Hello, Here is one of my test examples for stiff ode solvers (attached). I tried many lsoda based solvers in different software and they usually can solve this problem with their default parameter values (accuracy etc.). By using ode() from deSolve, it seems that I have to reduce the values of atol and rtol in order to get the expected profiles. There is no method using the default accuracy values I tried which will give me the profile I expect. The problem with these equations become problematic when times riches around 16.33. After that time the value of B should be constant and for S should be zero. I just wanted to send this example to the mailing list and hope there might be any comment about it. In the first place, I hope that the authors of deSolve might consider this example regarding the default values of atol and rtol. Best Regards, Radovan
[R-sig-dyn-mod] stiff ode - accuracy default values
3 messages · Thomas Petzoldt, Radovan Omorjan
Dear Radovan, attachments are not allowed in this mailing list, so please re-send your question again, text only with reproducible code. See: https://www.r-project.org/posting-guide.html Thomas Petzoldt (mailing list moderator)
On 29.04.2016 14:39, Radovan Omorjan wrote:
Hello, Here is one of my test examples for stiff ode solvers (attached). I tried many lsoda based solvers in different software and they usually can solve this problem with their default parameter values (accuracy etc.). By using ode() from deSolve, it seems that I have to reduce the values of atol and rtol in order to get the expected profiles. There is no method using the default accuracy values I tried which will give me the profile I expect. The problem with these equations become problematic when times riches around 16.33. After that time the value of B should be constant and for S should be zero. I just wanted to send this example to the mailing list and hope there might be any comment about it. In the first place, I hope that the authors of deSolve might consider this example regarding the default values of atol and rtol. Best Regards, Radovan
_______________________________________________ R-sig-dynamic-models mailing list R-sig-dynamic-models at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-dynamic-models
Dear Thomas, Thank you very much for your prompt response. I am sorry for this inconvenience, and I will do as you said regarding the message and the code. I will resend it again this moment. Best Regards, Radovan
On 04/29/2016 02:51 PM, Thomas Petzoldt wrote:
Dear Radovan, attachments are not allowed in this mailing list, so please re-send your question again, text only with reproducible code. See:
https://www.r-project.org/posting-guide.html Thomas Petzoldt (mailing list moderator) On 29.04.2016 14:39, Radovan Omorjan wrote: Hello, Here is one of my test examples for stiff ode solvers (attached). I tried many lsoda based solvers in different software and they usually can solve this problem with their default parameter values (accuracy etc.). By using ode() from deSolve, it seems that I have to reduce the values of atol and rtol in order to get the expected profiles. There is no method using the default accuracy values I tried which will give me the profile I expect. The problem with these equations become problematic when times riches around 16.33. After that time the value of B should be constant and for S should be zero. I just wanted to send this example to the mailing list and hope there might be any comment about it. In the first place, I hope that the authors of deSolve might consider this example regarding the default values of atol and rtol. Best Regards, Radovan _______________________________________________ R-sig-dynamic-models mailing list R-sig-dynamic-models at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-dynamic-models _______________________________________________ R-sig-dynamic-models mailing list R-sig-dynamic-models at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-dynamic-models