/**
* Estimate tolerance vectors for an AdaptativeStepsizeIntegrator.
*
* <p>The errors are estimated from partial derivatives properties of orbits, starting from a
* scalar position error specified by the user. Considering the energy conservation equation V =
* sqrt(mu (2/r - 1/a)), we get at constant energy (i.e. on a Keplerian trajectory):
*
* <pre>
* V² r |dV| = mu |dr|
* </pre>
*
* So we deduce a scalar velocity error consistent with the position error. From here, we apply
* orbits Jacobians matrices to get consistent errors on orbital parameters.
*
* <p>The tolerances are only <em>orders of magnitude</em>, and integrator tolerances are only
* local estimates, not global ones. So some care must be taken when using these tolerances.
* Setting 1mm as a position error does NOT mean the tolerances will guarantee a 1mm error
* position after several orbits integration.
*
* @param dP user specified position error (m)
* @param orbit reference orbit
* @return a two rows array, row 0 being the absolute tolerance error and row 1 being the relative
* tolerance error
* @exception PropagationException if Jacobian is singular
*/
public static double[][] tolerances(final double dP, final Orbit orbit)
throws PropagationException {
return NumericalPropagator.tolerances(dP, orbit, OrbitType.EQUINOCTIAL);
}
