In his Applied Differential Geometry book, William Burke says the following after telling that the action should be the integral of a function L:
A line integral makes geometric sense only if it's integrand is a 1-form. Is Ldt a 1-form? Well, that's the wrong question. The correct question is: On what space is Ldt a 1-form? It is not a 1-form on configuration space, the space of positions, because it can have a non-linear dependence on velocities. A 1-form must be a linear operator on the tangent vectors. The correct space for Ldt is the line-element contact bundle of the configuration space.
Now, why intuitively the correct setting for lagrangian mechanics is on the contact bundle? I understand the contact bundle as pairs (p,[v]) where p is a point in configuration space and [v] is an equivalence class of vectors, explicitly v∼kv.
Thinking not on all that arguments for selecting the space on which Ldt is a 1-form, physically, how can we intuit that the contact bundle is useful for that? I mean, is there some observation in classical mechanics that guides us in building the theory on that space?
This post imported from StackExchange Physics at 2014-04-01 12:39 (UCT), posted by SE-user user1620696