Travou nisso? Resolva com IA
Calculus · real student question

Is the dilogarithm Li2(x) differentiable at x = 1? Justify your answer.

Question

Is the dilogarithm

Li⁡2(x)=∑n=1∞xnn2\operatorname{Li}_{2}(x)=\sum_{n=1}^{\infty}\frac{x^{n}}{n^{2}}

differentiable at x=1x=1? Justify your answer.

Step-by-step solution

  1. Check that the function is even defined at x=1x=1. The series has radius of convergence 11, and at the endpoint x=1x=1 it becomes the convergent pp-series with p=2p=2:

    Li⁡2(1)=∑n=1∞1n2=π26≈1.644934.\operatorname{Li}_{2}(1)=\sum_{n=1}^{\infty}\frac{1}{n^{2}}=\frac{\pi^{2}}{6}\approx1.644934.

    So the function is defined there, and by Abel's theorem it is also continuous from the left. Differentiability is a strictly stronger requirement.

  2. Differentiate the series inside the disc. For ∣x∣<1|x|<1 term-by-term differentiation is valid:

    Li⁡2′(x)=∑n=1∞n xn−1n2=1x∑n=1∞xnn.\operatorname{Li}_{2}'(x)=\sum_{n=1}^{\infty}\frac{n\,x^{n-1}}{n^{2}}=\frac{1}{x}\sum_{n=1}^{\infty}\frac{x^{n}}{n}.

    The inner sum is the Mercator series −ln⁡(1−x)-\ln(1-x), so

    Li⁡2′(x)=−ln⁡(1−x)x,∣x∣<1.\operatorname{Li}_{2}'(x)=-\frac{\ln(1-x)}{x},\qquad |x|<1.

    Numerical check at x=0.99x=0.99: a central difference of the series gives 4.6516874.651687 and −ln⁡(0.01)0.99=4.651687-\frac{\ln(0.01)}{0.99}=4.651687 ✓.

  3. Examine the derivative as x→1−x\to1^{-}. As xx approaches 11 from below, 1−x→0+1-x\to0^{+} so ln⁡(1−x)→−∞\ln(1-x)\to-\infty, hence

    −ln⁡(1−x)x⟶+∞.-\frac{\ln(1-x)}{x}\longrightarrow+\infty.

    The slope grows without bound: at x=0.9x=0.9 it is 2.562.56, at x=0.99x=0.99 it is 4.654.65, at x=0.999x=0.999 it is 6.916.91 — increasing like ln⁡11−x\ln\frac{1}{1-x}, slowly but without limit.

  4. Conclude with the definition of the derivative. Since Li⁡2\operatorname{Li}_{2} is differentiable on (0,1)(0,1) with derivative tending to +∞+\infty, the mean value theorem forces the difference quotient Li⁡2(1)−Li⁡2(x)1−x\frac{\operatorname{Li}_{2}(1)-\operatorname{Li}_{2}(x)}{1-x} to diverge as well. So the left derivative is +∞+\infty and no finite derivative exists:

    Li⁡2 is continuous but not differentiable at x=1.\operatorname{Li}_{2}\text{ is continuous but not differentiable at }x=1.

    Geometrically the graph has a vertical tangent there.

  5. See it in the local expansion. Landen's identity Li⁡2(x)+Li⁡2(1−x)=π26−ln⁡xln⁡(1−x)\operatorname{Li}_{2}(x)+\operatorname{Li}_{2}(1-x)=\frac{\pi^{2}}{6}-\ln x\ln(1-x) gives, with y=1−x→0+y=1-x\to0^{+},

    Li⁡2(x)=π26+(1−x)ln⁡(1−x)−(1−x)+O ⁣((1−x)2).\operatorname{Li}_{2}(x)=\frac{\pi^{2}}{6}+(1-x)\ln(1-x)-(1-x)+O\!\left((1-x)^{2}\right).

    Numerically at y=0.001y=0.001: the expansion gives 1.637026311.63702631 against the true 1.637022611.63702261 ✓. The yln⁡yy\ln y term is what does the damage — its derivative ln⁡y+1\ln y+1 blows up, confirming the failure of differentiability from a second direction.

Answer

No. Li⁡2(1)=π26 and Li⁡2 is continuous at 1, but Li⁡2′(x)=−ln⁡(1−x)x→+∞ as x→1−, so no finite derivative exists.\text{No. }\operatorname{Li}_{2}(1)=\frac{\pi^{2}}{6}\text{ and }\operatorname{Li}_{2}\text{ is continuous at }1,\text{ but }\operatorname{Li}_{2}'(x)=-\frac{\ln(1-x)}{x}\to+\infty\text{ as }x\to1^{-},\text{ so no finite derivative exists.}

Need to solve a different problem like this? Open the solver →