leding coefficient =1). 1 1 P(x)P n (x) dx = 0 for ll polnomil P of degree less thn n. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 168

Legendre polnomils The first five Legendre polnomils: P 0 (x) = 1 P 1 (x) = x P (x) = x 1 3 P 3 (x) = x 3 3 5 x P 4 (x) = x 4 6 7 x + 3 35 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 169

Guss qudrture nd Legendre polnomil Theorem (Obtin Guss qudrture b Legendre pol.) Suppose x 1,..., x n re the roots of the nth Legendre polnomil P n (x), nd define c i = 1 1 n j=1 j i x x j x i x j dx If P(x) is n polnomil of degree less thn n, then 1 1 P(x)dx = n c i P (x i ) i=1 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 170

Guss qudrture n Roots r n,i Coefficients c n,i 0.57735069 1.0000000000-0.57735069 1.0000000000 3 0.774596669 0.5555555556 0.0000000000 0.8888888889-0.774596669 0.5555555556 4 0.8611363116 0.3478548451 0.3399810436 0.651451549-0.3399810436 0.651451549-0.8611363116 0.3478548451 5 0.9061798459 0.36968850 0.5384693101 0.478686705 0.0000000000 0.5688888889-0.5384693101 0.478686705-0.9061798459 0.36968850 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 171

Exmple Exmple (Guss qudrture) Approximte 1 1 ex cos x dx using Guss qudrture with n = 3. Solution. We need to use the roots of Legendre polnomil nd coefficient vlues for n = 3: n Roots r n,i Coefficients c n,i 3 0.774596669 0.5555555556 0.0000000000 0.8888888889-0.774596669 0.5555555556 1 1 e x cos x dx 0.5e 0.7745969 cos(0.77459669) + 0.8 cos(0) + 0.5e 0.7745969 cos( 0.77459669) =1.9333904 True vlue is 1 1 ex cos x dx = 1.933414. Our error is 3. 10 5. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 17

Guss qudrture on rbitrr intervl So fr the Guss qudrture is onl considered on [ 1, 1]. To find Guss qudrture on rbitrr x [, b], just do chnge of vrible: t = x b b x = 1 [(b )t + + b] Then t [ 1, 1] nd the integrl is f (x) dx = 1 1 f ( ) (b )t + (b + ) (b ) dt Then ppl Guss qudrture to the right side. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 173

The techniques discussed in the previous sections cn be modified for use in the pproximtion of multiple integrls. Consider the double integrl f(x, ) da, Multiple integrls Now we consider multiple integrl where R = {(x, ) x b, b c d d }, for some constnts, b, c, nd d, is rectngulr region in the plne. (See Figure 4.18.) c Figure 4.18 R f (x, ) d dx z z f (x, ) c d x b R Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 174

This pproximtion is of order O (b )(d c) (b ) + (d c). Figure 4.19 shows grid with the number of functionl evlutions t ech of the nodes used in the pproximtion. Multiple integrls.19 First consider grid on the domin [, b] [c, d]: 1 d (c d) c 1 1 4 1 1 1 ( b) b x Here k = d c nd h = b. ll Rights Reserved. M not be copied, scnned, or duplicted, in whole or in prt. Due to electronic rights, some third prt content m be suppressed from the ebook nd/or echpter(s). ressed content does not mterill ffect the overll lerning experience. Cengge Lerning reserves the right to remove dditionl content t n time if subsequent rights restrictions require it. Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 175

Multiple integrls We first pproximte the inner integrl using composite Trpezoidl rule: d c f (x, ) d = c+k c f (x, ) d + d c+k f (x, ) d k (f (x, c) + f (x, c + k)) + k (f (x, c + k) + f (x, d)) = k (f (x, c) + f (x, c + k) + f (x, d)) =: g(x) Then pproximte the outer integrl: g(x) dx = h (g() + g( + h) + g(b)) Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 176

