Cholesky factorization failed - predict function

1

Hi all,
I am trying to use predict function for a model. however I am getting the following error:

image
image1337×403 38.4 KB

This is my code for your reference:
Rhee = @model begin
@param begin
tvcl ∈ RealDomain(lower=0)
tvvc ∈ RealDomain(lower=0)
tvka ∈ RealDomain(lower=0)
Ω ∈ PSDDomain(2)
σ_add ∈ RealDomain(lower=0)
end
@random begin
η ~ MvNormal(Ω)
end
@covariates WT eGFR
@pre begin
Ka=tvka
Cl=tvcl* (WT/70)^0.7 * (eGFR/90)^0.44 exp(η[1])
Vc=tvvc (WT/70) * exp(η[2])
end
@dynamics begin
Depot’= -Ka*Depot
Central’ = -(Cl/Vc)Central + Ka * Depot
end
@derived begin
cp = @.Central/Vc
DV ~ @.Normal(cp, sqrt(cp^2σ_add))
end
end

param_rhee= (tvcl=3.9,
tvvc=65.3,
tvka=2.44,
Ω=[0.144 0.0758;
0.0758 0.0357],
σ_add=0.395)

leve_pop_1= read_pumas(data_lev, observations = [:DV], covariates=[:WT,:eGFR],amt = :AMT,
id = :ID, evid = :EVID, cmt = :CMT, time = :TIME, route=:ROUTE, ii=:II,addl=:ADDL)

rhee_pred = predict(Rhee, leve_pop_1, param_rhee)

2

Your Ω matrix is not positive definite, it has a negative eigenvalue.

julia> eigen([0.144 0.0758;
       0.0758 0.0357]).values
2-element Vector{Float64}:
 -0.0033050454886905246
  0.1830050454886905
3

Ok I get it. Thank you Mohamed.