[R] i suspect that there a memory leak in "vmmin"?

[Sundar Dorai-Raj]

Vumani Dlamini wrote:
> Dear listers,
> Am currently using MCMC approaches to estimate some parameters of my model. 
> One parameter has to be updated using a tuned gamma distribution. So at each 
> iteration I estimate the mean and variance of the density of the gamma 
> approximation using "vmmin" (i also supply the gradient argument). For 
> moderate replications the procedure works, but if I increase them R crashes. 
> If instead of the tuned gamma density i use the adaptive rejection sampling 
> procedure by Gilks and Wild (downloaded from their website) the procedure 
> returns the results for any number of replications, thus I suspected that 
> there was a memory leak in "vmmin". I am not an expert in "C"  and am only a 
> windows user. Have tried some of the debugging tools available to us windows 
> users: gdb, mpatrol, duma without success. Am not sure whether this is a 
> bug.
> 
> Version 2.3.1 (2006-06-01)
> Windows XP
> 
> Regards, Vumani
> 
> 
> 
> Rgui.exe caused an Access Violation at location 100dec2f in module R.dll 
> Reading from location 3ff7c27d.
> 
> Registers:
> eax=000063ac ebx=01e773a0 ecx=3ff7c27a edx=00000002 esi=01e254b4 
> edi=00000002
> eip=100dec2f esp=00e0ecb0 ebp=00e0ed78 iopl=0         nv up ei pl nz na pe 
> nc
> cs=001b  ss=0023  ds=0023  es=0023  fs=003b  gs=0000             
> efl=00000202
> 
> Call stack:
> 100DEC2F  R.dll:100DEC2F  SET_TAG
> 100E0527  R.dll:100E0527  Rf_allocVector
> 100E0D37  R.dll:100E0D37  R_alloc
> 100F2ACA  R.dll:100F2ACA  vmmin
> 023F5E96  gem.dll:023F5E96  weibullMH
> 10088C33  R.dll:10088C33  do_dotcall
> 100B8827  R.dll:100B8827  Rf_eval
> 100BA4D2  R.dll:100BA4D2  do_set
> 100B86CB  R.dll:100B86CB  Rf_eval
> 100BA5D5  R.dll:100BA5D5  do_begin
> 100B86CB  R.dll:100B86CB  Rf_eval
> 100BB8EB  R.dll:100BB8EB  Rf_applyClosure
> 100B85F8  R.dll:100B85F8  Rf_eval
> 100BA4D2  R.dll:100BA4D2  do_set
> 100B86CB  R.dll:100B86CB  Rf_eval
> 100DB51C  R.dll:100DB51C  Rf_ReplIteration
> 100DBAA6  R.dll:100DBAA6  run_Rmainloop
> 004013CF  Rgui.exe:004013CF
> 00401316  Rgui.exe:00401316
> 00401518  Rgui.exe:00401518
> 00401236  Rgui.exe:00401236
> 00401288  Rgui.exe:00401288
> 7C816D4F  kernel32.dll:7C816D4F  RegisterWaitForInputIdle
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html

It's not clear to my why you suspect "vmmin" (aka BFGS) rather than how 
you are using said function? Can you provide a re-producible example as 
the posting guide asks?

Also, are you using "optim" in R? Or are you calling vmmin from C-code? 
Again, an example would clear up many of the questions you have asked.

--sundar