Return this book on or before the Latest Date stamped below. University of Illinois Library SEP 7 m m JUN 18 Rl \a FEB 12 1973 •lANy’O r| ■^V/V g C'O % L161 — H4l 51 5 , 58 V88^ :E 1919 Digitized by the Internet Archive in 2017 with funding from University of Illinois Urbana-Champaign Alternates https://archive.org/details/lecturesonintegrOOvolt 1 1-0 di. h’IRST LECTURE' Solution of equations by passing to the limit from Approximations.; The problems that arise in algebra are of a finite character. These meet their first generalization in the study of infinite series in which the notion of sum is generalized from the finite to the infinite case.; We meet a similar generalization in the Riemann definition of integral.’ In this well*“known case an interval of integration^ from to Xn j is divided into ri sub— interval s^ the interval from Xj^ having a lent^th ,* 2 ^. The function value of the integrand is taken at some point in the interval^ and the sum formed ^ 1,. C, n. The limit of this sum if it exists^, for any method of subdivision into n parts as n increases indefinitely^ the length of the maximum 'hj decreasing indefinitely with 1 ^ n, is the Riemann integral.' Likewise in Jefinin^ a derivative we pass from a finite case to an infinite case.; Usually the derivative is the inverse of the integral defined above^ but it was pointed out by Dini that there were functions which had derivatives but such that the derivative was not inte^rable in the Riemann sense,' On the basis of his own and also investigations of Sorely Lebes^ue «jave a definition of intCiJral which was much 1 reader in extent^ and took care of some of these exceptional cases ; The definition of derivative is usually ^iven in the form F( X* h)-F{ x) This can be determined approximately by taking \ as small as possible 4U ^iifilaa^o %<,i isnoiifEupa lo noiJuio2 S«floli£(£iixo'iaqA atot^ iiiail ' 3 'i ‘'\ H, *5^iffx3 £ ![o 97K ut^ajda ni &8ti& ^utrl^ eojaldoiq S :Wji .SaiTae «-‘i(.Mni to ,iuJ8 ni noxias ileisnss Xanii nifux,,, 9.-IJ oJ sXini^ sdj „otl tosilaianss si lua }o notion erfj iioHw n ■■•j ■■ ' li ^ nna^nsifl srIJ ni noiissiXe-isnss isliaia a Jssn sW i.saat aiinilni RV, io Xavisini na asao n»onH-XX,» airii nl ;.Xeis,s.ini io poilinilal arii ..sla.naini-dua „ oini iabivit^si oi ,x »on1 .noixan^ain) »tlav noixonai aPT .jP rfx.nnj , jx oX ,.jx noiT Xa.neXa^ fcns ,i3 Jaanaini arix „i xnioo aaoe Xa naSaX si tn.n^^fnt ,rtx Ir b^OIlO^ nog 8(iy =J noiaivifcdua tortXaa yna no! ,.alaixa xi tl Ln ,i,1x i, XiaixVl »aaixax. arfx lo rixpnaX arfx .vXaXiniiafcni easaaaoni n sa aXnac x oJnX 'V ^ -.flansaxni nnaaaiR anx ^i V i ylaxlniXatni Jnisaanaafa asaa aXini.i a »oai aaaq a« ayfXavlnab *■ JninilaX ni aeiwaJtU , ..Xi 50 asnavni aPx ai avixavinat a,.X yllanaO ,aaaa aXinlXxfi na oX oxanx XartX inxQ yp x„o iaXnioq aaw ii' XoX- ..avoXa ianilat Xansaxni _oviXaviaai aXX Xa.iX Pana Xnd saviXavinai fad daid. anoiXonni i ’■ iild -o Sid 10 sisad.adX „0 ;.aanas nnaoaiXi adx ni aXdaa.axni x\n' ..a 50 noixinilat a avas anvsadaj .,lano8 1o snoiXa sixsayn i .^'xa .b„^ ■io a.os 50 anaa dooX ina .,x„,xxa ni nataon - dono san doidw lanyaLi lsnoiSo 90 X 3 0 B 9 fiJ _,x.dJX adX nl navxi ylXsuay ai aylXa»inal >o noiX-iniXai edl ' tx ♦x i'** I i&Il. " Vd Vlalaaixodooa iani.