Error: Hessian could not be computed

[rifatzahan]

I tried running both the chicken diet model, as well as, a well-being and smartphone use model, that you shared the code online. However, while working on this,
#fit the model fit_frequency<-ctStanFit(datalong = data_freq, ctstanmodel = model_tot, plot=10,verbose=0,cores=cores,nopriors=F),
the model starts running and then throws the following error:

Maximum a posteriori estimation requested Individual variation in T0VAR parameters not possible, removed Using 6/8 logical CPU cores starting worker pid=68659 on localhost:11022 at 00:30:00.032 starting worker pid=68681 on localhost:11022 at 00:30:00.285 starting worker pid=68711 on localhost:11022 at 00:30:00.533 starting worker pid=68738 on localhost:11022 at 00:30:00.763 starting worker pid=68761 on localhost:11022 at 00:30:01.000 starting worker pid=68779 on localhost:11022 at 00:30:01.237 Doing 1st pass optimize... ................................................................................................. Error in .local(object, ...) : the model object is not created or not valid Error in .local(object, ...) : the model object is not created or not valid Warning in FUN(X[[i]], ...) : NA gradient encountered at param 104 Problems computing Hessian... Error in eigen(if (doDykstra) R else Y, symmetric = TRUE) : infinite or missing values in 'x' Error in (function (standata, sm, init = "random", initsd = 0.01, sampleinit = NA, : Hessian could not be computed Optimization cancelled -- restart from current point by including this argument: init = c(0.003, 0.00522, 0.01099, 0.00609, 0.0045, -0.00096, -0.00962, 0.00591, 0.00441.....................)

Any idea as to what's going on? Or how I can solve this issue? Thanks in advance.

[cdriveraus]

You seem to have some general problem with the installation, most likely related to rstan / stanheaders. I would try removing those as well as ctsem, reinstall (not from source if it asks you) and see how things go.

…

On Mon, 3 Jul 2023, 08:36 Rifat Zahan, ***@***.***> wrote: I tried running both the chicken diet model, as well as, a well-being and smartphone use model, that you shared the code online. However, while working on this, #fit the model fit_frequency<-ctStanFit(datalong = data_freq, ctstanmodel = model_tot, plot=10,verbose=0,cores=cores,nopriors=F), the model starts running and then throws the following error: Maximum a posteriori estimation requested Individual variation in T0VAR parameters not possible, removed Using 6/8 logical CPU cores starting worker pid=68659 on localhost:11022 at 00:30:00.032 starting worker pid=68681 on localhost:11022 at 00:30:00.285 starting worker pid=68711 on localhost:11022 at 00:30:00.533 starting worker pid=68738 on localhost:11022 at 00:30:00.763 starting worker pid=68761 on localhost:11022 at 00:30:01.000 starting worker pid=68779 on localhost:11022 at 00:30:01.237 Doing 1st pass optimize... ................................................................................................. Error in .local(object, ...) : the model object is not created or not valid Error in .local(object, ...) : the model object is not created or not valid Warning in FUN(X[[i]], ...) : NA gradient encountered at param 104 Problems computing Hessian... Error in eigen(if (doDykstra) R else Y, symmetric = TRUE) : infinite or missing values in 'x' Error in (function (standata, sm, init = "random", initsd = 0.01, sampleinit = NA, : Hessian could not be computed Optimization cancelled -- restart from current point by including this argument: init = c(0.003, 0.00522, 0.01099, 0.00609, 0.0045, -0.00096, -0.00962, 0.00591, 0.00441.....................) Any idea as to what's going on? Or how I can solve this issue? Thanks in advance. — Reply to this email directly, view it on GitHub <#49>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA7LCBL6WI2DZJOGJEQWRPLXOJSAJANCNFSM6AAAAAAZ373VKU> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[rifatzahan]

Yes, it worked. Thanks.