Simulation result files and input files

[chraibi]

In Gitlab by @gjaeger on Jul 4, 2019, 08:06 [origin]

jpscore:

JuPedSim - JPScore

Current date   : Jul  2 2019 19:07:41
Version        : 0.8.4
Commit hash    : v0.8.4-68-ga33375a
Commit date    : Mon Jul 1 15:21:05 2019
Branch         : develop
----

inifile in /data:

<header>
  <seed>4313</seed>
  <max_sim_time>80</max_sim_time>
  <num_threads>48</num_threads>
  <show_statistics>true</show_statistics>
  <logfile>log_RiMEA_Test_04_7-0_om_2</logfile>
  <!--<progressbar/>-->
  <trajectories format="plain" fps="2">
    <file location="traj_RiMEA_Test_04_7-0_om_2.txt" />
  </trajectories>

  <geometry>geo_RiMEA_Test_04.xml</geometry>
</header>

...

batch_script in $HOME:

#!/bin/bash -x
#SBATCH -J R70
#SBATCH --account=ias-7
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=48
#SBATCH --output=data/out_%j.txt
#SBATCH --error=data/err_%j.txt
#SBATCH --time=6:00:00
#SBATCH --mail-user=g.jaeger@fz-juelich.de
#SBATCH --mail-type=ALL

export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}

srun ./jpscore/bin/jpscore data/ini_RiMEA_Test_04_7-0_om_2.xml

The logfile is stored in the root directory ($HOME) as datalog_RiMEA_Test_04_7-0_om_2.txt. Also the statistic files are stored there.

Both files should be stored in the directory data (like the trajectory files).

Note:
The function <progressbar/> creates a very large file (5-15 GB) and should be disabled.

[anna-braun]

I started a simulation with Jureca today and I have a similar problem. The Trajectories have been written to the directory "Simulationen/test-01", but the statistic files and goal.vtk have been written to the root directory:

@chraibi @schroedtert Is there any possibility to set a file location to "show_statistics"? If not, we should really add this function. For the KapaKrit simulations we will have many statistic files (100+/simulation), which will lead to total chaos, if all the files will be be written into the same/root directory.

batch.sh.txt

[chraibi]

During the process of fixing #448, I would propose to save these files (log and flow*) in the same directory as the trajectory file.

[anna-braun]

@zeroset

1. As discussed today the structure for the output could look like this: With an new output folder for every simulation, which contains all output files like trajectory files and flow statistics and a copy of all input files.
2. Also important is the paths in the trajectory file. At the moment these are absolute paths: But it should better be relative paths to have more flexibility when changing a folders name or sharing the data.

[schroedtert]

Can this be closed? See #649