14
$\begingroup$

I asked a question at M.SE a couple of years ago about polylogarithms $\!^{[1]}$$\!^{[2]}$$\!^{[3]}$$\!^{[4]}$ where I conjectured $$720\,\operatorname{Li}_4\!\left(\tfrac12\right)-2160\,\operatorname{Li}_4\!\left(\tfrac13\right)+2160\,\operatorname{Li}_4\!\left(\tfrac23\right)+270\,\operatorname{Li}_4\!\left(\tfrac14\right)+540\,\operatorname{Li}_4\!\left(\tfrac34\right)+135\,\operatorname{Li}_4\!\left(\tfrac19\right)\\ =19\pi^4+30\left(\pi^2-\beta^2\right)\left(10\alpha^2-12\alpha\beta+3\beta^2\right)-30\,\alpha^2\left(19\alpha^2-24\alpha\beta+8\beta^2\right)\!,\\ \text{where $\alpha=\ln2,\,\,\beta=\ln3$.}$$ I put a bounty on it recently, but it still remains unanswered. So, I decided to re-post it here. Also, I recently discovered several similar conjectures for $\operatorname{Li}_5(z)$ involving powers of the golden ratio $\phi=\frac{1+\sqrt5}2$: $$4455\,\operatorname{Li}_5\left(\phi^{-2}\right)-1215\,\operatorname{Li}_5\left(\phi^{-4}\right)-360\,\operatorname{Li}_5\left(\phi^{-6}\right)+15\,\operatorname{Li}_5\left(\phi^{-12}\right)\\ =3015\,\zeta(5)-702\ln^5\phi+180\,\pi^2\ln^3\phi-38\,\pi^4\ln\phi$$


$$45000\,\operatorname{Li}_5\left(\phi^{-2}\right)-16875\,\operatorname{Li}_5\left(\phi^{-4}\right)-144\,\operatorname{Li}_5\left(\phi^{-10}\right)+9\,\operatorname{Li}_5\left(\phi^{-20}\right)\\ =28944\,\zeta(5)-6000\ln^5\phi+1600\,\pi^2\ln^3\phi-356\,\pi^4\ln\phi$$


$$15660\,\operatorname{Li}_5\left(\phi^{-2}\right)-19440\,\operatorname{Li}_5\left(\phi^{-4}\right)+7680\,\operatorname{Li}_5\left(\phi^{-6}\right)+2430\,\operatorname{Li}_5\left(\phi^{-8}\right)-15\,\operatorname{Li}_5\left(\phi^{-24}\right)=5025\,\zeta(5)+2088\ln^5\phi-240\,\pi^2\ln^3\phi-28\,\pi^4\ln\phi$$

How can we prove these conjectures? Are there other similar identities of this kind?

$\endgroup$
10
$\begingroup$

(Updated answer):

Upon further research, it turns out your three equations involving $\phi$ are special cases of three polylogarithm ladders of index $12,\,20,\,24$ that can be found in "The Polylogarithm in Algebraic Number Fields" (1985) by M. Abouzahra and L. Levin. Let $\color{red}{\rho = \frac1{\phi}}$.

I. Ladder of index 12

$$A = \frac{\text{Li}_n(\rho^{12})}{12^{n-1}}- \frac32\frac{\text{Li}_n(\rho^6)}{6^{n-1}}- \frac{\text{Li}_n(\rho^4)}{4^{n-1}}+ \frac{11}{48}\frac{\text{Li}_n(\rho^2)}{2^{n-1}}$$

$$B = -\frac{13}{48}\frac{\ln^n\rho}{n!}+\frac{\zeta(2)}{48}\frac{\ln^{n-2}\rho}{(n-2)!}-\frac{19\,\zeta(4)}{1728}\frac{\ln^{n-4}\rho}{(n-4)!} $$

then for $n=1,2,3,4,5$,

$$A+B= \frac{67\,\zeta(5)}{6912}\frac{\ln^{n-5}\rho}{(n-5)!} $$

with $A+B = 0$ for $n<5$. If $n=5$, then we recover your experimentally found equation. Incidentally, a consequence (or is it the other way round?) of this ladder is the high-degree cyclotomic equation satisfied by the golden ratio,

$$(1 - \phi^{-12}) = (1 - \phi^{-6})^{3/2}(1 - \phi^{-4})(1 - \phi^{-2})^{-11/48}\,\phi^{13/48}$$

Notice that the powers of $(1 - \phi^{k})$ are coefficients in $A,B$.

II. Ladder of index 20

Similar to the ladder above. For $n=5$ yields the OP's second formula but, with modifications, can be extended to $n=7$.

III. Ladder of index 24

Likewise. If modified, can be extended even to $n=9$. This also explains the index $24$ formulas below for $n=6,7$ that I found empirically.



(Old answer): As to your question if there are similar identities, it seems there is a whole family,

$$5\,\text{Li}_2(\phi^{-1}) =-5\ln^2\phi+\,3\,\color{blue}{\zeta(2)}$$


$$15\,\text{Li}_3(\phi^{-2})=10\ln^3\phi\,-2\pi^2\ln\phi+12\,\color{blue}{\zeta(3)}$$


$$144\text{Li}_4(\phi^{-1})-9\text{Li}_4(\phi^{-2})-64\text{Li}_4(\phi^{-3})+4\text{Li}_4(\phi^{-6})=27\ln^4\phi-6\pi^2\ln^2\phi+80\,\color{blue}{\zeta(4)}$$


$$4455\,\text{Li}_5\left(\phi^{-2}\right)-1215\,\text{Li}_5\left(\phi^{-4}\right)-360\,\text{Li}_5\left(\phi^{-6}\right)+15\,\text{Li}_5\left(\phi^{-12}\right)\\ =-702\ln^5\phi+180\,\pi^2\ln^3\phi-38\,\pi^4\ln\phi+3015\,\color{blue}{\zeta(5)}$$


$$m_1\text{Li}_6(\phi^{-2})+m_2\text{Li}_6(\phi^{-4})+m_3\text{Li}_6(\phi^{-6})+m_4\text{Li}_6(\phi^{-8})+m_5\text{Li}_6(\phi^{-12})+m_6\text{Li}_6(\phi^{-24})\\ =m_7\ln^6\phi+m_8\pi^2\ln^4\phi+m_9\pi^4\ln^2\phi +32155\,\color{blue}{\zeta(6)}$$


$$n_1\text{Li}_7(\phi^{-2})+n_2\text{Li}_7(\phi^{-4})+n_3\text{Li}_7(\phi^{-6})+n_4\text{Li}_7(\phi^{-8})+n_5\text{Li}_7(\phi^{-12})+n_6\text{Li}_7(\phi^{-24})\\ =n_7\ln^7\phi+n_8\pi^2\ln^5\phi+n_9\pi^4\ln^3\phi+n_{10}\pi^6\ln\phi +44689995\,\color{blue}{\zeta(7)}$$


and where the integers $m_i$ and $n_i$ were found using Integer Relations but are too tedious to write here.

Does this go on forever? Note the exponents of $\phi$ in $\text{Li}_k(\phi^{-n})$ match up for $k=4$ as $c=1\times6=2\times3$ then for $k=5$ as $c=2\times12=6\times4$ then for $k=6,7$ as $c=2\times24=4\times12 = 6\times8$.

P.S. Based on Abouzahra and Levin's paper cited in the update, I guess this can't be brought to arbitrarily high $k$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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