-.axoi ,d n«, though not «ctuaUy reaching the llmitini case of '^-o,, Such approx- imations are quite useful in aopZiei mathematic a,! and for many such purposes satisfactory^ If now we consider the simple differential equation dy/dx • f(xyjf with the usual conditions on the functional Cauchy showed that an lotetjtral always exi'sts.j The method generally used for the mechanical integration of this equation is an apDroxiiDation!,i which pushsl to the limiting ^ase shows the eadstence of the function called the integrsl!.J This can be done graphically by strating from a given point acoi,i yo-r^ calculating by the equation the slope of an integral curve at this point.,1 then continuing in the corresponding direction for a short distance tc which point a new slope is calculated*] This process repeated for short enough intervals will lead to a close approximation to the integral curve.! In symbols we may Indicate this process as followsi, W(- ^ •«i-4 from which we have a set of equations of the form y{xi)-y(xi^lhf(xi^i,yi^l)(xi- If the successive values are substituted in the equations we arrive finally at the value of y(xn) in terms of a sum of products of the form *e may extend this method to aoply also to the oase of systems of equations.) For instance let iy/dx - dx^'iix ” •t,i(x)y * Proceeding to exoress the approximate results for n different values of *, and multiolyini out the results we have yn® (l + .’Zn yThe passage f ro’n Zr._i to Un j Zn is by rceans nf the substitution **12 " ■ ... # ,"'To pass then frofu \io Zq to jt,>, Zn woul .1 require a succession of these substitutions • 3 2 3 1 . cr -n ^ We nay write 3 n in the forno 1 +’:^^ where is the matrix of the u's ! alone^ and the product of the succession of substitutions takes the j f orm . 1 ■*■ t . vy ( x • M ) + m '2 ^ yiy ( ;c • j f X y j * • • • iHhe summation bein^ from 1 to n. If we pas:s to the limit, we ar^'ive at inte->'rals of a substitution 1 + f ( x>) ( ) dx ^ r r t f Jc I ) f X 2 ) ( ) ti X idx 2 + * • • [See Volterra: Leccns sur les functions des li.>'ne 3 ^ 00 36 — 42 ] f' Continuing our develoonrent^ let us notice next the (■^equation d^^^dx = f(x)u(x), of whic'. we may write the integral as an lithe solution of an inte^'^al equation.^ namely 1 h(x) = + c. |lr a more general fo'^m we would have to consider as linear integral ■equations the forms ^ix) = f{x)-( x)d^ the Volterra equation of second kind^ x)= f Q^(t)F(hjx) di. the Volterra equation of first Kind^ x) = f(x)~\^ 0 ^>{^-)^'i^,x)dF the Preihclm equation of the seccn 1 kind, L '^{x)=\f (i^^f{t)F(^.x) dh the Fredholm equation of the first kind. ! To these we may aooly the same metho ds of constructiru' a solution by j the methods of aoproximation under discussion^ viz. those which i consider the values of the quantities at n ooinl, in the region under u consideration^ passing ttience to the region by lettin.? n becoiiie ^'infinite. An exaiole anicnu tne eanliest integral equations is that due to a problem of Abells. The pro b lei; is this. deter vine that curve in a vertical plane iowti which a xovin.? 00 dy would pass iin such a way that it would reach the lowest point in a time ^iven by th« function whore '*1 is the initial heij^htui the initial velocity bein? zero.J If wOh) i's constant for all heiafhts the carve is the tautochrone,\ The integral equation which has to be solved is / x) " To ^ ✓ ( Solution;, uix)*’ ^ A dx /Kx-hJ y‘u(x) is the equation of the curve. | 'ISee Bocherij.t Linear Intesfral Equationstji Volterral,' L'econs sur les equations inte|rale9.j] A generalization of this is the eqaati<}n f(x) * f xj { dB, where 0 and second derivative as to y is •~^f(y). IS in !'• Nr '•j TEIUD LECTURE t, The process utilized to solve the equation last considered is an example of a very general process by which an integral .equation with a ?iven nucle.us may be reduced to anoL.