S2 Labs
  • Introduction
  • Getting Started
    • Quickstart
  • Command Line Options
    • isciml
    • isciml: generate-mesh tdem-loop
    • isciml: generate-models
      • inspect-mesh
      • initialize-mu-sigma
      • one-pipe
    • isciml: generate
    • isciml: train
    • isciml: inference
  • isciml tdem
  • Distributed computing
    • Overview
    • Data generation using Slurm
    • Multi-GPU Training
  • Contact Us
Powered by GitBook
On this page
  • Overview
  • Usage
  • Sequential
  • Multiple Cores
  • Distributed
  • Options
  • Description
  • Example Usage
  • Notes
  • Related Commands
  • See Also
  1. Command Line Options

isciml: generate

Overview

The generate subcommand in isciml is used for creating synthetic data or response files based on the models created with generate-models. This command is essential for generating electromagnetic field data, sensor responses, or synthetic measurements that can be used as input for AI models or simulations.

Usage

Sequential

singularity exec isciml.sif isciml generate [OPTIONS]

Multiple Cores

singularity exec isciml.sif mpirun -np <NUMBER_OF_CORES> isciml generate [OPTIONS]

Distributed

Options

Option
Description
Default
Required

--vtk PATH

Mesh in VTK file format

-

Yes

--receiver_file PATH

Receiver locations file in CSV format

-

Yes

--receiver_file_has_header

Flag to indicate if the receiver file has a header

False

No

--perturb_receiver_z

Flag to perturb receiver z location

False

No

--input_folder PATH

Folder with files containing magnetic properties in numpy format

-

Yes

--ambient_field <FLOAT FLOAT FLOAT>...

Ambient magnetic field values (x, y, z)

820.5, 16241.7, 53380.0

No

--output_folder PATH

Folder for output files (adjoints or forward) in numpy format

-

Yes

--output_prefix TEXT

Prefix for output file names

"adjoint" or "forward"

No

--solver [adjoint, forward]

Solver mode: adjoint or forward

adjoint

No

--help

Show the help message and exit

-

No

Description

The generate subcommand creates synthetic data based on the physical models and parameters you provide. It can operate in two modes: adjoint and forward, allowing for different types of data generation depending on your needs.

Example Usage

To generate adjoint data with default settings:

singularity exec isciml.sif isciml generate \
    --vtk /path/to/mesh.vtk \
    --receiver_file /path/to/receivers.csv \
    --input_folder /path/to/magnetic_properties \
    --output_folder /path/to/output

To generate forward data with custom ambient field and output prefix:

singularity exec isciml.sif isciml generate \
    --vtk /path/to/mesh.vtk \
    --receiver_file /path/to/receivers.csv \
    --input_folder /path/to/magnetic_properties \
    --output_folder /path/to/output \
    --ambient_field 800 16000 53000 \
    --output_prefix custom_forward \
    --solver forward

Notes

  1. The VTK file should contain the mesh geometry created using the generate-models command.

  2. The receiver file should be in CSV format, containing the locations of receivers.

  3. If --receiver_file_has_header is not set, the CSV is assumed to have no header.

  4. The --perturb_receiver_z option can be used to add some randomness to the z-coordinate of receivers, which may be useful for simulating real-world imperfections.

  5. The input folder should contain files with magnetic properties (likely generated from generate-models).

  6. The ambient field represents the background magnetic field in the x, y, and z directions.

  7. Output files will be saved in the specified output folder, with names based on the input files and the chosen prefix.

  8. The solver mode determines whether the command generates adjoint or forward data.

Related Commands

  • generate-models: Used to create the physical models before generating data.

  • train: Used to train AI models on the generated data.

  • inference: Used to apply trained models to new data.

See Also

For more information on data generation techniques and parameters, refer to the isciml documentation on synthetic data creation and electromagnetic field modeling.

Previousone-pipeNextisciml: train

Last updated 9 months ago