Online Encyclopedia

CONTINUED FRACTIONS

Online Encyclopedia
Originally appearing in Volume V07, Page 33 of the 1911 Encyclopedia Britannica.
Spread the word: del.icio.us del.icio.us it!
CONTINUED FRACTIONS. In mathematics, an expression of the form a't b2 astb3 b4 a3 a4#b5 reduced to the forms a4 ala2+b2 r' a z asa3a4+a4ba+a2b4 are called the successive convergents to the general continued fraction. Their numerators are denoted by p,, p2, p3, p4... ; their de-nominators by qi, q2, q,, q4.. . We have the relations pn = anpn_3 +bnpn-2, qn = angn_1+bngn-2• In the case of the fraction a,—b2 b3 b4 we have the a2—a,—a4 relations p,, =anpn-l—bnpn-2, qn=angn-i —bngn-z• Taking the quantities al..., bs... to be all positive, a continued fraction of the form a,+a?+a3+...is called a continued fraction of the first class; a continued fraction of the form bs ba b4 is az—a3—a4— • • . called a continued fraction of the second class. A continued fraction of the form al-f-azi r ? +s+a4+..., where al, as, as, a4. . . are all positive integers, is called a simple continued fraction. In the case of this fraction al, a2, a,, a4. . . are called the successive partial quotients. It is evident that, in this case, P1, P2, p3... , qi, are two series of positive integers increasing without limit if the fraction does not terminate. A2b2 A,X1b3 x3x4b4 al+x222+ ~3a3 r A4a4 +• • where X2, 4,, X4, ... are any quantities whatever, so that by choosing X,b2 = r, A,X,b, = r, &c., it can be reduced to any equivalent continued fraction of the form al+d2+d3-I:a4+ • • • Simple Continued Fractions. r. The simple continued fraction is both the most interesting and important kind of continued fraction. Any quantity, commensurable or incommensurable, can be expressed uniquely as a simple continued fraction, terminating in the case of a commensurable quantity, non-terminating in the case of an incommensurable quantity. A non-terminating simple continued fraction must be incommensurable. In the case of a terminating simple continued fraction the number of partial quotients may be odd or even as we please by writing the last partial quotient, an as an — 1+=. The numerators and denominators of the successive convergents obey the law p,,gn_,—pn_lgn(—r)^, from which it follows at' once that every convergent is in its lowest terms. The other principal properties of the convergents are: The odd convergents form an increasing series of rational fractions continually approaching to the value of the whole continued fraction; the even convergents form a decreasing series having the same property. Every even convergent is greater than every odd convergent; every odd convergent is less than, and every even convergent greater than, any following convergent. Every convergent is nearer to the value of the whole fraction than any preceding convergent. Every convergent is a nearer approximation to the value of the whole fraction than any fraction whose denominator is Iess than that of the convergent. The difference between the continued fraction and the nth con- vergent is less than r , and greater than an+z These, limits gnqn+~ gnqn+2 may be replaced by the following, which, though not so close, are simpler, viz. 2 and q q.(qn+qn+l)' Every simple continued fraction must converge to a definite limit ; for its value lies between that of the first and second convergents and, since f'n p=' Lt. f'n = Lt. p=' 4n _ q,,-1 gnqn-1' qn qn-1' so that its value cannot oscillate. The chief practical use of the simple continued fraction is that by means of it we can obtain rational fractions which approximate to any quantity, and we can also estimate the error of our a5$.. where al,a2,a3, . . . and b2,b3,b4, . . . are any quantities whatever, positive or negative, is called a " continued fraction." The quantities al . . . ,b2 . . . may follow any law whatsoever. If the continued fraction terminates, it is said to be a terminating continued fraction; if the