{\er with a simpler nucleus.; In fact if we have the equation r(xJ = fo^ si,{OF(lx)dl and multiply on both sides by a ^iven function ^ (xy) then integrate as to X, we have (xy)f(x)dx=fo y§(xy)dxf o^^>{OF{tx) dt = f o^^fiOdtf ^^F(E.x)^( xyJdx = /’o^^y(e) ^{ly)di We may set the result of the composition Ff=^(xy), thus arriving at a new equation with a new nucleus.' An example is the previous case where the function equation with nucleus [ 1 oi ( jc-£. ) into one with nucleus -(][/-£.).' We now purpose to find a function ^ which will convert the nucleus \oi( x-i.)*0, where C is the Kuler constant We start with the formula r ' X dt (y-xj a + b— 1 Tu re r(a*e) The meabers of this equation are icultipliei by • the parts on the left beini separated • We then repeat the process used before of differentiation as to u and integration as to b form to . If 1 t \ 9 - T we reoresent by P the expression B B pf, then we have r(a+e) Nowseta=l B=landwehavethedesiredresult fx^[\oiU-x)-V'l/Tl +^]P(£f^;cia=-(.v-Jc) lln these equations if we do not take a=l we may ccaiDlicate the nucleus lito a considerable extent,; 3y further differentiation as to a and in- |te?ration as to b it is possible to arrive at expressions involving the square of the lo^arithiu^ and even higher powers and polynorcials in the lo^.; This leads us to inquire .what the nost general integral equation may be like. One case studies by Lalesco (Equations Inte»?rales p 127 et seq) i s lj(xj = f This is not linear^ and indeed F may be quite 4eneral Another general form would be ^>(x)=^y(x)+fo "-F i{E.x)di.*f o^'f 0 "-dS. i,di. s{e,j.x ^x) j.) yii s) ^ We need not multiply these special cases but pass at once to the consideration of the most general case that can occur.; We find necessary here the general conception of function of a line, and that the most ieneral integral equation takes the form ^)(x)^ ^ [yit), x] the riJht hand side representing a function of all the values 4iven by the function yU) where ^ takes every value from 0 to 1, and a parameter ;c, which may also take every value from 0 to 1.; This equation may now be studies as an equation which is a limiting case of an infinity of equations depending upon an infinity of variables.; See Volterra, Archiv der Math/ und Phys.; (3)23(1914/5) Sitzber.; Berlin Math Ges.; 13(1914) pp 130-150. ; ' 0 ^ ■ /fr U.. -10- ' V « 4 !*• Let u(x) define a curve^^ then any set of numbers determined by some process of determination such that for each function u(x) there is one number, ‘ are the function values of a function of a line.' The argument in this case of the function is the line in a certain sense, or we may say is the totality of all the function values of yix) for every vaOue of jjc from 0 to 1.' A simole example is the area under the curve,' another is the length of the curve.; In physics we have many such cases,- as for example the ootenytial at a ^iven point in space due to a loop carrying an electric current is a function of the loop.; We may study such functions by our method of development depending on a set of n selected points and the corresponding ordinates,' so that we would have a function ^ ( Ij u Us j ' '' ■' Afterwards we determine the limit of the expressions as n approaches infinity.; We may also have a set of such function ^ y xj y 2/ '.Vn ) “ z 1 ^ 9^y X} y 2 j * i'n ) - z 2 r • ? • r • ♦ • f ^ y Xf y 2 j ‘ ‘yn Zn If we attach the values z to the points alon^ the line we may look upon them as values of a function z( x) , and must then con adder that in place of the index of the F we must introduce jc as a parameteif. we have as the limitini? case FlyiOjx] - z(x) • • where the left side is a function of a line depending upon a . parameter x. This is the most general form of an integral equation ** ( of the Fredholm type,- but this includes all the types.) ^ In order to solve an equation of this form we expand it || in a form similar to a Taylor Series ; In fact the exoansic -i is a ft generalization of Taylor's series from the case of variables to > that of an infinity of variables.' We exoand the function a 1 14 . he in the form . ^ 'J ^ z(x)=y( x) +f 0 o^f o F git\^ gx)'dCl'iE,.y+" - It should be noticed that if we stoo with the first Integral we' have the linear integral equation.