Constante de Catalan

Page d’aide sur l’homonymie

Pour les articles homonymes, voir Catalan (homonymie).

En mathématiques, la constante de Catalan, portant le nom du mathématicien Eugène Charles Catalan, est le nombre défini par : K = β ( 2 ) = n = 0 ( 1 ) n ( 2 n + 1 ) 2 0,915   965   594 {\displaystyle K=\beta (2)=\sum _{n=0}^{\infty }{\frac {(-1)^{n}}{(2n+1)^{2}}}\approx 0{,}915~965~594} β {\displaystyle \beta } est la fonction bêta de Dirichlet.

Ses décimales sont répertoriées par la suite A006752 de l'OEIS.

On ne sait pas si la constante K {\displaystyle K} est rationnelle ou irrationnelle.

Autres expressions

La constante K {\displaystyle K} de Catalan est aussi égale à :

Expressions intégrales

  • I 1 = 0 1 arctan u u d u = T i 2 ( 1 ) {\displaystyle I_{1}=\int _{0}^{1}{\arctan u \over u}\,\mathrm {d} u=\mathrm {Ti} _{2}(1)} , où Ti2 désigne la fonction arc tangente intégral
  • I 2 = 1 2 0 π / 2 u sin u d u {\displaystyle I_{2}={1 \over 2}\int _{0}^{\pi /2}{\frac {u}{\sin u}}\mathrm {d} u}
  • I 3 = 0 1 ln u 1 + u 2 d u = 1 + ln u 1 + u 2 d u {\displaystyle I_{3}=-\int _{0}^{1}{\ln u \over 1+u^{2}}\,\mathrm {d} u=\int _{1}^{+\infty }{\ln u \over 1+u^{2}}\,\mathrm {d} u}
  • I 4 = 0 π / 4 ln ( tan u ) d u = 0 π / 4 ln ( cot u ) d u = u 2 u 1 4 0 π / 2 ln ( 1 + cos u 1 cos u ) d u {\displaystyle I_{4}=-\int _{0}^{\pi /4}{\ln(\tan u)\,\mathrm {d} u}=\int _{0}^{\pi /4}{\ln(\cot u)\,\mathrm {d} u}{\overset {u\leftrightarrow 2u}{=}}{\frac {1}{4}}\int _{0}^{\pi /2}\ln \left({\frac {1+\cos u}{1-\cos u}}\right)\,\mathrm {d} u}
  • I 5 = 0 1 0 1 d x d y 1 + x 2 y 2 {\displaystyle I_{5}=\int _{0}^{1}\int _{0}^{1}{\frac {\mathrm {d} x\mathrm {d} y}{1+x^{2}y^{2}}}}
  • I 6 = 0 1 / 2 arcsin u u ( 1 u 2 ) d u {\displaystyle I_{6}=\int _{0}^{1/{\sqrt {2}}}{\arcsin u \over {u(1-u^{2})}}\mathrm {d} u}
  • I 7 = 0 ln ( 1 + 2 ) arccos ( sh u ) d u = sh u u 0 1 arccos u 1 + u 2 d u {\displaystyle I_{7}=\int _{0}^{\ln(1+{\sqrt {2}})}\arccos {(\operatorname {sh} u)}\,\mathrm {d} u{\overset {\operatorname {sh} u\leftrightarrow u}{=}}\int _{0}^{1}{\frac {\arccos u}{\sqrt {1+u^{2}}}}\,\mathrm {d} u}
  • I 8 = 0 π / 2 argsh ( sin u ) d u = sin u u 0 1 argsh u 1 u 2 d u {\displaystyle I_{8}=\int _{0}^{\pi /2}\operatorname {argsh} {(\sin u)}\,\mathrm {d} u{\overset {\sin u\leftrightarrow u}{=}}\int _{0}^{1}{\frac {\operatorname {argsh} u}{\sqrt {1-u^{2}}}}\,\mathrm {d} u}
  • I 9 = 1 2 0 π / 2 argth ( sin u ) d u = sin u u 1 2 0 1 argth u 1 u 2 d u {\displaystyle I_{9}={1 \over 2}\int _{0}^{\pi /2}\operatorname {argth} {(\sin u)}\,\mathrm {d} u{\overset {\sin u\leftrightarrow u}{=}}{1 \over 2}\int _{0}^{1}{\frac {\operatorname {argth} u}{\sqrt {1-u^{2}}}}\,\mathrm {d} u}
  • I 10 = 0 + arctan ( e u ) d u {\displaystyle I_{10}=\int _{0}^{+\infty }\arctan(e^{-u})\,\mathrm {d} u}
  • I 11 = 1 2 0 + u ch u d u {\displaystyle I_{11}={\frac {1}{2}}\int _{0}^{+\infty }{\frac {u}{\operatorname {ch} u}}\,\mathrm {d} u}
  • I 12 = 1 2 0 1 F ( k ) d k {\displaystyle I_{12}={1 \over 2}\int _{0}^{1}F(k)\,\mathrm {d} k\quad } F ( k ) = 0 π 2 d φ 1 k 2 sin 2 φ {\displaystyle \quad F(k)=\int _{0}^{\frac {\pi }{2}}{\frac {\,\mathrm {d} \varphi }{\sqrt {1-k^{2}\sin ^{2}\varphi }}}} est l'intégrale elliptique complète de première espèce
  • I 13 = 1 2 + 0 1 E ( k ) d k {\displaystyle I_{13}=-{1 \over 2}+\int _{0}^{1}E(k)\,\mathrm {d} k\quad } E ( k ) = 0 π 2 1 k 2 sin 2 φ d φ {\displaystyle \quad E(k)=\int _{0}^{\frac {\pi }{2}}{\sqrt {1-k^{2}\sin ^{2}\varphi }}\,\mathrm {d} \varphi } est l'intégrale elliptique complète de deuxième espèce
  • I 14 = 1 4 0 1 0 1 d x d y ( x + y ) ( 1 x ) ( 1 y ) {\displaystyle I_{14}={1 \over 4}\int _{0}^{1}\int _{0}^{1}{\frac {\mathrm {d} x\mathrm {d} y}{(x+y){\sqrt {(1-x)(1-y)}}}}}
  • I 15 = 1 / 2 1 argth u u 2 u 2 1 d u = u cos u 0 π / 4 tan u argth ( cos u ) cos 2 u d u {\displaystyle I_{15}=\int _{1/{\sqrt {2}}}^{1}{\frac {\operatorname {argth} u}{u{\sqrt {2u^{2}-1}}}}\,\mathrm {d} u{\overset {u\leftrightarrow \cos u}{=}}\int _{0}^{\pi /4}{\frac {\tan u\operatorname {argth} (\cos u)}{\sqrt {\cos 2u}}}\,\mathrm {d} u}
  • I 16 = 0 1 / 2 argth u ( 1 u 2 ) 1 2 u 2 d u = u sin u 0 π / 4 argth ( sin u ) cos u cos 2 u d u {\displaystyle I_{16}=\int _{0}^{1/{\sqrt {2}}}{\frac {\operatorname {argth} u}{{(1-u^{2})}{\sqrt {1-2u^{2}}}}}\,\mathrm {d} u{\overset {u\leftrightarrow \sin u}{=}}\int _{0}^{\pi /4}{\frac {\operatorname {argth} (\sin u)}{\cos u{\sqrt {\cos 2u}}}}\,\mathrm {d} u}
Démonstrations de ces égalités

