Re: time-step of integer size 1 not allowed

From: Goddard, Julianne K. <Julianne.Goddard_at_uky.edu>
Date: Thu, 29 Jul 2021 12:06:20 +0000

Dear Volker,
Thank you very much for taking the time to look over my simulation and provide feedback. I am quite new to Gadget and still learning.
I did not explain clearly last time what type of simulation I am running. I am running a zoom-in cosmological simulation with a period boundary. The zoom region in the simulation is 1.5 Mpc in radius, and the effective resolution here is 4096^3. The adaptive softening for DM in the zoom region is 80pc, for the SPH and stars it is 50pc. For the boundary region region it is 1.2 kpc. I did run this same simulation with dark matter only and did not have any errors from z=99 to z=0, so I am inclined to think that the small time-stepping is coming from the gas, but I have yet to confirm this with the recommended debugging steps. I have not yet decreased the value of CritPhysDensity for star formation which us 4 cm^{-3}, I do get stars starting around z=20 and more than 100,000 by z=10.
These parameter values have been run previously with GIZMO.
I had turned on ENABLE_DYNAMIC_RANGE_IN_TIME and now get a new error:
“Code termination on task=91, function drift_particle(), file src/time_integration/predict.cc, line 334: no prediction into past allowed: time0=62628182318120960 time1=62557813573943296”
I have not yet implemented your previous suggestions, as I thought I may have misled you by not explaining clearly. Please let me know if you think those same steps may help with this new error, or if perhaps there is something else I need to look at. Your help and advice is greatly appreciated.
Sincerely,
Julianne
Received on 2021-07-29 14:06:33