Tracking for TPC/DC without surfaces

Using a TPC/DC for measuring trajectories of charged particles is related to a setup in which the detecting element is located at a different position than the origin of the measurement. Hence the scenario is different to e.g. modules in a tracking detector. A current method for tracking in TPC/DC relies on the construction of surfaces for each measurement at its origin such that the Kalman filter formalism can be applied. However this involves the construction/destruction of many surfaces as well as coordinate transformations of the particle's parametrisation for each surface. In order to avoid this workload, further development is required to allow track reconstruction without surface dependencies for TPCs/DCs.
The therefore required steps can be grouped as:
1. Measurement description
- [x] Free measurement description
Our current measurements are restricted to a bound parametrisation. In order to handle a free measurement, appropriate adaptions are required
- [ ] Event storage
As the measurements need modifications the same would account to the storage of their collection.
- [ ] Calibrator
A calibrator for the measurements is required that locates the measurements including corrections like space charge distortions.
2. Track extrapolation
- [ ] Propagation needs to handle FreeParameters
The current Propagator and stepper classes can only handle bound parametrisation. This concept needs to be extended such that FreeParameters can be used.
- [ ] Actor to stop propagation at right position
An actor is required to stop the propagation at a given position for applying the Kalman filter formalism.
3. Track finding & fitting
- [ ] Kalman filter for free parameters
The current Kalman filter is able to handle only bound parametrisations and must be extended therefore by FreeParameters.
- [ ] TrackState
The current TrackState stores bound parametrisations and jacobians from bound to bound parametrisations. When FreeParameters should become possible, the dimensions are not fixed anymore.
- [ ] CKF
Similar to the Kalman filter the CKF needs to learn when and how to act.

As these modifications strongly interfere with common tracking algorithms, the existing functionality should be preserved and moreover, as the TPC/DC can be considered as special cases, the modifications should be explicitly demanded by the user. The regular applications outside the TPC/DC should be as less affected from the changes as possible.