5
$\begingroup$

(For brevity, the level-6 functions have been migrated to another post.)

I. Level-10 functions

Given the Dedekind eta function $\eta(\tau)$. To recall, for level-6,

$$j_{6A} = \left(\sqrt{j_{6B}} + \frac{\color{blue}{-1}}{\sqrt{j_{6B}}}\right)^2 =\left(\sqrt{j_{6C}} + \frac{\color{blue}8}{\sqrt{j_{6C}}}\right)^2 = \left(\sqrt{j_{6D}} + \frac{\color{blue}9}{\sqrt{j_{6D}}}\right)^2-4$$

For level-10,

$$j_{10A} = \left(\sqrt{j_{10D}} + \frac{\color{blue}{-1}}{\sqrt{j_{10D}}}\right)^2 = \left(\sqrt{j_{10B}} + \frac{\color{blue}4}{\sqrt{j_{10B}}}\right)^2 = \left(\sqrt{j_{10C}} + \frac{\color{blue}5}{\sqrt{j_{10C}}}\right)^2-4$$

where,

\begin{align} j_{10B}(\tau) &= \left(\frac{\eta(\tau)\,\eta(5\tau)}{\eta(2\tau)\,\eta(10\tau)}\right)^{4}\qquad \\ j_{10C}(\tau) &= \left(\frac{\eta(\tau)\,\eta(2\tau)}{\eta(5\tau)\,\eta(10\tau)}\right)^{2} \\ j_{10D}(\tau) &= \left(\frac{\eta(2\tau)\,\eta(5\tau)}{\eta(\tau)\,\eta(10\tau)}\right)^{6} \\ j_{10E}(\tau) &= \left(\frac{\eta(2\tau)\,\eta^5(5\tau)}{\eta(\tau)\,\eta^5(10\tau)}\right) \end{align}

Conway and Norton found these moonshine functions obey (or a version thereof),

$$j_{10A}+2j_{10E} = j_{10B}+j_{10C}+j_{10D}+6$$


II. Sequences

Just like for level-6, we can use the relations above to get four sequences. In Cooper's paper, "Level 10 Analogues of Ramanujan's series for 1/Pi", he discussed $s_{10}=s_{10A}$ and two related sequences found by Zudilin (p.10), but not the other three below,

\begin{align} s_{10A}(k) &=\sum_{m=0}^k \binom{k}{m}^4\\ s_{10D}(j) &=\sum_{k=0}^j (-u)^{j-k}\binom{j+k}{j-k}\,s_{10A}(k)\\ s_{10B}(j) &=\sum_{k=0}^j (-v)^{j-k}\binom{j+k}{j-k}\,s_{10A}(k)\\ s_{10C}(n) &=\sum_{j=0}^n\sum_{k=0}^j (-w)^{n-j}\binom{n+j}{n-j}\binom{j}{k}\binom{2j}{j}\binom{2k}{k}^{-1}s_{10A}(k) \end{align}

where $u = \color{blue}{-1}$, $v = \color{blue}4$, $w = \color{blue}5$. Using the variable $h$ for uniformity, the first few terms are,

\begin{align} s_{10A}(h) &=1, 2, 18, 164, 1810, 21252,\ldots\\ s_{10D}(h) &=1, 3, 25, 267, 3249, 42795, 594145,\ldots\\ s_{10B}(h) &=1, -2, 10, -68, 514, -4100, 33940,\ldots\\ s_{10C}(h) &=1, -1, 1, -1, 1, 23, -263, 1343, -2303,\ldots \end{align}

such that all $s_{10}(0) = 1.$ The sequences $(s_{10A}, s_{10D}, s_{10B}, s_{10C})$ have an $m$-term recurrence relation with $m=3,5,5,7$ (with the last one courtesy of G. Edgar's answer below).


III. Pi formulas

A. These four sequences can be used to generate new Ramanujan-Sato formulas for $1/\pi$ of level 10. For example, let $\tau = \sqrt{-19/10}$, then,

\begin{align} j_{10A}(\tau) &= 76^2\\ j_{10D}(\tau) &= (2+\sqrt5)^6\\ j_{10B}(\tau) &= 4(3+\sqrt{10})^4\\ j_{10C}(\tau) &= 5(1+\sqrt{2})^8\qquad \end{align}