* A resolvent function;bis founds and we then have yi x) = z(x) + f dti + f ^zilx^gx) citidts-^' See Volterra Fon' tions des li^nes p.; 69 et seq.; FOURTH LKCTURE Inte^ ro-di f f erenti al equations * The equations contain the unknown function not only unier the integration si^n but also differentiated .partially or totally with respect to the arguments it contains * As examples df* dx'-^ to which ^ives the motion of an elastic cor 1 when hereiitary effects are considered.,- that is when the motion changes the character of the elasticity progressively.,* oroduciriij an effect similar to hysteresis in magnetism.’ See Volterra Rend.; Ac | Real ; lei Lincei 191^.,'/ Fonctions des li^nes p 97.; Another example from physics is t £.kiLl = n(t)*fo\'('^)F('it)d^ dt'^ to which the solution of the precedinu is reduced.; From analysis we easily find examples.; We shall consider here only t that one arising in the solution-of the problem of finding all the functions oermutable with a ^iven function.' We recall the definition of permutability (xOfO-uidf- ks an intpolactorv Raamole which will suiiest the general aethol of attaol^ let as take that is we oropose to fini the functions oermutablewith v , \ ^ ^(xy) = ^(xOdt I If we differentiate as to k we have at once if Differentiatini as to y we have also 3f 31/ i ^ »U these we eliminate the funotioh and have a partial 'i - 12 - for whence 4 / * F(y-x) " lere F is arbitrary but differentiable.; These functions f urt hertnox e< ia1*e a>Jl oermutable with one another,' and in this sense foriE a ^rouo. his is called for physical reasons the s^roup of closed cycle, (Volterra Fonctions des li^nes Chap.'VIl),; The auestions as to whether it is true that the iroup of functions peirmutab le with 3 ^iven function are therefore permutable with one another has been answered affirma- tively by Vessio’t^ C.;R.; 1912 p 682.; In this i^eneral case we make use of a transformation upon the whole sis follows,' Let F(xy) be one of the 5 roup.; Transform this by Fi(xy) * f( x)F( xy)( y) and set dxi-=dx^r(x)^>(x) x=K(xi.) Xj,= ii(x) y^^(yi) Transform now $ ( xy) in the same way i?ivin;? $x( xy)=‘f(x)'^( xy)^( y) Then we have ^ i{xy)= f ^ x) dlir f ^ J r ( x) F ( xt)^>it) f it) Hty) ^^( y)dti But from the value of d^i we have at once that this = r ^^f(x)F{xt)^ity)w(y^dt=f(x)Fi^>(y) X 1 That is ^^J,(xy)^r(x)^{ xy)^>( y) It is evident no» that if is the result of a oerautable oo^nosition By a orooer choice of functions we can reduce any function P from the Jeneral font to a fore such that P(xx)^\, F xy) = dP^ ox, P.( xu)-dF/Su and Fjxx) =■ 0 F,{xx) '0.; „ . We now derive the condition of pennutability of a function .tth a iiven function as the solution of an intejro-iif ferential equation.; Let ^ dif f erentiaf e as to x and as to y 4 ivin^ iof which integral epuaUcn.- i as unknown we wiU let T. he the resolvent. ■y 1 ' X of which we will let f, be the resolvent • .Therefore we will have froo these two < 1 . X A ♦ 4 .Jf, Hl3n by subtraction we ellfflinate the unknown I 'quatiop'.for f' ‘ and have an ae nte^ratin? the ri^ht hand side by parts we free the function ^ from the differentiation as to and therefore have a froa The solution given here and the final treatment are found in Foncticns de lignes pp 61 et seq«,| Let FIFTH lecture; Integror-dif ferential equations.] We found the general form of integral equation