Open EnzoHG opened 6 years ago
@EnzoHG sorry for getting back to you so late.
Problem is with sig0
. Currently you are setting it so that all the values are uniform(0,1) which produces a nonsymmetric covariance matrix. This does not match your model specification where sig0 = A'A
@bdeonovic no problem, at this point i'm doing for my self. i come to this solution ``
A=Logical(2,
sig0 -> transpose(sig0)*sig0
),
sig0=Stochastic(2,
(I,J) ->UnivariateDistribution[
Uniform(0,1)
for i in 1:I, j in 1:J
],
false
) ´´
but this doesn't seem to work, it's no longer the symmetric issue but this:
LoadError: DimensionMismatch("incompatible distribution for stochastic node") while loading C:\Users\Enzo Hernandez\Downloads\tarea2_mamba.jl, in expression starting on line 89 in at base\
in #mcmc#29 at Mamba\src\model\mcmc.jl:29 in setinits! at Mamba\src\model\initialization.jl:24 in setinits! at Mamba\src\model\initialization.jl:11 in setinits! at Mamba\src\model\dependent.jl:171
i followed the example in the birats model with the normal distribution matrix.
i really appreciate that you take the time to help me out 👍 thanks again
Your model is still wrong. You cannot have sig0 as a covariance matrix because a matrix filled with random values is not guaranteed to be positive definite. Also your new code still doesn't follow your model that you have a picture of in original post.
your model is something like this:
A=Stochastic(1,
()-> Uniform(0,1)
),
sig0=Logical(2,
(A) -> A'A,
false
)
This is my very first time in julia, and i can't seem to follow the traceback and crack the problem. i have visited number of forums and i haven't found a similar issue. i checked out the birats example as a guide for this
![distribution2](https://user-images.githubusercontent.com/32920164/31754165-31e84fc8-b46c-11e7-909c-d2587ba421ee.PNG)
and i think i have all the syntax right( mu = R should be mu_cero). this is my code: `
where the error and the traceback are:
I really want to learn to use mamba, so any piece of advice would be very much appreciate it.