Copright 010 Cengge Lerning. Numericl All Rights Reserved. Anlsis M not I be copied, Xiojing scnned, or Ye, duplicted, Mth in whole & or in Stt, prt. Due Georgi to electronic rights, Stte some third Universit prt content m be suppressed from the ebook nd/or echpter(s). 177 ditoril review hs deemed tht n suppressed content does not mterill ffect the overll lerning experience. Cengge Lerning reserves the right to remove dditionl content t n time if subsequent rights restrictions require it. Multiple integrls + Combine the two to obtin: ( d Figure 4.19 c = + f(b, c) + f b, + f(b, d) 4 4 [ (b )(d c) f(, c) + f(, d) + f(b, c) + f(b, d) 16 ( ( ) ( ) ( + b + b f, c + f, d + f, c + d ) ( )) ( )] + b ) { (b+ f )(d b, c + d c) + 4f f (x, ) d dx = f (, c) +, c + d f (, d) + f (b, c) + f (b, d) 16 [ ( ) ( ) ( + b + b + f, c + f, d + f, c + d ) ( + f b, c + d ) ] ( + b + 4f, c + d ) } This pproximtion is of order O ( (b )(d c) [ (b ) + (d c) ]). Figure 4.19 shows grid with the number of functionl evlutions t ech of the nodes used in the pproximtion. d 1 (c d) c 1 1 4 1 1 1 ( b) b x

the nodes (x i, j ), where i = 0, 1,, 3, 4 nd j = 0, 1,. It lso shows the coefficients w i,j of f(x Multiple i, i ) = ln(xintegrls i + i ) the sum tht gives the Composite Simpson s rule pproximtion to the integrl. We cn lso consider 4 grid on the domin [, b] [c, d]: 1.50 1 4 4 1 1.5 4 16 8 16 4 1.00 1 4 4 1 1.40 1.55 1.70 1.85.00 x Here k = d c 4 nd h = b The pproximtion is..0 1.5 (0.15)(0.5) 4 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 178

Guss qudrture for non-rectngulr region We cn lso use Guss qudrture for non-rectngulr region: d(x) Figure 4.1 c(x) c(x) we will use the bsic Simpson s rule to integrte with respect to both vribles. The step size for the vrible x is h = (b )/, but the step size for vries with x (see Figure 4.1) nd is written k(x) = f (x, ) d dx d(x) c(x). z z f (x, ) d() d(b) d(x) A(x) k() c(b) c() k( h) c(x) h b k(b) x x b c(x) R d(x) () Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 179 (b)

Composite Simpson s rule on non-rectngulr region Now we consider multiple integrls on non-rectngulr regions: d(x) c(x) f (x, ) d dx For ech integrl set k(x) = d(x) c(x), then d(x) c(x) k(x) f (x, ) d dx [f (x, c(x)) + 4f (x, c(x) + k(x)) + f (x, d(x))] dx 3 h { k() [f (, c()) + 4f (, c() + k()) + f (, d())] 3 3 4k( + h) + [f ( + h, c( + h)) + 4f ( + h, c( + h) 3 + k( + h)) + f ( + h, d( + h))] + k(b) [ ]} f (b, c(b)) + 4f (b, c(b) + k(b)) + f (b, d(b)) 3 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 180

Guss qudrture for non-rectngulr region We cn lso use Guss qudrture for non-rectngulr region: d(x) c(x) f (x, ) d dx For ech x [, b], trnsform [c(x), d(x)] into vrible t in [ 1, 1]: ( ) (d(x) c(x))t + d(x) + c(x) f (x, ) = f x, d(x) c(x) d = dt Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 181

Guss qudrture for non-rectngulr region So the inner integrl cn be pproximted b Guss qudrture: d(x) c(x) f (x, ) d = d(x) c(x) d(x) c(x) =: g(x) 1 1 f ( x, n c n,j f j=1 ) (d(x) c(x))t + d(x) + c(x) dt ( x, (d(x) c(x))r n,j + d(x) + c(x) Then we ppl Guss qudrture to the outer integrl: ) d(x) c(x) f (x, ) d dx g(x) dx = ( ) (b )t + (b + ) (b ) g dt 1 m ( ) (b )rm,i + (b + ) (b ) c m,i g 1 i=1 Numericl Anlsis I Xiojing Ye, Mth & Stt, Georgi Stte Universit 18