On passe de I 1 {\displaystyle I_{1}} à K {\displaystyle K} en développant arctan {\displaystyle \arctan } en série entière ;

on passe de I 1 {\displaystyle I_{1}} à I 2 {\displaystyle I_{2}} en posant u = tan v 2 {\displaystyle u=\tan {v \over 2}} ;

on passe de I 1 {\displaystyle I_{1}} à I 3 {\displaystyle I_{3}} en intégrant par parties (dériver arctan {\displaystyle \arctan } et intégrer u 1 / u {\displaystyle u\mapsto 1/u} ) et le passage entre les deux expressions se fait en changeant u {\displaystyle u} en 1 / u {\displaystyle 1/u}  ;

on passe de I 3 {\displaystyle I_{3}} à I 4 {\displaystyle I_{4}} en posant u = tan v {\displaystyle u=\tan v} ;

on passe de I 5 {\displaystyle I_{5}} à I 1 {\displaystyle I_{1}} en posant x = u y {\displaystyle x={u \over y}} ;

on passe de I 6 {\displaystyle I_{6}} à I 2 {\displaystyle I_{2}} en posant u = sin v 2 {\displaystyle u=\sin {v \over 2}} ;

on passe de I 7 {\displaystyle I_{7}} à I 8 {\displaystyle I_{8}} en intégrant par parties ;

démonstration de I 8 = I 1 {\displaystyle I_{8}=I_{1}}  :

