Big list of challenges and things to demonstrate

[EiffL]

As we work on this some more, we are going to run into specific questions we want to answer. We can use this issue to keep a running list. Ultimately a subset of these questions can go into a first paper.

• [ ] : Accelerate inference of shear posterior (naive HMC takes 50000 iterations)
• [ ] #3
• [ ] : Demonstrate model bias can be eliminated with deep morphology models
• [ ] : (optional) learn morphology model on the fly
• [ ] : Simultaneously infer PSF errors
• [ ] : Move inference to the level of the convergence map
• [ ] : Develop a procedure for handling detection bias
• [ ] : Apply on realistic simulations