Blog Archive

Showing posts with label Palomar Registry. Show all posts
Showing posts with label Palomar Registry. Show all posts

Thursday, November 19, 2026

Cartesian max norm – universal Lipschitz constant for Newtonian gravity

Cartesian max norm – universal Lipschitz constant for Newtonian gravity

Let's consider N bodies. We do not know their exact masses, but we can estimate a maximum value M for each of them.

Let's define cubic volumes (side length 2d) around these bodies, with sides parallel to the Cartesian axes. These are like "safe exclusion zones", in the sense that we assume, somehow, that we know that  the asteroids remain outside all these cubes.

For the exact Newtonian gravitational field generated by these bodies, we have three related mathematical results.

The important point is that we use the Cartesian maximum norm (the maximum absolute value among the three coordinates), rather than the usual Euclidean norm. The exclusion zones are cubes rather than spheres.

Let's define:

delta_g = max(|gx(P)-gx(Q)|, |gy(P)-gy(Q)|, |gz(P)-gz(Q)|)

delta_pos = max(|Px-Qx|, |Py-Qy|, |Pz-Qz|)

where gx, gy, gz are the Cartesian components of gravitational acceleration, and Px, Py, Pz are the coordinates of P (similarly for Q).

1. Local sensitivity for N=1 (sharp)

Let's first consider a single body of mass m.

For an infinitesimally small movement of the observation point outside its excluded cube, the change in gravitational acceleration is bounded, to first order, by:

delta_g <= (G * m * C3 / d^3) * delta_pos

where G is the gravitational constant and C3 is approximately 2.57756.

The coefficient C3 is optimal (or "sharp"), meaning that no smaller constant would work for all admissible positions.

Note that if we had worked with the Euclidean norm and spherical exclusion zones, the value of the constant would have been 2, because the maximum local derivative of the gravitational field is along the radial direction.

2. Local sensitivity for N bodies

If we consider N bodies, with each mass not greater than M, the total gravitational field is the sum of the individual fields.

We obtain the upper bound:

delta_g <= (N * G * M * C3 / d^3) * delta_pos

This bound is valid for infinitesimally small movements of the observation point outside all the excluded cubes, assuming the masses and positions of the N bodies are fixed.

The coefficient is also optimal when considering all possible configurations, although for a particular configuration of N bodies the bound may be conservative.

Note again that, if we had worked with the Euclidean norm and spherical exclusion zones, the value of the sharp local constant would have been 2.

3. Global sensitivity for N bodies

Furthermore, the interesting result is that exactly the same inequality holds globally, for ANY two points P and Q outside all the excluded cubes, no matter how far apart they are:

delta_g <= (N * G * M * C3 / d^3) * delta_pos

This remains true even if the straight line between P and Q crosses one of the excluded cubes.

The coefficient C3 is still optimal in this global case, in the sense that no smaller coefficient would work universally. However, for a particular configuration of N bodies, the bound may be conservative.

Note: if we had worked with the Euclidean norm and spherical exclusion zones, I believe the constant is also 2 globally, but I have not formally verified that case.

The point here is the Cartesian maximum norm and the cubic exclusion zones, especially the global result.

Possible practical interest ?

Is this just a mathematical curiosity, or could it be interesting for numerical methods that divide space into small cubic cells, perhaps to obtain more reliable bounds on gravitational variations?

The mathematical result has been formally verified in Lean and is registered in the Palomar Registry:

https://palomar-registry.org/entry?id=PALOMAR-2026-11-19-000007&version=1

The related GitHub repository, with the Lean proofs, is here:

https://github.com/alexodas/NewtonianLipschitzOptimality