argsh x = 0 x d u 1 + u 2 = u = t x 0 1 x d t 1 + t 2 x 2 {\displaystyle \operatorname {argsh} x=\int _{0}^{x}{\frac {\mathrm {d} u}{\sqrt {1+u^{2}}}}\,{\overset {u=tx}{=}}\int _{0}^{1}{\frac {x\,\mathrm {d} t}{\sqrt {1+t^{2}x^{2}}}}\,} donc I 8 = 0 π / 2 0 1 sin u 1 + t 2 sin 2 u d t d u = 0 1 0 π / 2 sin u d u 1 + t 2 sin 2 u d t {\displaystyle I_{8}=\int _{0}^{\pi /2}\int _{0}^{1}{\frac {\sin u}{\sqrt {1+t^{2}\sin ^{2}u}}}\,\mathrm {d} t\,\mathrm {d} u=\int _{0}^{1}\int _{0}^{\pi /2}{\frac {\sin u\,\mathrm {d} u}{\sqrt {1+t^{2}\sin ^{2}u}}}\,\mathrm {d} t}

or 0 π / 2 sin u 1 + t 2 sin 2 u d u = v = cos u 0 1 d v 1 + t 2 t 2 v 2 = 1 t arcsin t 1 + t 2 = arctan t t {\displaystyle \int _{0}^{\pi /2}{\frac {\sin u}{\sqrt {1+t^{2}\sin ^{2}u}}}\,\mathrm {d} u{\overset {v=\cos u}{=}}\int _{0}^{1}{\frac {\mathrm {d} v}{\sqrt {1+t^{2}-t^{2}v^{2}}}}={1 \over t}\arcsin {t \over {\sqrt {1+t^{2}}}}={\arctan {t} \over t}} donc I 8 = I 1 {\displaystyle I_{8}=I_{1}}  ;

variante : posons f ( x ) = 0 π / 2 argsh ( x sin u ) d u {\displaystyle f(x)=\int _{0}^{\pi /2}\operatorname {argsh} {(x\sin u)}\,\mathrm {d} u}  ; alors f ( x ) = 0 π / 2 sin u 1 + x 2 sin 2 u d u = arctan x x {\displaystyle f'(x)=\int _{0}^{\pi /2}{\frac {\sin u}{\sqrt {1+x^{2}\sin ^{2}u}}}\,\mathrm {d} u={\arctan x \over x}} comme ci-dessus, donc I 8 = f ( 1 ) = 0 1 f ( x ) d x = 0 1 arctan ( x ) x d x = I 1 {\displaystyle I_{8}=f(1)=\int _{0}^{1}f'(x)dx=\int _{0}^{1}{\frac {\arctan(x)}{x}}\mathrm {d} x=I_{1}} ;

démonstration de I 4 {\displaystyle I_{4}} = I 9 {\displaystyle I_{9}}

0 π / 4 ln ( cot ( u ) ) d u = 1 2 0 π / 2 ln ( cot ( 2 u ) ) d u = 1 2 0 π / 2 ln 1 + cos u 1 cos u d u = 1 2 0 π / 2 argth cos u d u = 1 2 0 π / 2 argth sin u d u {\displaystyle \int _{0}^{\pi /4}{\ln(\cot(u))\,\mathrm {d} u}={1 \over 2}\int _{0}^{\pi /2}{\ln(\cot(2u))\,\mathrm {d} u}={1 \over 2}\int _{0}^{\pi /2}{\ln {\sqrt {{1+\cos u} \over {1-\cos u}}}\,\mathrm {d} u}={1 \over 2}\int _{0}^{\pi /2}{\operatorname {argth} \cos u\,\mathrm {d} u}={1 \over 2}\int _{0}^{\pi /2}{\operatorname {argth} \sin u\,\mathrm {d} u}}

on passe de I 10 {\displaystyle I_{10}} à I 2 {\displaystyle I_{2}} en posant e u = tan v 2 {\displaystyle e^{-u}=\tan {v \over 2}} ;

on passe de I 11 {\displaystyle I_{11}} à I 3 {\displaystyle I_{3}} en posant u = ln v {\displaystyle u=\ln v} ;

on passe de I 12 {\displaystyle I_{12}} et I 13 {\displaystyle I_{13}} à I 2 {\displaystyle I_{2}} en intervertissant les signes d'intégration;

passage de I 14 {\displaystyle I_{14}} à I 15 {\displaystyle I_{15}}  :

