I have been chugging through some proofs regarding the dilogarithm, also known as [Spencer's function][1].
\begin{alignat}{2} & \operatorname{Li}_2(z) + \operatorname{Li}_2(-z) = \frac{1}{2} \operatorname{Li}_2(z^2) && \text{(Double Identity)} \tag{1} \\ & \operatorname{Li}_2(z) + \operatorname{Li}_2(1-z) = \frac{\pi^2}{6} - \log z \log (1 - z) \ && \text{(Eulers reflection formula)} \tag{2} \\ & \operatorname{Li}_2(-z) + \operatorname{Li}_2\left( \frac{z}{1+z} \right) = -\frac{1}{2} \log^2(z+1) && \text{(Landen's Identity)} \tag{3} \\ & \operatorname{Li}_2(z) + \operatorname{Li}_2\left( \frac{1}{z} \right) = - \frac{\pi^2}{6} - \frac{1}{2}\log^2(-z) \ \ && \text{(Inversions formula)} \tag{4} \end{alignat} With the basis of $\text{Abel's Identity}$ (see Proving Abel's identity for the Dilogarithm.). I have been able to prove all the the identities except the last. (My proof for $(3)$, was somewhat convoluted, so hints there would be appreciated..). I have yet to prove $(4)$, although I have given it two attempts below
Attempt 1
In the Abel Identity let $x=y=1-z$ and divide by $2$ to obtain $$ \frac{1}{2}\log(-z)^2 = \operatorname{Li}_2\left( -\frac{1+z}{z}\right) - \frac{1}{2} \operatorname{Li}_2\left( \left[ \frac{1+z}{z} \right]^2 \right) - \operatorname{Li}_2(1+z) $$ By now using $(1)$ with $z = [1+1/z]^2$ and inserting it into the equation above I obtain $$ \frac{1}{2}\log(-z)^2 = - \left[ \operatorname{Li}_2\left( \frac{1}{z} + 1\right)- \operatorname{Li}_2(1 + z)\right] $$ and from here I am stuck. It is close to what I want but I can not find any way to transform the right hand side.
Attempt 2
From chat the suggestion was to instead look at the integral definition, this gives \begin{align*} \operatorname{Li}_2(z) + \operatorname{Li}_2\left( \frac{1}{z}\right) & = - \int_0^z \frac{\log(1-t)}{t}\,\mathrm{d}t - \int_0^{1/z} \frac{\log(1-t)}{t}\,\mathrm{d}t \\ & = - \int\limits_0^1 {\frac{{\log \left( {z - t} \right) + \log \left( {1 - zt} \right) - \log z}}{t}dt} \\ & = -\zeta(2) + \int_0^1 \frac{\log z - \log(1-zt)}{t}\,\mathrm{d}t \end{align*} The last step used that $$ \int_0^1 \frac{\log(1-t)}{t}\,\mathrm{d}t = \int_0^1 -\frac{1}{t} \sum_{n=1}^\infty \frac{t^n}{n} \,\mathrm{d}t = \sum_{n=1}^\infty \frac{1}{n^2} = \zeta(2) $$ And this is where I stopped. I think this argument can be finished by series expansion, but I got lost in the algebra. If possible I would very much like to prove this identity from Abel's Identity and the three equations stated above. Any hint or solutions is much appreciated as always =) [1]: http://en.wikipedia.org/wiki/Spence%27s_function