number of the quantities al ..., b2 .. . is infinite it is said to be a non-terminating or infinite continued fraction. If b2/as, b3/a3..., the component fractions, as they are called, recur, either from the commencement or from some fixed term, the continued fraction is said to be recurring or periodic. It is obvious that every terminating continued fraction reduces to a commensurable number. The notation employed by English writers for the general continued fraction is a4 t bs bs b4 a2~asta4~' Continental writers frequently use the notation albs b3tb4...,oral#bJb4l ••• a2. as a4 la2 la3 IU'4 The terminating continued fractions al, albs, a1+— bs al+b2 bs b4 W2 a2+aa a2-1-23+a4' ... ala,a,+bsa3+b2a1 a,a3+b3 ' ala2a,a4+b2a3a4+bsala4+b4ala2 +b,b4 The general continued fraction al+b2 bs b4 a2+a3+a4+ equal, convergent by convergent, to the continued fraction is evidently approximation. Thus a continued fraction equivalent ton (the ratio of the circumference to the diameter of a circle) is I I I I I I 3+7+15+I+292+I+I+.. . of which the successive convergents are 1 22 333 355 103993 &c I' 7' 106' 113' 33102' ' the fourth of which is accurate to the sixth decimal place, since the error lies between 1/gags or •0000002673 and as/q'ags or .x000002665. Similarly the continued fraction given by Euler as equivalent to Ea --1) (e being the base of Napierian logarithms), viz. I I I I I 1+6+io+14+18+ .. may be used to approximate very rapidly to the value of e. For the application of continued fractions to the problem " To find the fraction, whose denominator does not exceed a given integer D, which shall most closely approximate (by excess or defect, as may be assigned) to a given number commensurable or incommensurable," the reader is referred to G. Chrystal's Algebra, where also may be found details of the application of continued fractions to such interesting and important problems as the recurrence of eclipses and the rectification of the calendar (q.v.). Lagrange used simple continued fractions to approximate to the solutions of numerical equations; thus, if an equation has a root between two integers a and a+I, put x=a+I/y and form the equation in y; if the equation in y has a root between b and b+I, put y=b+I/z, and so on. Such a method is, however, too tedibus, compared with such a method as Horner's, to be of any practical value. The solution in integers of the indeterminate equation ax+by=c may be effected by means of continued fractions. If we suppose a/b to be converted into a continued fraction and p/q to be the pen-ultimate convergent, we have bq—bp=+1 or -1, according as the number of convergents is even or odd, which we can take them to be as we please. If we take aq—bp=+1 we have a general solution in integers of ax+by=c, viz. x=cq—bt, y=at—cp; if we take aq—bp= -1, we have x=bt—cq, y=cp—at. An interesting application of continued fractions to establish a unique correspondence between the elements of an aggregate of m dimensions and an aggregate of n dimensions is given by G. Cantor in vol. 2 of the Acta Mathematica. Applications of simple continued fractions to the theory of numbers, as, for example, to prove the theorem that a divisor of the sum of two squares is itself the sum of two squares, may be found in J. A. Serret's Cours d'Algebre Superieure. 2. Recurring Simple Continued Fractions.—The infinite continued fraction a1+ I I .1~ I I i I I I I I a2+a3+ I an+bl+b2+•.• +bn+bl+b2+• +b„+bi+.