0 1 d x ( x + y ) ( 1 x ) ( 1 y ) = [ 2 argth 1 x 1 + y 1 y 2 ] 0 1 = 2 argth 1 1 + y 1 y 2 {\displaystyle \int _{0}^{1}{\frac {\mathrm {d} x}{(x+y){\sqrt {(1-x)(1-y)}}}}=\left[{\frac {-2\operatorname {argth} {\sqrt {1-x \over {1+y}}}}{\sqrt {1-y^{2}}}}\right]_{0}^{1}={\frac {2\operatorname {argth} {\sqrt {1 \over {1+y}}}}{\sqrt {1-y^{2}}}}}

donc I 14 = 1 4 0 1 0 1 d x d y ( x + y ) ( 1 x ) ( 1 y ) = 0 1 argth 1 1 + y 2 1 y 2 d y = u = 1 / 1 + y 1 / 2 1 argth u u 2 u 2 1 d u = I 15 {\displaystyle I_{14}={1 \over 4}\int _{0}^{1}\int _{0}^{1}{\frac {\mathrm {d} x\mathrm {d} y}{(x+y){\sqrt {(1-x)(1-y)}}}}=\int _{0}^{1}{\frac {\operatorname {argth} {\sqrt {1 \over {1+y}}}}{2{\sqrt {1-y^{2}}}}}\mathrm {d} y{\overset {u=1/{\sqrt {1+y}}}{=}}\int _{1/{\sqrt {2}}}^{1}{\frac {\operatorname {argth} u}{u{\sqrt {2u^{2}-1}}}}\,\mathrm {d} u=I_{15}}

pour I 14 {\displaystyle I_{14}} et I 16 {\displaystyle I_{16}} , voir la référence Bradley ci-dessous, pages 23 et 11.

Développements en série

Cette constante peut aussi être définie par la fonction de Clausen : K = Cl 2 ( π 2 ) = n = 1 sin ( n π 2 ) n 2 {\displaystyle K=\operatorname {Cl} _{2}\left({\frac {\pi }{2}}\right)=\sum _{n=1}^{\infty }{\frac {\sin(n{\frac {\pi }{2}})}{n^{2}}}} , ce qui nous donne les formules suivantes :

  • K = 0 π 2 ln ( 2 sin u 2 ) d u {\displaystyle K=-\int _{0}^{\frac {\pi }{2}}\ln {\biggl (}2\sin {u \over 2}{\biggr )}\,\mathrm {d} u} ,
  • K = π 2 ( 1 ln ( π 2 ) + n = 1 ζ ( 2 n ) n ( 2 n + 1 ) ( 1 4 ) 2 n ) {\displaystyle K={\frac {\pi }{2}}\left(1-\ln \left({\frac {\pi }{2}}\right)+\sum _{n=1}^{\infty }{\frac {\zeta (2n)}{n(2n+1)}}\left({\frac {1}{4}}\right)^{2n}\right)} ,
  • K = π 2 ( 3 ln ( 15 π 32 ) 4 ln ( 5 3 ) + n = 1 ζ ( 2 n ) 1 n ( 2 n + 1 ) ( 1 4 ) 2 n ) {\displaystyle K={\frac {\pi }{2}}\left(3-\ln \left({\frac {15\pi }{32}}\right)-4\ln \left({\frac {5}{3}}\right)+\sum _{n=1}^{\infty }{\frac {\zeta (2n)-1}{n(2n+1)}}\left({\frac {1}{4}}\right)^{2n}\right)} .

Puisque K {\displaystyle K} est l'image de 2 par la fonction bêta, nous avons un lien avec le polylogarithme : Li 2 ( i ) = π 2 48 + i K {\displaystyle \operatorname {Li} _{2}(\mathrm {i} )=-{\frac {\pi ^{2}}{48}}+\mathrm {i} K} , d'où : K = ( Li 2 ( i ) ) {\displaystyle K=\Im (\operatorname {Li} _{2}(\mathrm {i} ))} .

Utilisation

K apparaît en combinatoire, ainsi que dans les valeurs de la fonction polygamma de deuxième ordre, aussi appelée la fonction trigamma : ψ 1 ( 1 4 ) = π 2 + 8 K {\displaystyle \psi _{1}\left({\frac {1}{4}}\right)=\pi ^{2}+8K} , ψ 1 ( 3 4 ) = π 2 8 K {\displaystyle \psi _{1}\left({\frac {3}{4}}\right)=\pi ^{2}-8K} .

Simon Plouffe donne une famille infinie d'identités entre la fonction trigamma, π 2 {\displaystyle \pi ^{2}} et la constante de Catalan.

K apparaît aussi dans la loi sécante hyperbolique.

