Let
Is differentiable at ? Is it differentiable at ?
Find the radius of convergence and a closed form. The coefficients are , so the radius is . For , factor out and use the logarithm series:
with the removable value . Having a closed form is what makes the endpoint questions answerable, since the differentiated series behaves badly there.
Settle by checking convergence first. At the series becomes , the harmonic series, which diverges. So does not exist as a number, and the question of differentiability at does not even arise. The closed form agrees: as .
Settle the value at . Here the series is , the alternating harmonic series, convergent by Leibniz's test with sum
Abel's theorem guarantees this equals , so the closed form extends continuously to .
Differentiate the closed form and evaluate at . From ,
so
Because and , the closed form is real-analytic on a whole neighbourhood of , which is why this one-sided derivative genuinely exists.
Note the trap and check numerically. Differentiating the series term by term gives , whose terms at have magnitude and therefore do not tend to — the differentiated series diverges at . That does not contradict the result: it only shows term-by-term differentiation is invalid at the endpoint, while the function itself is still differentiable there. Numerically, and .
Need to solve a different problem like this? Open the solver →