mit18_03s10_chapter_20

Upload: jackiechieng

Post on 05-Apr-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 MIT18_03S10_chapter_20

    1/6

    10120. Laplacetransformtechnique: coverup

    I

    want

    to

    show

    you

    some

    practical

    tricks

    which

    will

    help

    you

    to

    find the inverse Laplace transform of a rational function. These arerefinementsonthemethodofpartialfractionswhichyoustudiedwhenyou learnedhowto integraterationalfunctions. Someofthiswillusecomplexnumbers.20.1. Simplecase. First,letsdoaneasycase:

    1F(s) =

    2 .s 4Tobegin,factorthedenominator,andwrite

    1 a b(1) F(s) = = +(s2)(s+2) s2 s+ 2forasyetunknownconstantsaandb. Onewaytoproceed istocrossmultiply and collect terms in the numerator. That is fine but thefollowingismorefun.

    To find a,firstmultiply through by the correspondingdenominator,(s2)inthiscase. Youget

    1=a+ (s2)(otherterms),

    s+ 2b

    inwhichtheotherterms(namely, )donthaveafactorof(s2)s+ 2inthedenominator. Thensets=2:

    1=a+(22)(otherterms)=a

    2 + 2sincethesecondtermvanishes. Soa= 1/4. Inexactlythesameway,youcanmultiply(1)throughbys+2andthensets=2,tofind

    1=b

    22orb=1/4. Thus

    1/4 1/4F(s) = .

    s+ 2s2

    Thetablesthenshowthatf(t)=(1/4)(e2te2t).

    This approach to partial fractions has been called the cover-upmethod; you cover up the denominators of the terms you wish tocompute. You can do it without writing anything down;just cover

  • 7/31/2019 MIT18_03S10_chapter_20

    2/6

    102up the denominators and write down the answer. It works well butdoes not completely handle the case in which the denominator has arepeatedroot.20.2. Repeatedroots. Forexample lookat

    sF(s) = .

    s2 + 2s+ 1Thedenominatorfactorsas(s + 1)2. Wewanttofindaandbsuchthat

    s a bF(s) = = + .

    (s+1)2 s+ 1 (s+1)2Wecanbegintousethecoverupmethod: multiply throughby(s+1)2and

    set

    s=

    1:

    The

    left

    hand

    side

    is

    just

    1;

    the

    first

    term

    vanishes;

    andthesecondtermisb: sob=1. Wecantgetathisway,though.Onewaytofindaistosetstosomeothervalue. Anyothervaluewilldo, and we might as well make our arithmetic as simple as possible.Letstakes=0: thenwehave

    a+10 =

    1 1soa=1:

    1 1F(s) =

    s+ 1(s+1)2.Nowthetablesshow

    f(t) =ettet.20.3. Completing thesquare. Suppose

    1F(s) = .

    s2 + 2s+ 2The first part of the method here is to complete the square in thedenominator,andrewrite thenumerator inthesameterms:

    s a(s+1)+b= .

    s2 + 2s+ 2 (s+1)2 + 1Thisworkswitha=1andb=1:

    F(s) = (s+1)1.(s+1)2 + 1

    Now the s-shift rule applies, since F(s) is written in terms of sa(whereherea=1). Thesecondpartofthismethodgivesyouawaytousethes-shiftrulewithoutgettingtooconfused. Youshouldinvent

  • 7/31/2019 MIT18_03S10_chapter_20

    3/6

    103a newfunction namesay G(s)and use it to denote the functionsuchthatthatF(s) =G(sa). Thus,here,

    G(s) = s1.s2 + 1

    Nowthes-shiftrulesaysthatifg(t)G(s)thenetg(t)G(s+1)=F(s),which istosaythatf(t) =etg(t). Thetablesgive

    g(t)=costsintso

    f(t) =et(costsint).20.4. Complex coverup. Now lets take an example in which thequadratic

    factor

    does

    not

    occur

    alone

    in

    the

    denominator:

    say

    1

    F(s) = .s3 +s22

    Thedenominator factorsass3 +s22 = (s1)(s2 + 2s+2). Intheexampleabovewelearnedthatthefactors2+ 2s+ 2shouldbehandledbycompletingthesquareandgroupingthe(s+1)inthenumerator:

    1 a b(s+1)+cF(s) = = + .