••, where, after the nth partial quotient, the cycle of partial quotients b1, b2,. • • r b,, recur in the same order, is the type of a recurring simple continued fraction. The value of such a fraction is the positive root of a quadratic equation whose coefficients are real and of which one root is negative. Since the fraction is infinite it cannot be commensurable and there-fore its value is a quadratic surd number. Conversely every positive quadratic surd number, when expressed as a simple continued fraction, will give rise to a recurring fraction. Thus 2—s/3=3+I+2+1+2+ ..., 1128=5+I I I I _I I I I 3+2+3+10+3+2+3+10+ • • . The second case illustrates a feature of the recurring continued fraction which represents a complete quadratic surd. There is only one non-recurring partial quotient a1. If b1, b2, ..., b,, is the cycle of recurring quotients, then b,, =2a1,bib2=bn_2,b3&c. In the case of a recurring continued fraction which represents N, where N is an integer, if n is the number of partial quotients in the recurring cycle, and pnr/qnr the nrth convergent, then p2nr—Ng2nr (— I)", whence, if n is odd, integral solutions of the indeterminate equation x2 —Ny2 = =1 (the so-called Pellian equation) can be found. If n is even, solutions of the equation x2—Ny2=+1 can be found. The theory and development of the simple recurring continued fraction is due to Lagrange. For proofs of the theorems here stated and for applications to the more general indeterminate equation x2 —Ny2 = H the reader may consult Chrystal's Algebra or Serret's Cours d'Algebre Superieure; he may also profitably consult a tract by T. Muir, The Expression of a Quadratic Surd as a Continued Fraction (Glasgow, 1894). The General Continued Fraction. 1. The Evaluation of Continued Fractions.—The numerators and denominators of the convergents to the general continued fraction both satisfy the difference equation un=a,,un_i+b,,un-2• When wecan solve this equation we have an expression for the nth convergent to the fraction, generally in the form of the quotient of two series, each of n terms. As an example, take the fraction (known as Brouncker's fraction, after Lord Brouncker) I I2 32 52 72 ?tin+1 =2un+(2n-.I)2un-i, n„+1—(272+1)24 =—(21t —I){un—(211—I)ZGn_i}, and we readily find that —=I—3—+5 n -7+ ... = 2n+I' whence the value of the fraction taken to infinity is 17r. It is always possible to find the value of the nth convergent to a recurring continued fraction. If r be the number of quotients in the recurring cycle, we can by writing down the relations connecting the successive p's and q's obtain a linear relation connecting pnr+m{ P(n-1)r+mr p(n-2)Nm, in which the coefficients are all constants. Or we may proceed as follows. (We need not consider a fraction with a non-recurring part). Let the fraction be al a2 ar a1 b.l+b2+... +br+61+ . qn qn_i gngn-1 hand side is not necessarily zero. The tests for convergency are as follows: Let the continued fraction of the first class be reduced to the form d'+d2+a3+a4+ then it is convergent if at least one of the series d3+d5+d7+ ..., d2+da+ds+ . . . diverges, and oscillates if both these series converge. For the convergence of the continued fraction of the second class there is no complete criterion. The following theorem covers a large number of important cases. " If in the infinite continued fraction of the second class a,. b,,+T. for all values of n, it converges to a finite limit not greater than unity.,, 3. The Incommensurability of Infinite Continued Fractions.—There is no general test for the incommensurability of the general infinite continued fraction. Two cases have been given by Legendre as follows; If as, as, • ., an, b2, b3, ..., b,, are all positive integers, then I. The infinite continued fraction bz bn con- a2+a3+ +an+ ... verges to an incommensurable limit if after some finite value of n the condition a,, need not always occur but must occur infinitely often. Continuants. The functions p,, and q,,, regarded as functions of al, .. b2, ..., b,, determined by the relations hh Ps = an pn-1 +bnYn-2, qn=a,qn-l+bngn-2, with the conditions Pi =al, Po =I; q2 =a2, qi = I, qo =0, have been studied under the name of continuants. The notation adopted is pn=K (al, b2, • b,,) as, . , an and it is evident that we have qn = (a2, a3, ... , an) The theory of continuants is due in the first place to Euler. The reader will find the theory completely treated in Chrystal's Algebra, where will be found the exhibition of a prime number of the form 4p+I as the actual sum of two squares by means of continuants, a result given by H. J. S. Smith. Here we have whence Let un pn'+m ; then us _ al b2 a' , leading to an qnr+m bt++ ... +br+uni equation of the form Au,un_1+Bun+Cu„_i+D=o, where A,B,C,D are independent of n, which is readily solved. 