to get (the first one is known),

\begin{align} \frac1{\pi} &= \frac{5}{\sqrt{95}}\,\sum_{n=0}^\infty s_{10A}(n)\,\frac{\;408n+47}{(76^2)^{n+1/2}}\\[4pt] \frac1{\pi} &= \frac{2\sqrt{95}}{17\sqrt{5}}\sum_{n=0}^\infty s_{10D}(n)\,\frac{408n+47-\psi_1}{\big((2+\sqrt5)^6\big)^{n+1/2}}\\[4pt] \frac1{\pi} &= \frac{\sqrt{95}}{6\sqrt{10}}\sum_{n=0}^\infty s_{10B}(n)\,\frac{408n+47+\psi_2\;}{\big(4(3+\sqrt{10})^4\big)^{n+1/2}}\\[4pt] \frac1{\pi} &= \frac{1}{\sqrt{95}}\;\sum_{n=0}^\infty s_{10C}(n)\,\frac{An+B+\psi_3}{\;\big(5(1+\sqrt{2})^8\big)^{n+1/2}}\\[4pt] \end{align}

where $\psi_1 = \frac{157}{38(2+\sqrt5)^3},$ and $\psi_2 = \frac{157}{19(3+\sqrt{10})^2}.$ (The fourth to be added later.)

B. Furthermore, if within the radius of convergence, it seems that,

$$\sum_{h=0}^\infty s_{10A}(h)\,\frac{1}{\;\big(j_{10A}\big)^{h+1/2}} = \sum_{h=0}^\infty s_{10B}(h)\,\frac{1}{\;\big(j_{10B}\big)^{h+1/2}} = \\ \sum_{h=0}^\infty s_{10C}(h)\,\frac{1}{\;\big(j_{10C}\big)^{h+1/2}} = \sum_{h=0}^\infty s_{10D}(h)\,\frac{1}{\;\big(j_{10D}\big)^{h+1/2}}\;$$


IV. Questions

  1. What is the recurrence relation for $s_{10C}$?
  2. Using the four given sequences of level $10$, is the last relation really true? And do their closed-forms have simpler versions, just like for level-6?
$\endgroup$
5
  • $\begingroup$ You could include $j_{10A} = j_{10C} + 6 + 25/j_{10C}.$ $\endgroup$
    – Somos
    Commented May 12, 2023 at 22:23
  • $\begingroup$ @Somos Done. And I have migrated level 6 to another post since their sequences are more well-studied anyway. $\endgroup$ Commented Jun 13, 2023 at 16:07
  • $\begingroup$ @Somos The relations for levels 12, 18, 30 are nicely consistent. Kindly see this new post. $\endgroup$ Commented Jun 14, 2023 at 11:14
  • $\begingroup$ Thanks for the link. You may be interested in the new improved version of my code for sequence recursion here. $\endgroup$
    – Somos
    Commented Jun 14, 2023 at 11:53
  • $\begingroup$ @Somos I’ll see if i can figure it out. Even your original code which i’ve saved was of great help. With it, i found that the four level-10 and four level-12 sequences both have $m$-term recurrences for $m = 3, 6,6,7$. It’s nice to be able to ascertain these things. $\endgroup$ Commented Jun 14, 2023 at 12:24

3 Answers 3

3
$\begingroup$

For $s_{10C}$, Maple finds this $7$-term recurrence:

{(15625*n^3 + 46875*n^2 + 46875*n + 15625)*u(n) + (11250*n^3 + 61875*n^2 + 115625*n + 73125)*u(n + 1) + (4575*n^3 + 36600*n^2 + 99150*n + 90950)*u(n + 2) + (1116*n^3 + 11718*n^2 + 41434*n + 49322)*u(n + 3) + (183*n^3 + 2379*n^2 + 10371*n + 15157)*u(n + 4) + (18*n^3 + 279*n^2 + 1445*n + 2501)*u(n + 5) + (n^3 + 18*n^2 + 108*n + 216)*u(n + 6), u(0) = 1, u(1) = -1, u(2) = 1, u(3) = -1, u(4) = 1, u(5) = 23}

