30 Day 30 (May 7)
30.1 Announcements
Mathematical statistics workshop (link)
Teaching evaluations
Questions/clarifications from journals
Assignments for the tutorial and report should be available on Canvas.
“I am still trying to understand is how to properly validate complex mechanistic space-time models, especially after the fruit fly diffusion example.”
“Something that I was struggling to understand is that although the paper used a negative binomial model for the observed trap counts, the underlying dispersion process was still modeled over a continuous surface. Since the space between the traps never had a recording, I was wondering whether the trap spacing/sampling design affects how well the diffusion process can be estimated or interpreted. I would assume that smaller gaps between traps may not matter as much, but if the gaps increase, would that have to be accounted for more directly in the model? If so, could a diffusion model still be used, but in a more discrete way based on the trap locations?”
“Two questions that I still have are related to terminology. First, what does it mean for a process to be stationary, and what is the opposite of stationarity? Is this idea related to the difference between descriptive and dynamic models, or whether these are separate concepts?”
“Two questions that I still have are related to terminology. First, what does it mean for a process to be stationary, and what is the opposite of stationarity? Is this idea related to the difference between descriptive and dynamic models, or whether these are separate concepts?” See paper here
“I’m still wrapping my head around the boundary conditions and how the choice between absorbing, noflux/reflective, or unknown boundaries actually affects the propagator matrix and the resulting inference. I understand that no-flux means the gradient at the boundary is zero and absorbing means intensity goes to zero there, but I’m not confident yet in how to decide which is biologically or physically appropriate for a given application, and how sensitive the posterior estimates of mu and omega are to that choice.” See paper here
“I’m