Unexpectedly slow simulation with zvode

[TorkelE]

I have a simple model I want to simulate. I try the various solvers lsoda zvode and vode. For some reason, the lsod solver is waaay faster than the other two. Like, thing are fine because I can simualte quickly, but I wanted to check whenever I was doing something wrong causing the slow down.

Here is the code:

import sys, os, numpy, libsbml, gillespy2
import timeit
import numpy
from gillespy2 import ODESolver, ODECSolver

model_multistate = gillespy2.core.import_SBML('../Data/multistate.xml')[0]

leng = 1000
%time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='vode',integrator_options={"nsteps":1000000})
%time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='zvode',integrator_options={"nsteps":1000000})
%time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='lsoda',integrator_options={"nsteps":100000})
0

Here is a screenshot of the output:

(lsoda 100 times faster)

I am attaching the model (renamed as .txt file to make GitHub happy)
multistate.txt

[briandrawert]

I think you have a typo on the nsteps for lsoda, it is not the same as the other two. Can you please verify and let me know.

On Thu, Mar 2, 2023 at 2:22 PM Torkel ***@***.***> wrote: I have a simple model I want to simulate. I try the various solvers lsoda zvode and vode. For some reason, the lsod solver is waaay faster than the other two. Like, thing are fine because I can simualte quickly, but I wanted to check whenever I was doing something wrong causing the slow down. Here is the code: import sys, os, numpy, libsbml, gillespy2 import timeit import numpy from gillespy2 import ODESolver, ODECSolver model_multistate = gillespy2.core.import_SBML('../Data/multistate.xml')[0] leng = 1000 %time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='vode',integrator_options={"nsteps":1000000}) %time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='zvode',integrator_options={"nsteps":1000000}) %time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='lsoda',integrator_options={"nsteps":100000}) 0 Here is a screenshot of the output: [image: image] <https://user-images.githubusercontent.com/18099310/222572607-ead15084-44d5-48fd-842d-4ea679c61433.png> (lsoda 100 times faster) I am attaching the model (renamed as .txt file to make GitHub happy) multistate.txt <https://github.com/StochSS/GillesPy2/files/10876289/multistate.txt> — Reply to this email directly, view it on GitHub <#927>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKZDIXV4GX6OEYOSYWEDWDW2EMT5ANCNFSM6AAAAAAVN6D7JM> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Sincerely, Brian Drawert Brian Drawert, Ph.D. (he/him) Visiting Researcher National Environmental Modeling and Analysis Center (NEMAC) University of North Carolina Asheville

[TorkelE]

Away from computer, but can check tomorrow. You mean that for lsoda nstep lack one 0? Generally, having that argument seems to make no difference for lsoda. Even for the otherones it only sets the limit and shouldn't have any effect unless that limit is hit (which it is not)?

[TorkelE]

I can confirm that setting nsteps all equal does not make any change

import sys, os, numpy, libsbml, gillespy2
import timeit
import numpy
from gillespy2 import ODESolver, ODECSolver

model_multistate = gillespy2.core.import_SBML('../Data/multistate.xml')[0]

leng = 1000
%time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='vode', integrator_options={"nsteps":10000})
%time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='zvode',integrator_options={"nsteps":10000})
%time model_multistate.run(solver=ODESolver,t=leng,increment=leng/50.,integrator='lsoda',integrator_options={"nsteps":10000})
0