\begin{align} 0 = &\left( 15625\,{n}^{3}+46875\,{n}^{2}+46875\,n+15625 \right) u \left( n \right) \\ &+ \left( 11250\,{n}^{3}+61875\,{n}^{2}+115625\,n+ 73125 \right) u \left( n+1 \right) \\ &+ \left( 4575\,{n}^{3}+36600\,{n}^{ 2}+99150\,n+90950 \right) u \left( n+2 \right) \\ &+ \left( 1116\,{n}^{3}+ 11718\,{n}^{2}+41434\,n+49322 \right) u \left( n+3 \right) \\ &+ \left( 183\,{n}^{3}+2379\,{n}^{2}+10371\,n+15157 \right) u \left( n+4 \right) \\ &+ \left( 18\,{n}^{3}+279\,{n}^{2}+1445\,n+2501 \right) u \left( n+5 \right) \\ &+ \left( {n}^{3}+18\,{n}^{2}+108\,n+216 \right) u \left( n+6 \right) , \\ &u \left( 0 \right) =1,u \left( 1 \right) =-1,u \left( 2 \right) =1,u \left( 3 \right) =-1,u \left( 4 \right) =1,u \left( 5 \right) =23 \end{align}
or \begin{align} 0 = & \,5^6\, \left( k-2 \right) ^{3}u \left( k-3 \right) \\ &+ 5^4 \left( 18\,k^3 - 63\,k^2 + 77\,k -33 \right) u \left( k-2 \right) \\ &+ 5^2 \left( 183\,k^3 - 183\,k^2 + 123\,k - 25 \right) u \left( k-1 \right) \\ &+2\, \left( 2\,k+1 \right) \left( 279\,{k}^{2}+279\,k+175 \right) u \left( k \right) \\ &+ \left( 183\,{k}^{3}+732\,{k}^{2}+1038\,k +514 \right) u \left( k+1 \right) \\ &+ \left( 18\,{k}^{3}+117\,{k}^{2}+ 257\,k+191 \right) u \left( k+2 \right) \\ &+ \left( k+3 \right) ^{3}u \left( k+3 \right) \end{align}

$\endgroup$
4
  • 1
    $\begingroup$ Thanks! Somos taught me some Mathematica code in this post that showed $s_{10B}$ was a $5$-term recurrence, just like the one for $s_{10D}$ you found years ago in this MO post. But I didn't know how to tweak the code for general $n$-term recurrence, and I suspected $s_{10C}$ was $n>5$. $\endgroup$ Commented May 12, 2023 at 12:38
  • 1
    $\begingroup$ I used Maple's gfun library for this. $\endgroup$ Commented May 12, 2023 at 16:37
  • $\begingroup$ It’s interesting that two of the coefficients are cubes, and the central coefficient factors just like the relation for the Apery numbers. Tried to use it in a cfrac hoping to get a new cfrac for $\zeta(3)$ or $\zeta(5)$ but no luck. $\endgroup$ Commented May 12, 2023 at 17:26
  • $\begingroup$ I just found out the $5$-term recurrences for sequences $s_{10B}$ and $s_{10D}$ (the latter of which you found years ago in this MO post) have a $6$-term alternative with a nice symmetry. See second answer to this post. $\endgroup$ Commented Jun 8, 2023 at 6:18
0
$\begingroup$

(This supplements G. Edgar's answer.) Given the integer sequences and their formulas in the post,

\begin{align} \alpha &= s_{10A}(k) =1, 2, 18, 164, 1810, 21252,\ldots\\ \beta &= s_{10B}(k) =1, -2, 10, -68, 514, -4100, 33940,\ldots\\ \gamma &= s_{10C}(k) =1, -1, 1, -1, 1, 23, -263, 1343, -2303,\ldots\\ \delta &= s_{10D}(k) =1, 3, 25, 267, 3249, 42795, 594145,\ldots \end{align}

Their four level-6 counterparts are a bit simpler, all having only a 3-term recurrence relation. The first level-10 also has a 3-term,

$$(n+1)^3\alpha_{n+1} = 2(2n+1)(3n^2+3n+1)\alpha_n + 4n(16n^2-1)\alpha_{n-1}$$

As mentioned in the post, two of the level-10 have a 5-term recurrence instead. However, turns out they have an alternative 6-term recurrence with a nice symmetry. So the beta sequence obeys,

$$0=(n+2)^3\beta_{n+2}\\ +2(12n^3+45n^2+59n+27)\beta_{n+1}\\ +2^2(52n^3+66n^2+39n+7)\beta_{n+0}\\ +2^4(52n^3-66n^2+39n-7)\beta_{n-1}\\ +2^7(12n^3-45n^2+59n-27)\beta_{n-2}\\ +2^{10}(n-2)^3\beta_{n-3}\\\\$$

with $\beta_{-5} = \beta_{-4} = \beta_{-3} = \beta_{-2} = \beta_{-1} =0$ and $\beta_{0} = 1.$ The delta sequence has,

$$0=(n+2)^3\delta_{n+2}\\ -3(7n^3+30n^2+44n+22)\delta_{n+1}\\ +2(29n^3+42n^2+18n-1)\delta_{n+0}\\ -2(29n^3-42n^2+18n+1)\delta_{n-1}\\ +3(7n^3-30n^2+44n-22)\delta_{n-2}\\ -(n-2)^3\delta_{n-3}$$

with the same initial conditions. Anybody knows the reason for such symmetry?

