Faulty integration
Hallo.
I'm running a specific setup with one fixed planet with different eccentricities in (0,0.01,0.02,0.03,0.04,...,0.09,0.10), one simulation for each eccentricity. All simulations with eccentricities except (0.02,0.03) run normally. With e=0.02, 0.03, the simulation has some issues.
Here's my config.info file for the setup I am running; the only quantity that I change in the different runs is the eccentricity, going from 0 to 0.1 in 0.01 increments.
Disk true
{
Grid
{
Radii 1024
{
Min 0.3
Max 3.3
Spacing ARITHMETIC
}
Sectors 3000
Layers 1
Sector 6.2831853071795862
Half false
Opening 1.4486232791552935
}
AspectRatio 0.05
Density
{
Slope 1
Minimum 1.e-20
Start 0.0001
}
StarAccretion
{
Type constant
Rate 0.1e-8
}
FlaringIndex 0
Viscosity {
Type Alpha
Value 0.005
MRIToDz {
MRIAlpha 0.05 ; implicit value if none specified
OuterTemperature 15 ; implicit value if none specified
MRITemperature 100 ; implicit value if none specified
TransitionRadius 1
}
}
Smoothing
{
Change false
Flat true
Size 2
Taper 0.
}
Referential
{
Type Constant
Transport Fast
IndirectForces true
Omega 0
TrackPlanet false
}
Velocities
{
AddRandomNoise false
}
Boundaries evanescent2D
}
PlanetarySystem
{
Planets {
pl1 {
Name pl1
Init
{
Mass TAPER
{
InitialMass 0
FinalMass 1.e-5
BeginTime 0
EndTime 20
}
SemiMajorAxis 1.49
Eccentricity ... #Value in (0,0.01,0.02,0.03,0.04,...,0.09,0.10)
Inclination 0
}
AccretionTimeInverse 0.0
Feel
{
Disk false
Others false
}
}
}
AccreteOntoPlanet false
ExcludeHill true
}
Output
{
TimeStep 0.2856913611345460123495943
NbStep 4000
Trackers {
step_timer {}
hdf5_logger {
modulo 100
ofile_fmt ./disk%1%.h5
restart_modulo 1
verbose true
}
gas_txt_logger {
ofile ./monitor.dat
}
orbital_elements_txt_logger {
ofile_fmt ./orbit%1%.dat
}
planets_txt_logger {
ofile_fmt ./planet%1%.dat
when step
}
Hill_force_txt_logger {
ofile_fmt ./tqwk%1%.dat
}
Hill_force_hdf5_logger {
ofile tqwk.h5
when step
verbose true
}
gas_tracer {}
}
}
For Eccentricity 0.02
, the OUT file looks like this:
Got disk0.h5 of index 0
Will start at ./disk0.h5
Creating './tqwk.h5'
Updating planet log ./planetpl1.dat
Disk Momentum : 0.000472558
Disk total Mass : 0.000315642
....................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
Updating './tqwk.h5'
Disk Momentum : -nan
Disk total Mass : -nan
Writing disk file './disk1.h5'
Updating planet log ./planetpl1.dat
In step 1: 305.43 seconds
.
Updating './tqwk.h5'
Disk Momentum : -nan
Disk total Mass : -nan
Writing disk file './disk2.h5'
Removed restart file './disk1.h5'.
Updating planet log ./planetpl1.dat
In step 2: 0.23 seconds
.
and so on. So the Disk Momentum and the Disk total Mass became nans, but the code went on integrating as if nothing was happening. This is probably not good, I think there should be a check somewhere that if some quantity is a nan the integration should stop with an error.
For Eccentricity 0.03
, the OUT file looks like this:
----------------------------------------------------------------------------
Mon Mar 22 09:51:11 2021
Hello Gabriele Pichierri,
Welcome to the ISAAC cluster.
============================================================================
Data in /u ($HOME) is quoted to 1TB per user.
- IMPORTANT: /isaac/ptmp is NOT backed up on this system.
You have to archive your data in those directories by yourself.
============================================================================
----------------------------------------------------------------------------
/u/gpichierri/.bashrc: line 44: sourcefargOCA: command not found
Git LFS initialized.
SLURM_NTASKS = 30
SLURM_CPUS_PER_TASK = 4
SLURM_JOB_NODELIST = isaac[044-046]
rm: cannot remove 'disk*': No such file or directory
Start job
Input data 'disk0.h5' has been prepared in directory '.'.
You can launch a simulation by running:
[mpi command] fargOCA .
Created disk0.h5
Got disk0.h5 of index 0
Will start at ./disk0.h5
Creating './tqwk.h5'
Updating planet log ./planetpl1.dat
Disk Momentum : 0.000472558
Disk total Mass : 0.000315642
................................................................................................................................................................................................................................................................
and so on with just dots. So the integration got stuck somewhere and it's not going anywhere.
As an example of a "normal" simulation, e.g. with Eccentricity 0.05
, I get:
Got disk0.h5 of index 0
Will start at ./disk0.h5
Creating './tqwk.h5'
Updating planet log ./planetpl1.dat
Disk Momentum : 0.000472558
Disk total Mass : 0.000315642
............................................................................................................................................................................................................
Updating './tqwk.h5'
Disk Momentum : 0.000472559
Disk total Mass : 0.000315645
Writing disk file './disk1.h5'
Updating planet log ./planetpl1.dat
In step 1: 56.76 seconds
............................................................................................................................................................................................................
Updating './tqwk.h5'
Disk Momentum : 0.00047256
Disk total Mass : 0.000315645
Writing disk file './disk2.h5'
Removed restart file './disk1.h5'.
Updating planet log ./planetpl1.dat
In step 2: 53.27 seconds
............................................................................................................................................................................................................
and so on until a normal end to the integration.
In short, there's the issue of the nan's not being detected by the code, and then I would also like to understand why these two specific simulations give problems. But I don't know how to debug this.
I also tried with like Eccentricity 0.02999
to see if a slight change in the parameter makes a difference but not really.