PhyslibAlpha.ClassicalMechanics.NortonDome.Sqrt
Calculus of the real square root near zero
i. Overview
Two facts about the real square root at zero, where Mathlib's calculus does not reach: the cube `√y ^ 3` is differentiable everywhere, with derivative `(3/2) √x`, and `√` is not Lipschitz on any interval `[0, ε]`. The first gives the derivative of the potential of the Norton dome, the second the failure of the Picard–Lindelöf hypothesis for its force.
ii. Key results
- `hasDerivAt_sqrt_pow_three` is the derivative of `√y ^ 3` at every real number.
- `not_lipschitzOnWith_sqrt` is the failure of the Lipschitz property of `√` on `[0, ε]`.
iii. Table of contents
- A. The derivative of the cube of the square root
- B. The square root is not Lipschitz at zero
iv. References
- Norton, J. D., *The dome: an unexpectedly simple failure of determinism*, Philosophy of Science 75 (2008), 786–798.
A. The derivative of the cube of the square root
To the left of zero the square root vanishes, to the right the chain rule applies, and at zero the difference quotient of `√t ^ 3 = t √t` is `√t`, which tends to zero.
B. The square root is not Lipschitz at zero
A bound `√y ≤ K y` on `[0, ε]` gives `1 ≤ K √y` for every positive `y ≤ ε`, which fails at `y = min ε (1 / (2K + 1)²)`.
2 declarations
The derivative of is for all
For any real number , the function is differentiable at with a derivative equal to . In particular, although the square root function itself is not differentiable at , its cube is differentiable at with derivative . (Note: for , is defined to be ).
The real square root is not Lipschitz on for
For any non-negative real number and any , the real square root function is not Lipschitz continuous with constant on the closed interval .
