Answer: The Integral That Swaps Its Limits
Key Idea / Intuition
The integrand has no elementary antiderivative in terms of x alone โ but it has a hidden parameter waiting to be introduced. The classic Feynman/Leibniz trick is to replace the numerator xโ1 with xtโ1 for a parameter t, differentiate under the integral sign with respect to t, evaluate the resulting clean integral, then integrate back in t. The result is a logarithm that evaluates to ln2.
Formal Proof / Solution
Step 1: Introduce a parameter.
Define
I(t)=โซ01โlnxxtโ1โdx,tโฅ0.
Note I(0)=0 and I(1)=I (our target).
Step 2: Differentiate under the integral sign.
Iโฒ(t)=dtdโโซ01โlnxxtโ1โdx=โซ01โโtโโ(xtโ1)โ
lnx1โdx.
Since โtโโxt=xtlnx, we get
Iโฒ(t)=โซ01โlnxxtlnxโdx=โซ01โxtdx=t+11โ.
Step 3: Integrate back.
I(t)=โซ0tโs+11โds+C=ln(t+1)+C.
Using I(0)=0: C=0. So
I(t)=ln(t+1).
Step 4: Evaluate at t=1.
I=I(1)=ln2.
Verification of the trick's validity: The interchange of differentiation and integration is justified by dominated convergence on [0,1], since โฃxtlnxโฃ is integrable uniformly for t in any compact set. Also note that lnxxtโ1โโ0 as xโ0+ and equals t at x=1 by L'Hรดpital, so the integrand is bounded and continuous on (0,1].
Answer:
I=ln2โ