Introduce the position and velocity variables and . Then the required system of ordinary differential equations and initial condition are
Thus . Conversely, the first equation identifies , and the second recovers the original second-order ordinary differential equation and both initial data.
Work with the L1 norm on . The scalar Lipschitz hypothesis implies
We may take . The compact set and continuity give a finite bound . Choose
Let . The space of continuous maps is complete in the supremum norm: it is a closed subset of the Banach space . Define the Picard operator on a closed ball by
For any , the integral triangle inequality gives
The integrand is continuous, so the result is continuous and remains in the prescribed closed ball. Thus is a self-map of the complete space. Moreover,
It is a contraction mapping. The contraction mapping theorem gives a unique fixed point . By the fundamental theorem of calculus, the fixed-point identity makes continuously differentiable and gives and . Its first component is twice continuously differentiable, because its derivative is the second component.
Conversely, integrating any solution that remains in the ball makes it a fixed point, so it equals . A solution starting at cannot first reach the boundary of this ball during : before such an exit the same estimate gives distance at most , contradicting distance at the first exit. This proves existence and uniqueness on , not merely formal convergence of an iteration. It is the local Picard-Lindelöf theorem construction.
In position-velocity variables , the scalar right-hand side is
Its only relevant singularity is . The initial state is separated from it. For an explicit application of part (a), take and in the L1 norm. On the resulting closed ball, , and .
The partial derivatives obey
The ball is convex, so integrating these partial derivatives along a segment between two states gives the uniform Lipschitz condition
Continuity holds on the whole slab, and
Thus , , and the choice satisfies the self-map and contraction bounds in part (a). This proves a unique solution on , with bounded away from zero there.
There is also a useful exact representation. Where , set , the logarithmic derivative. Differentiation cancels the quadratic term:
Solving this first-order linear differential equation gives , and a second integration yields
This function is positive and smooth for every real , satisfies both initial data, and verifies the original ordinary differential equation by direct differentiation. It therefore extends the unique local solution to the entire real line. The logarithmic-derivative reduction of a second-order equation explains why this apparently nonlinear example is explicitly integrable by integration.

Articles by others on the same topic (0)

There are currently no matching articles.