What is the Output Directory in QUEENS?#
The output directory, commonly referred to as output_dir
, is a designated space where all results from iterators are stored. You must specify this directory within the GlobalSettings
object.
Note: This directory does not include the outputs from model simulations, such as simulation data. While this distinction might seem confusing at first, remember that analysis outputs are consistently stored locally, whereas model outputs can reside on remote machines. For more information, consult Where are my simulations? and How is the output of a simulation model structured?.
Where Are My Simulations?#
If you utilize a simulation model with a Cluster
or Local
scheduler, you need to set the experiment_base_dir
. If an experiment directory isn’t specified, it defaults to ~/queens-experiments/
. This location is where all simulation outputs are stored:
└── <experiment_base_dir>/
└── <experiment_name>/
├── 0
├── 1
├── 2
├── 3
├── ...
└── <n_sims-1>
In this structure, 0, ..., <n_sims-1>
represent the outputs of individual simulations. The numbers correspond to job IDs provided by the scheduler, enumerated in the order the model was called.
How is the Output of a Simulation Model Structured?#
When employing the JobScript
driver or its derived variants such as Mpi
or FourC
, the output structure is as follows:
└── <job_dir>/
├── input_file.<extension>
├── metadata.yaml
├── jobscript.sh
└── output/
Here’s a breakdown of each component:
input_file.<extension>
: The input file for your simulation code. Note that multiple input files can be utilized.jobscript.sh
: Contains the commands required to execute your code.metadata.yaml
: Provides metadata from the simulation run, detailing timings, inputs, outputs, and additional information.output
: Includes the actual output files generated by the solver, including the log file created by QUEENS.
The specific job_dir
is assigned by a scheduler. For more details, refer to Where are my simulations?.