[R-sig-dyn-mod] memory issues with lsoda

[Thomas Petzoldt]

Hi,

lsoda etc. can use resp. numerically estimate the Jacobian. This speeds 
considerably up, at the cost of a much higher memory footprint. The 
Runge-Kuttas don't make use of the Jacobian, so they are less memory 
consuming but (in most cases) slower.

Thomas P.

Am 04.07.2017 um 12:45 schrieb Rampal S. Etienne:
> Hi,
>
> When trying desolve with lsoda, I often get an error saying that I use
> lots of memory, e.g.
>
> Error in lsoda(y, times, func, parms, ...) : cannot allocate memory
> block of size 134217728 Tb
> In addition: Warning message: In lsoda(y, times, func, parms, ...) :NAs
> introduced by coercion to integer range
>
>
> With a different solver (e.g. ode45), I don't get this problem. What is
> different in these methods that they use the memory so differently, and
> can the lsoda error somehow be fixed?
>
> Cheers,
>
> Rampal
>