    (s1)((s+1)2 +1) s1 (s+1)2 + 1Findajust asbefore: multiply throughby s1andthen set s=1,to get a = 1/5. To find b and c, multiply through by the quadraticfactor(s+1)2+ 1andthensetsequaltoarootofthatfactor. Havingalreadycompletedthesquare, itseasytofindaroot: (s+1)2 =1,sos+ 1 =iforexample,sos=1 +i. Weget:

    1(1 +i)1 =b((1 +i)+1)+c

    or,rationalizingthedenominator,2i

    =c+bi5

    Sincewewantbandcreal,wemusthavec=2/5andb=1/5:

    1 1 (s+1)+2F(s) = .5 s1(s+1)2 + 1

    Wereinpositiontoappealtothes-shiftrule,usingthetricksdescribedin20.3,andfind

    1 f(t) = etet(sint+2cost) .

    5

  • 7/31/2019 MIT18_03S10_chapter_20

    4/6

    10420.5. Completepartialfractions. Thereisanotherwaytodealwithquadraticfactors:justfactorthemoverthecomplexnumbersandusethe coverup method in its original form as in Section 20.1. I dontrecommend using this in practice, but its interesting to see how itworksout,andwewillusetheseideasinSection22. Usingtheexample

    1F(s) =

    s3 +s22again,wecanfindcomplexconstantsa,b,csuchthat

    a b c(2) F(s) = + +

    s1 s(1 +i) s(1i)Expectthatawillbereal,andthatbandcwillbecomplexconjugatesofeachother.

    Find ajust as before; a = 1/5. To find b, do the same: multiplythroughbys(1 +i)toget

    1(s1)(s(1i)) =b+ (s(1 +i))(otherterms)

    andthensets=1 +itosee1

    =b(2 +i)(2i)

    orb= 1/(24i) = (1 + 2i)/10. Thecoefficientccanbecomputedsimilarly. Alternatively, you can use the fact that the two last termsin(2)mustbecomplexconjugatetoeachother(inorderforthewhole

    expressiontocomeoutreal)andsodiscoverthatc=b= (12i)/10:F(s) = 1/5 + (1 + 2i)/10+(12i)/10.

    s1 s(1 +i) s(1i)Thenumeratorsa,b,andc,inthisexpressionarecalledtheresidues

    ofthepolesofF(s);see22.1below.Its perfectly simple to find the inverse Laplace transforms of the

    termshere:f(t) = 1 t +1 + 2i (1+i)t +12i (1i)te e e .

    5 10 10Thelasttwotermsarecomplexconjugatesofeachother,sotheirsumistwicetherealpartofeach,namely,

    t te e2 Re((1 + 2i)(cost+isint)) = (cost2sint).

    10 5Wewindupwiththesamefunctionf(t)asbefore.

  • 7/31/2019 MIT18_03S10_chapter_20

    5/6

    105ListofpropertiesoftheLaplace transform1.L

    islinear:

    af(t) +

    bg(t)

    aF

    (s) +

    bG(s).

    2. F(s)essentiallydeterminesf(t).3. s-shifttheorem: Iff(t)F(s),theneatf(t)F(sa).4. t-shifttheorem: Iff(t)F(s),thenfa(t)easF(s),where

    fa(t) = f(ta) ift > a .0 ift < a5. s-derivativetheorem: Iff(t)F(s),thentf(t)F(s).6. t-derivativetheorem: Iff(t)F(s),thenf(t)sF(s)f(0+)wheref(t)iscontinuousfort >0andthenotationf(t)indicatestheordinaryderivativeoff(t).7. Iff(t)F(s)andg(t)G(s),thenf(t) g(t)F(s)G(s).8. (t)1.

  • 7/31/2019 MIT18_03S10_chapter_20

    6/6

    MIT OpenCourseWarehttp://ocw.mit.edu

    18.03 Differential Equations Spring 2010

    For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

    http://ocw.mit.edu/http://ocw.mit.edu/termshttp://ocw.mit.edu/termshttp://ocw.mit.edu/termshttp://ocw.mit.edu/