2,1

The LG1 matrix coefficients are defined by LG1[2m,1] = 2*lambda(2m+1) for m = 1, 2, .. , and the recurrence relation LG1[2*m,n] = LG1[2*m-2,n-1]/((2*n-3)*(2*n-1)) - (2*n-3)*LG1[2*m,n-1]/(2*n-1) with m = .. , -2, -1, 0, 1, 2, .. and n = 1, 2, 3, .. , under the condition that n <= m. As usual lambda(m) = (1-2^(-m))*zeta(m) with zeta(m) the Riemann zeta function. For the LG2 matrix, the even counterpart of the LG1 matrix, see A008956.

These two formulae enable us to determine the values of the LG1[2*m,n] coefficients, with m all integers and n all positive integers, but not for all. If we choose, somewhat but not entirely arbitrarily, LG1[0,1] = gamma, with gamma the Euler-Mascheroni constant, we can determine them all.

The coefficients in the columns of the LG1 matrix, for m => 1 and n => 2, can be generated with GFL(z;n) = (hg(n)*CFN2(z;n)*GFL(z;n=1) + LAMBDA(z;n))/pg(n) with pg(n) = 6*(2*n-3)!!*(2*n-1)!!*A160476(n) and hg(n) = 6*A160476(n). For the CFN2(z;n) and the LAMBDA(z;n) see A160487.

The values of the column sums cs(n) = sum(LG1[2*m,n], m = 0.. infinity), for n => 2, can be determined with the first Maple program. In this program we have made use of the remarkable fact that if we take LGx[2*m,n] = 2, for m=>0, and LGx[ -2,n] = LG1[ -2,n] and assume that the recurrence relation remains the same we find that the column sums of this new matrix converge to the same values as the original cs(n).

The LG1[2*m,n] matrix coefficients can be generated with the second Maple program.

The LG1 matrix is related to the LS1 matrix, see A160487 and the formulae below.

M. Abramowitz and I. A. Stegun, eds., Handbook of Mathematical Functions, National Bureau of Standards, Applied Math. Series 55, Tenth Printing, 1972, Chapter 23, pp. 811-812.

a(n) = numer(cs(n)) and denom(cs(n)) = A162449(n).

with cs(n) = sum(LG1[2*m,n], m = 0 .. infinity) for n => 2.

GFL(z;n) = sum( LG1[2*m,n]*z^(2*m-2),m=1..infinity)

GFL(z;n) = (LG1[ -2,n-1])/((2*n-3)*(2*n-1))+(z^2/((2*n-3)*(2*n-1))-(2*n-3)/(2*n-1))*GFL(z;n-1) with GFL(z;n=1) = -2*Psi(1-z)+Psi(1-(z/2))-(Pi/2)*tan(Pi*z/2)

LG1[ -2,n] = (-1)^(n+1)*4*(A061549(n-1)/A001790(n-1))*(A002197(n-1)/A002198(n-1))

LG1[2*m,n] = (4^(n-1)/((2*n-1)*binomial(2*n-2,n-1)))*LS1[2*m,n]

The first few generating functions GFL(z;n) are:

GFL(z;2) = (6*(z^2-1)*GFL(z;1)+(1))/18

GFL(z;3) = (60*(z^4-10*z^2+9)*GFL(z;1)+(-107+10*z^2))/2700

GFL(z;4) = (1260*(z^6-35*z^4+259*z^2-225)*GFL(z;1)+(59845-7497*z^2+210*z^4))/ 1984500

restart; nmax:=10; mmax:=nmax: jn:=nmax: im:=nmax: for n from 1 to nmax do for i from 2 to im do cfn2[i, 1]:=0 end do: for j from 1 to jn do cfn2[1, j]:=1 end do: for j from 2 to jn do for i from 2 to im do cfn2[i, j]:= cfn2[i, j-1] + cfn2[i-1, j-1]*(2*j-3)^2 end do end do: Delta[n-1]:= sum((1-2^(2*k-1))*(-1)^(n+1)*(-bernoulli(2*k)/(2*k))*(-1)^(k+n)*cfn2[n-k+1, n], k=1..n)/ (2*4^(n-1)*(2*n-1)!) end do: for n from 1 to nmax do LG1[ -2, n]:= (-1)^(n+1)*4*Delta[n-1]* 4^(2*n-2)/binomial(2*n-2, n-1) od: for n from 1 to nmax do LGx[ -2, n]:=LG1[ -2, n] od: for m from 0 to mmax do LGx[2*m, 1]:= 2 od: for n from 2 to nmax do for m from 0 to mmax do LGx[2*m, n]:=LGx[2*m-2, n-1]/((2*n-3)*(2*n-1)) - (2*n-3)*LGx[2*m, n-1]/(2*n-1) od: od: for n from 2 to nmax do s(n):=0; for m from 0 to mmax-1 do s(n):=s(n)+LGx[2*m, n] od: od: seq(s(n), n=2..nmax);

restart; nmax:=5; ncol:=3; Digits:=20: mmax:=nmax: for m from 1 to mmax do LG1[ -2*m, 1]:=(((2^(2*m-1)-1)*bernoulli(2*m)/m)) od: LG1[0, 1]:=evalf(gamma): for m from 2 to mmax do LG1[2*m-2, 1]:= evalf(2*(1-2^(-2*m+1))*Zeta(2*m-1)) od: jn:=nmax: im:=nmax: for n from 1 to nmax do for i from 2 to im do cfn2[i, 1]:=0 end do: for j from 1 to jn do cfn2[1, j]:=1 end do: for j from 2 to jn do for i from 2 to im do cfn2[i, j]:= cfn2[i, j-1] + cfn2[i-1, j-1]*(2*j-3)^2 od: od: od: for m from -mmax+ncol-1 to mmax-1 do LG1[2*m, ncol]:=sum((-1)^(k+1)*cfn2[k, ncol]* LG1[2*m-(2*ncol-2*k), 1], k=1..ncol)/(doublefactorial(2*ncol-3)*doublefactorial(2*ncol-1)) od;

See A162449 for the denominators of the column sums.

The LAMBDA(z, n) polynomials and the LS1 matrix lead to the Lambda triangle A160487.

The CFN2(z, n), the cfn2(n, m) and the LG2 matrix lead to A008956.

The pg(n) and hg(n) sequences lead to A160476.

The LG1[ -2, n] lead to A002197, A002198, A061549 and A001790.

Cf. A001620 (gamma) and A079484 ((2n-1)!!*(2n+1)!!).

Cf. A162440 (EG1 matrix), A162443 (BG1 matrix) and A162446 (ZG1 matrix)

Sequence in context: A110776 A058392 A100369 this_sequence A024150 A052071 A157841

Adjacent sequences: A162445 A162446 A162447 this_sequence A162449 A162450 A162451

easy,frac,sign

Johannes W. Meijer (meijgia(AT)hotmail.com), Jul 06 2009