Write . The graph is a translated hyperbola, with vertical asymptote and horizontal asymptote . It passes through , lies below for and above it for , and has derivative on each component of its domain. The left panel below shows both branches.
Figure 1.
The rational hyperbola and the convex graph of the logarithmic product
. Details.
For real square-integrable , nonnegativity of gives a quadratic in the real number . If , choose to obtain
If , then almost everywhere and the same inequality is immediate. This proves the integral Cauchy-Schwarz inequality.
Put and . The forward speed satisfies , so . The change of variables gives . Apply the Cauchy-Schwarz inequality to and the constant function one:
Equality holds exactly when is constant almost everywhere. Equivalently, the time and distance averages of speed satisfy , which identifies the excess as a measure of speed variation.
For functions with the required derivatives, the Leibniz rule is
At it is immediate. If it holds at , differentiating each summand gives two sums. The coefficient of after regrouping is by Pascal's identity, including the two endpoint terms. This proves the formula by mathematical induction.
Apply it to and then expand the derivatives of . For , the resulting coefficient is . Consequently
where range over nonnegative integers. This is the three-factor generalized Leibniz rule.
If , divide numerator and denominator by . The expression becomes , which tends to since . If , the same argument gives . If , the expression equals for every . Thus
This is the two-term case of exponential concentration of a weighted mean, and holds for negative as well as positive parameters.
The coefficient matrix has determinant
A homogeneous linear system has a nonzero solution exactly when this is a singular matrix. Therefore
At this value all solutions are , with for a nonzero solution.

There are unlisted articles, also show them or only show them.