[julia-users] question about Hessian in ForwardDiff return InexactError()

[jamesmnason]

Hi All:

I am trying to compute a Hessian of a log likelihood function in a 
Metropolis-Hastings MCMC algorithm.  The coefficient vector of the model 
has seven elements. 

A call to

           ForwardDiff.hessian(PF_RE_AR1_outer_alt, parms)

returns

ERROR: LoadError: InexactError()
 in setindex! at array.jl:314
 in PF_RE_AR1_outer_alt at 
/home/jim_nason/jmn_work/smith/NS4/jl_code_Summer2015/RE_rho/MH_PF_test/PF_RE_AR1_outer_alt.jl:55
 in _calc_hessian at 
/home/jim_nason/.julia/v0.4/ForwardDiff/src/api/hessian.jl:98

Lines 52 and 55 of PF_RE_AR1_outer_alt.jl are

     cs00V = eye(dsts)
     cs22n = sigu*sigu
     cs22d = 1/(1.0 - rho1*rho1)
cs00V[2,2] = cs22n*cs22d

where sigu and rho1 are two coefficients of the model restricted as sigu > 
0 and -1 < rho1 < 1 and dsts = 2.

Suggestions or advice to fix the problem are welcome.

Thanks.

Best,

Jim