Dramatically different results from 2 runs #98
Description
Did the test twice, and got very different results. Rare events?
Here is the model:
using Logging
using StructuralIdentifiability
logger = Logging.SimpleLogger(stdout, Logging.Info)
global_logger(logger)
ode = @ODEmodel(
S'(t) = -beta_SA * S(t)(A(t) + P(t) ) - beta_SI S(t)In(t),
E'(t) = beta_SA S(t) * (A(t) + P(t) ) + beta_SI * S(t)In(t) - sigma_1E(t),
P'(t) = (1-phi)sigma_1E(t) - sigma_2P(t),
A'(t) = phisigma_1E(t) - M_ar A(t),
In'(t) = sigma_2P(t) - MIn(t),
H'(t) = M*In(t) - kappa_s *H(t),
D'(t) = H(t),
y(t) = D(t)
)
@time println(assess_identifiability(ode))
1st run: all are globally identifiable
Dict{Any, Symbol}(sigma_2 => :globally, M_ar => :globally, kappa_s => :globally, sigma_1 => :globally,
beta_SI => :globally, phi => :globally, M => :globally, beta_SA => :globally)
26381.402992 seconds (214.93 M allocations: 13.735 GiB, 0.04% gc time, 0.31% compilation time)
2nd run: took much longer to finish, and only 2 are globally identifiable
Dict{Any, Symbol}(sigma_2 => :locally, M_ar => :globally, kappa_s => :globally, sigma_1 => :locally,
beta_SI => :locally, phi => :locally, M => :locally, beta_SA => :locally)
67551.049371 seconds (214.83 M allocations: 13.644 GiB, 0.02% gc time, 0.12% compilation time)