[R-sig-dyn-mod] Possible bug specifying root functions for lsodes

[McGrath, Justin M]

Dear Thomas,
Thank you for the response. I'm modeling a system similar to one that has been modeled before, and the original authors recommended a stiff solver (they use ode15s in Matlab). I tried several of the solvers in deSolve (including nonstiff ones) and the results from some solvers were clearly bad (lsode didn't work well), and although some other solvers produced decent results (I forget which solvers), the results from lsodes seemed best. lsodes was also faster than most other solvers. Calculating roots isn't necessary - I was trying to do that so that the solver could stop when it reached steady state, which would make some things easier for me.

Best wishes,
Justin

________________________________________
From: r-sig-dynamic-models-bounces at r-project.org [r-sig-dynamic-models-bounces at r-project.org] on behalf of Thomas Petzoldt [Thomas.Petzoldt at TU-Dresden.de]
Sent: Tuesday, March 05, 2013 3:44 PM
To: Special Interest Group for Dynamic Simulation Models in R
Subject: Re: [R-sig-dyn-mod] Possible bug specifying root functions for lsodes

Dear Justin,

thanks for the bug report. The developer's web site can be found here:

http://desolve.r-forge.r-project.org/

containing a.o. addresses, links to developer homepages to the source
code and to this mailing list. In addition, full email addresses can
also be found in the package.

I was able to reproduce your problem and we will make efforts to fix it,
however we cannot do it immediately. For the meanwhile, I would
recommend to use another solver.

Is there any reason why you need lsodes (the sparse Jacobian ODE solver)
with root finding?

Thomas Petzoldt


--
Dr. Thomas Petzoldt
Technische Universitaet Dresden
Faculty of Environmental Sciences
Institute of Hydrobiology
01062 Dresden, Germany

E-Mail: thomas.petzoldt at tu-dresden.de
http://tu-dresden.de/Members/thomas.petzoldt

_______________________________________________
R-sig-dynamic-models mailing list
R-sig-dynamic-models at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-dynamic-models