Séries convergeant rapidement

Les deux formules suivantes convergent rapidement vers K et sont donc appropriées pour le calcul numérique :

K = {\displaystyle K=\,} 3 n = 0 1 2 4 n ( 1 2 ( 8 n + 2 ) 2 + 1 2 2 ( 8 n + 3 ) 2 1 2 3 ( 8 n + 5 ) 2 + 1 2 3 ( 8 n + 6 ) 2 1 2 4 ( 8 n + 7 ) 2 + 1 2 ( 8 n + 1 ) 2 ) {\displaystyle 3\sum _{n=0}^{\infty }{\frac {1}{2^{4n}}}\left(-{\frac {1}{2(8n+2)^{2}}}+{\frac {1}{2^{2}(8n+3)^{2}}}-{\frac {1}{2^{3}(8n+5)^{2}}}+{\frac {1}{2^{3}(8n+6)^{2}}}-{\frac {1}{2^{4}(8n+7)^{2}}}+{\frac {1}{2(8n+1)^{2}}}\right)-}
2 n = 0 1 2 12 n ( 1 2 4 ( 8 n + 2 ) 2 + 1 2 6 ( 8 n + 3 ) 2 1 2 9 ( 8 n + 5 ) 2 1 2 10 ( 8 n + 6 ) 2 1 2 12 ( 8 n + 7 ) 2 + 1 2 3 ( 8 n + 1 ) 2 ) {\displaystyle 2\sum _{n=0}^{\infty }{\frac {1}{2^{12n}}}\left({\frac {1}{2^{4}(8n+2)^{2}}}+{\frac {1}{2^{6}(8n+3)^{2}}}-{\frac {1}{2^{9}(8n+5)^{2}}}-{\frac {1}{2^{10}(8n+6)^{2}}}-{\frac {1}{2^{12}(8n+7)^{2}}}+{\frac {1}{2^{3}(8n+1)^{2}}}\right)}

et K = π 8 ln ( 3 + 2 ) + 3 8 n = 0 ( n ! ) 2 ( 2 n ) ! ( 2 n + 1 ) 2 {\displaystyle K={\frac {\pi }{8}}\ln \left({\sqrt {3}}+2\right)+{\frac {3}{8}}\sum _{n=0}^{\infty }{\frac {(n!)^{2}}{(2n)!(2n+1)^{2}}}} .

Les calculs théoriques pour cette série ont été donnés par Broadhurst[1].

Décimales connues

Le nombre de chiffres connus de la constante de Catalan a augmenté radicalement pendant les dernières décennies. Ceci est dû à l'augmentation des performances des ordinateurs et aux améliorations algorithmiques[2].

Nombres de chiffres connus de la constante de Catalan
Date Décimales Calculé par
2009 31 026 000 000 R. Shan et A. J. Yee
Octobre 2006 5 000 000 000 Shigeru Kondo[3]
2002 201 000 000 Xavier Gourdon et Pascal Sebah
2001 100 000 500 Xavier Gourdon et Pascal Sebah
4 janvier 1998 12 500 000 Xavier Gourdon
1997 3 379 957 Patrick Demichel
1996 1 500 000 Thomas Papanikolaou
29 septembre 1996 300 000 Thomas Papanikolaou
14 août 1996 100 000 Greg J. Fee et Simon Plouffe
1996 50 000 Greg J. Fee
1990 20 000 Greg J. Fee
1913 32 James W. L. Glaisher
1877 20 James W. L. Glaisher

Notes et références

(en) Cet article est partiellement ou en totalité issu de l’article de Wikipédia en anglais intitulé « Catalan's constant » (voir la liste des auteurs).
  1. (en) D. J. Broadhurst, Polylogarithmic ladders, hypergeometric series and the ten millionth digits of ζ(3) and ζ(5), arXiv : math.CA/9803067, 1998.
  2. (en) Constants and Records of Computation sur le site de X. Gourdon et P. Sebah.
  3. (en) Value of Catalan constant sur le site de Shigeru Kondo.

Voir aussi

Bibliographie

  • E. Catalan, « Mémoire sur la transformation des séries, et sur quelques intégrales définies : Extrait par l'auteur », CRAS, vol. 59,‎ , p. 618-620 (lire en ligne)
  • (en) Henri Cohen, Number Theory, vol. II : Analytic and Modern Tools, New York, Springer, , 596 p. (ISBN 978-0-387-49893-5, lire en ligne), p. 127
  • François Le Lionnais, Les nombres remarquables, Hermann, 1983 puis 1999 (ISBN 2-7056-1407-9)
  • (en) H. M. Srivastava et Choi Junesang, Series Associated With the Zeta and Related Functions, KluwerAcademic, , 388 p. (ISBN 978-0-7923-7054-3, lire en ligne), p. 30
  • (en) D.M. Bradley, Representations of Catalan's constant, KluwerAcademic, (lire en ligne)

Lien externe

(en) Eric W. Weisstein, « Catalan's Constant », sur MathWorld

  • icône décorative Arithmétique et théorie des nombres