Currently walkers are started in positions that are uniformly drawn from the Prior space.
~Perhaps users can pass an argument specifying an initial guess for each parameter and a sigma value and we can draw walker positions from this distribution.~
~For more advanced users, we should have a convenience function that lets users create their own list of start positions and then update sampler.curr_pos with this list.~
Decided to write a tutorial on how to modify this instead, for now.
sblunt
commented
5 years ago
Currently walkers are started in positions that are uniformly drawn from the Prior space.
~Perhaps users can pass an argument specifying an initial guess for each parameter and a sigma value and we can draw walker positions from this distribution.~
~For more advanced users, we should have a convenience function that lets users create their own list of start positions and then update
sampler.curr_poswith this list.~Decided to write a tutorial on how to modify this instead, for now.