P.S. In G. Edgar's answer, the remaining sequence (gamma) has 7-term recurrence, but I don't know if it has an alternative one.

$\endgroup$
0
$\begingroup$

As mentioned in the related post, according to \begin{eqnarray} \sum_{k=0}^{\infty}\frac{\gamma_{k}}{\left(j_{10C}(\tau)\right)^{k}}&=&{\sqrt{\frac{j_{10C}(\tau)}{j_{10D}(\tau)}}\sum_{k=0}^{\infty}\frac{\delta_{k}}{\left(j_{10D}(\tau)\right)^{k}}}\\&=&QM_{\gamma}\gamma+O\left(q^{4}\right)\\&=&QM_{\delta}\delta+O\left(q^{4}\right)\\Q&=&\left[\begin{array}{c} 1\\ q\\ q^{2}\\ q^{3} \end{array}\right]^{T},\gamma=\left[\begin{array}{c} \gamma_{1}\\ \gamma_{2}\\ \gamma_{3}\\ \gamma_{4} \end{array}\right],\delta=\left[\begin{array}{c} \delta_{1}\\ \delta_{2}\\ \delta_{3}\\ \delta_{4} \end{array}\right]\\M_{\gamma}&=&\left[\begin{array}{cccc} 1 & 0 & 0 & 0\\ -4 & 1 & 0 & 0\\ 4 & -10 & 1 & 0\\ 0 & 43 & -16 & 1 \end{array}\right],M_{\delta}=\left[\begin{array}{cccc} 1 & 0 & 0 & 0\\ 0 & 1 & 0 & 0\\ 0 & 2 & 1 & 0\\ 0 & 7 & 4 & 1 \end{array}\right] \end{eqnarray} we will have, \begin{eqnarray} \delta&=&T\gamma\\T&=&M_{\delta}^{-1}M_{\gamma}\\&&1\\&&-4,1\\&&12,-12,1\\&&-20,84,-20,1\\&&-36,-412,220,-28,1\\&&412,1380,-1700,420,-36,1\\&&-1428,-2028,9900,-4396,684,-44,1\\&&\cdots \end{eqnarray} $T$ is a matrix and we can observe that, \begin{eqnarray} T_{n,n-0}&=&1,n=0,1,\cdots\\T_{n,n-1}&=&4-8n,n=1,2,\cdots\\T_{n,n-2}&=&32n^{2}-88n+60,n=2,3,\cdots\\T_{n,n-3}&=&-\frac{256n^{3}}{3}+576n^{2}-\frac{3800n}{3}+900,n=3,4,\cdots\\T_{n,n-4}&=&\frac{512n^{4}}{3}-\frac{6400n^{3}}{3}+\frac{29344n^{2}}{3}\\&&-\frac{58376n}{3}+14140,n=4,5,\cdots\\T_{n,n-5}&=&-\frac{4096n^{5}}{15}+\frac{16384n^{4}}{3}-\frac{128512n^{3}}{3}\\&&+\frac{493952n^{2}}{3}-\frac{1549608n}{5}+228420,n=5,6,\cdots\\\cdots \end{eqnarray} It is obvious to guess that $T_{n,n-m}$ should be a polynomial of $n$ with a degree $m$. But it is really messy, I have no idea how to go on, please feel free to give your comments.

$\endgroup$
1
  • $\begingroup$ The hardest is $\color{red}{\gamma = s_{10C}}$ since it is a triple sum. If it can be reduced to a double sum, then that would be an improvement. \begin{align} \alpha = s_{10A}(k) &=\sum_{j=0}^k\binom{k}{j}^4=1, 2, 18, 164, 1810,\ldots\\ \beta = s_{10B}(n) &= \sum_{k=0}^n\sum_{j=0}^k(-4)^{n-k}\binom{n+k}{2k}\binom{k}{j}^4 =1, -2, 10, -68, 514,\ldots\\ \color{red}{\gamma = s_{10C}(n)} &= 1, -1, 1, -1, 1, 23, -263, 1343, -2303,\ldots\\ \delta = s_{10D}(n) &= \sum_{k=0}^n\sum_{j=0}^k(+1)^{n-k}\binom{n+k}{2k}\binom{k}{j}^4 =1, 3, 25, 267, 3249,\ldots \end{align} $\endgroup$ Commented Jun 14, 2023 at 12:59

Not the answer you're looking for? Browse other questions tagged or ask your own question.