You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
(Subproject for rigid contact in Simbody 4.0)
Witness functions currently trigger only on a sign change. Unilateral ones must trigger when the sign is wrong, regardless of its previous state. Sign changes should still be tracked when present since they can accelerate event isolation.
Witness functions need to be more flexible in several ways:
Should be able to specify as unilateral, and which sign causes triggering. Note that this implies which sign change should trigger the event, so likely we just need a boolean to say if it is unilateral.
Allow setting an event isolation requirement on the witness value as well as the time window.
Need to allow NaN for witness functions to mean that no value is available. For example, this would apply if all contacts were so far apart they didn't pass broad phase screening.
Consider whether +/- Infinity values would also be useful; those would work now except that they probably screw up the event isolation algorithm; should just binary chop in that case.
(Subproject for rigid contact in Simbody 4.0)
Witness functions currently trigger only on a sign change. Unilateral ones must trigger when the sign is wrong, regardless of its previous state. Sign changes should still be tracked when present since they can accelerate event isolation.
Witness functions need to be more flexible in several ways:
NaN
for witness functions to mean that no value is available. For example, this would apply if all contacts were so far apart they didn't pass broad phase screening.+/- Infinity
values would also be useful; those would work now except that they probably screw up the event isolation algorithm; should just binary chop in that case.The text was updated successfully, but these errors were encountered: