We present a Python package, PyTelPoint, that provides an open-source way to
model telescope pointing and visualize pointing performance. Models are
implemented that are analogous to those used by TPoint™ so that the results
can be easily adapted to existing control systems. The model fitting is performed
using PyMC (https://docs.pymc.io/). PyMC provides an easy-to-use syntax for building
and describing models along with cutting edge algorithms for fitting the models to data.
While slower than commonly-used linear regression techniques, the Bayesian probabilistic
approach that PyMC enables provides much more robust assessments of the uncertainties
in the model parameters, the correlations between parameters, and the intrinsic scatter in
the observations. Initial development is concentrated on modeling altitude-azimuth
telescopes like the MMTO, but the approach used is easily generalized to other designs
and outside contributions are encouraged.