2. The Convergence of Infinite Continued Fractions.—We have seen that the simple infinite continued fraction converges. The infinite general continued fraction of the first class cannot diverge for its value lies between that of its first two convergents. It may, how-ever, oscillate. ~ We have the relation p„q„_l —Ps_1q, = (—1)"b2ba.. b,,, from __. L:. l. Pn Ps-1.-r .\nb2b3. bn and the limit of the right- K b2, b3, ... , b,, ~ah a2, a3,. .. , a,, is also equal to the The continuant determinant al b2 o o o -1 a2 ba o o o – i a3 b4 o o o -1 a4 bs – u -1 an_1 bn O O - - O O -I a,, , from which point of view continuants have been treated by W. Spottiswoode, J. J. Sylvester and T. Muir. Most of the theorems concerning continued fractions can be thus proved simply from the properties of determinants (see T. Muir's Theory of Determinants, chap. iii.). Perhaps the earliest appearance in analysis of a continuant in its determinant form occurs in Lagrange's investigation of the vibrations of a stretched string (see Lord Rayleigh, Theory of Sound, vol. i. chap. iv.). The Conversion of Series and Products into Continued Fractions. I. A continued fraction may always be found whose nth convergent shall be equal to the sum to n terms of a given series or the product to n factors of a given continued product. In fact, a continued fraction aid a2+...+¢„+... can be constructed having for the numerators of its successive convergents any assigned quantities Pi, P2, Ps, . . , Pe, and for their denominators any assigned quantities qi, q2, qa, ... , q,,.. . The partial fraction b,,/a,, corresponding to the nth convergent can be found from the relations pn = anp,,-1+b,,p,,-2, qe= angn-1 +bngn-2; and the first two partial quotients are given by bl=P3, al=q1, bta2=p2, alaz+b2=q2. If we form then the continued fraction inwhich p,, p2, pa, ..., pn are ul, ul+u2, ul+u2+u3, ..., ul+u2+ ...' u,,, and q1, q2, q3, ... qn are all unity, we find the series ul+u2+ . . . +u„ equivalent to the continued fraction u2 us u_ ul u, u2 I I+,f1-I+ua-... -i d-ue-i which we can transform into ul u2 ulu3 u2U4 un-2un I -uj+u2-u2+u3-u3d-u4–...-nn I+unf a result given by Euler. 2. In this case the sum to n terms of the series is equal to the nth convergent of the fraction. There is, however, a different way in which a series may be represented by a continued fraction. We may require to represent the infinite convergent power series ao+alx+ aix2d- ... by an infinite continued fraction of the form flo t31x t32x $sx Here the fraction converges to the sum to infinity of the series. Its nth convergent is not equal to the sum to n terms of the series. Expressions for So, #1, 02, ... by means of determinants have been given by T. Muir (Edinburgh Transactions, vol. xxvii.). A method was given by J. H. Lambert for expressing as a continued fraction of the preceding type the quotient of two convergent power series. It is practically identical with that of finding the greatest common measure of two polynomials. As an instance leading to results of some importance consider the series z F(n'x) I +(7d-n) I!+(7+n)(7+n+1)2!+.. . We have F(n+i,x) –F(n,x) _ – (7+n) (7+n+I)F(n+2,x), whence we obtain F(I,x) x/Y(Y+I) x/(Y+I)(7+2) F(o,x –i I + I d- .., which may also be written y x x 7+7+1+7+2+... By putting x2/4 for x in F(o,x) and F(I,x), and putting at the same time 7=1/2, we obtain 2 2 z z z 2 tan x=x x x x tank x=x x x x 1– 3– 5– 7 –... 1+3+ 5+7 +. These results were given by Lambert, and used by him to (prove that 2r and 2r-2 are incommensurable, and also any commensurable power of e. Gauss in his famous memoir on the hypergeometric series F(a, x) = Ia0+c(ad-I)O(S+1) r2~ . . I.y I.z.7.