by passing to the llsi't from a set of n equations to be f 1 [y(^) x] \^z(x) 1*1 Similarly we arrive at an integrordlf ferential equationby the consideration of a sot of simultaneous dif fer??r44al equations. jy„ z) dy^/dz^F giui j/t '*t*f*! Un z) I* r I* f r I* r p :• f i* r r p '* r r r r r i* i*r r i dyn^dz"Fniyi Vn Vn z) If welet the number n increase now so as to become continuously infinite we find that we must substitute for the index a parameter X giving us x dy( z, x)^dz “ F\[y(z £ which is an integrordif ferential equation.] We may also arrive at equations of this character from partial differential equations^! but the classification is not at all easily settled.] These equations are solved by the methods of Cauchy (and earlier Laplace) called by Picard unethod of suocBssiue aoof'oxifnat ion [See Coursatiji Hedrick s Trans.] Vol 2 2 p.j 61 et seq.jl iGoursatiji Vol III Chap.] XXX) .] The types of integrordif ferential equations studies are classified under three typcs^i according to the o^araoterist io curves [See Bvansj,! Co'Jloquium lectures p 90 et seq) i The types are the elliptic^ hyperbolic/ and parabolic^ The entire fidd is an application of the theory of f i o n f functionals (Evans) font tionel 1 es Frechet),] functions in a Volterra (that is continuously infinite a uiil' er ’ o f s 11 j I? e s t s ^ iircensions) spac^, or t-he reporter of these led in brief,, Volterra functions. Derivative ani iifferential of Volterra functions. « In orier to pusch our frontiers as far bacK as possible^ we (Eust consiier next the ilea of variable Volterra function^ ani con- sepuentl.y continuity of such function ani hence iifferential of such function. A Volterra function ieoenclin:? upon f{x) is continuous (Volterra sense) if when we substitute for fix) the function fix) + where ^ix) is bouniei in absolute value,, then the variation of F\{fix)}\, that is ^ j [ ^ (jc j f jcj 1 1 -S’|[rf.x:)]| -rtn Solvinu the integral equation where G is an arhitra-,v function Golvi . : for G we have ( r \ .r Z(x) ^ y(x)-f z)y('^.)dc ^ ^ ^ n fm-tion Any Volterra function of xJ where ^ is the resolven function ■ .y is the result iesi’^ed. nolications of these theories to l-he-e are many aOO ^ Volterra ani his students. Mechanics. ?or *e se consu 1 t t he memoi r s '‘t.’Kv i Liu jk * ■ i - 16 - Latterly they have been applied to the study of Saturn ' sprir^fl*^ nd to nebulae. HISTORICAL.' , y The first work done alon^ these lines be^an with Volte^ra's thesis: Sopra alcuni problemi della teoria del ootenzial e Pisa 1833 In this infinite processes are considered.' In 1884 aooeared an * z integral equation of the first kind\, ^i>(x)=fo F ( ix,x)f ( z, a) da, in •Sopra un probletua i el etro st at ica^ ' Rend.; Ac.; Lincei (3) 6 dp 315—318;^ and in II nuove Cimeuto^ 16, (1834) 9 pp.; In 1387 the function of line (Volterra function) aooeared in Teoria delle equazione dif ferenzial e lineari, Soc.; Ital.‘ d Sci.; (Xl) 6(1887) 1C7 pp and 12 (1899) 69 DO.; There are other papers in the rend.' Lincei.' In 1896 the method of studying problems of this type by starting with a set of al.?ebraic equations, whoxe number is then made continuously infinite, was first used in Sulla inv-.'^^ione de^li inteifrali defirti, Atti Ac.. Torino 31 (1896) 311-523^. 400-408., 557-567., 693-708. See also Rend.- Lincei (5) l77-18o^ 5* 289-300. A summary of preceJin:? results and new results are to be found in Annali di matematica, (2) 25 (1897) 139-178' Inte^ro differential equations were studies in Rend Lincei 1909.; Since then many oaoe*s have appeared Lfp to 1914^ when work was suspended for the WAR. 5 ti ■ 'fc- r*o 'i -’• .~ir 'c»S ^ >■■' ,■ ' : >^-' ■‘H: ' ' 'nyJP' *> ^ v> r '