dfish

I'm not sure what happened. The run_all_benchmarks functions stopped working. I think the problem is that the module is not available on all workers. Do you have any suggestions for...

It works with DynamicHMC #master and Turing #master.

Good catch. It looks like something in my conversion to a Chain object broke. I'll look into that shortly. The most comprehensive source on the model can be found [here](http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.408.1948&rep=rep1&type=pdf)...

Yup. You are right. Sorry. I misunderstood. This seems to be working correctly:` p((v=fill(.5,Nc),A=.8,k=.2,tau=.4))` If you replace the the problem function with this ``` function (problem::LBAProb)(θ) @unpack data=problem @unpack v,A,k,tau=θ...

Thanks, Rob. Do you think the problem is with LBA_functions.jl or with DynamicHMC? My guess is that the problem is with DynamicHMC because it works with Turing.

Good point. Another difference that might be worth considering is that DynamicHMC uses different parameter values for NUTS (and this may also affect the initial values). Turing is now identical...

What parameter has the value of -.24? That might be a problem because only the v parameters can be negative.

Yeah. That is problematic. I suspect that part of DynamicHMC that enforces the allowable ranges of parameters is the source of the problem. For reference, here is the part of...

I'm not sure that the problem is due to numerical instability of the likelihood function. However, I want to make reference to a more numerically stable version of the likelihood...

I found some translation errors in new code for the likelihood function ([file here](https://github.com/StatisticalRethinkingJulia/MCMCBenchmarks.jl/blob/master/temp/test.jl)). As far as I can tell, it is not possible to convert all of the terms...