(7+I) gave the expression for F(a, j9+I, 7+1, x) =F(a, S, y, x) as a continued fraction, from which if we put /3=o and write y–I for y, we get the transformation a a(a+I) a(a+I)(a+2) I Rix Six +7x+7(7+I)x2+7(7+I)(7+2)x3+ ... =I_ I – I –...where I+I= a, R3= (a+I)7 ~2>,-i- (a+n-I)(7+n-2) 7 (7+I)(7+2)'.. (7+2n–3)(Y+2n-2)' Q2 = 7-a lj4 -2(7+I-a) t32n - n(7+n–I–a) 7(v+' ' '(7+2)(7+3)'. ' (7+2n-2)(7+2n-I). From this we may express several of the elementary series as continued fractions; thus taking a=l, y=2, and putting x for –x, we have log I x 12x 22x 22x 3( +x)=id 2 d 3I2x d 4 +!l+± 75 +. Taking y = i, writing x/a for x and increasing a indefinitely, we have es=l x _x x _x _x For some recent developments in this direction the reader may consult a paper by L. J. Rogers in the Proceedings of the London Mathematical Society (series 2, vol. 4). Ascending Continued Fractions. There is another type of continued fraction called the ascending continued fraction, the type so far discussed being called the descending continued fraction. It is of no interest or importance, though both Lambert and Lagrange devoted some attention to it. The notation for this type of fraction is b4+b5+ 1)3+ bed- a4 a3 al+ a2 It is obviously equal to the series al +a2 +a,2--a3 +a2a,a4+a2aba4a5 +.. Historical Note. The invention of continued fractions is ascribed generally to Pietro Antonia Cataldi, an Italian mathematician who died in 1626. He used them to represent square roots, but only for particular numerical examples, and appears to have had no theory on the subject. A previous writer, Rafaello Bombelli, had used them in his treatise on Algebra (about 1579), and it is quite possible that Cataldi may have got his ideas from him. His chief advance on Bombelli was in his notation. They next appear to have been used by Daniel Schwenter (1585–1636) in a Geometrica Practica published in 1618. He uses them for approximations. The theory, however, starts with the publication in 1655 by Lord Brouncker of the continued fraction 2 2 52 s } z + 2 + . , as an equivalent of 7r/4. This he is supposed to have deduced, no one knows how, from Wallis' formula for 4/7, viz. 3.3.5.5.7.7.. . 2.4.4.6.6.8... John Wallis, discussing this fraction in his Arithmetica Infinitorum (1656), gives many of the elementary properties of the convergents to the general continued fraction, including the rule for their formation. Huygens (Descriptio automati planetaria, 1703) uses the simple continued fraction for the purpose of approximation when designing the toothed wheels of his Planetarium. Nicol Saunderson (1682–1739), Euler and Lambert helped in developing the theory, and much was done by Lagrange in his additions to the French edition of Euler's Algebra (1795). Moritz A. Stern wrote at length on the subject in Crelle's Journal (x., 1833; xi., 1834; xxiii., 1838). The theory of the convergence of continued fractions is due to Oscar Schlomilch, P. F. Arndt, P. L. Seidel and Stern. O. Stolz, A. Pringsheim and E. B. van Vleck have written on the convergence of infinite continued fractions with complex elements. un–1 Irrational numbers there is P. Bachmann's Vorlesungen fiber die Natur der Irrationalzahnen (1892). For the application of continued fractions to the theory of lenses, see R. S. Heath's Geometrical Optics, chaps. iv. and v. For an exhaustive summary of all that has been written on the subject the reader may consult Bd. 1 of the Encyklopadie der mathematischen Wissenschaften (Leipzig). (A. E. J.)
End of Article: CONTINUED FRACTIONS
[back]
CONTINENTAL SHELF
[next]
CONTOUR

Additional information and Comments

There are no comments yet for this article.
» Add information or comments to this article.
Please link directly to this article:
Highlight the code below, right click and select "copy." Paste it into a website, email, or other HTML document.