Computing Errors for Random Fields

The Field data models menu option Random fields > Compute errors for random fields validate existing random field models with imported samples.

Before executing error computation:

  1. Ensure that a random field model has been calculated for the field quantities selected for error computation.

  2. Ensure that validation field samples have been imported for the selected field quantities.

When you select Field data modelsRandom fieldsCompute errors for random fields, error computation is performed for active field samples only. The calculated accuracy represents the random field model's ability to reproduce the samples as a relative value. When visualized, 100% means exact reproducibility. Values greater than 100% indicate overestimation of the true value.

After error computations are executed, the following new quantities are added to the database:

  • One field quantity named accuracy[quantity ident]

  • Three scalar quantities named err_rms[], err_max[], and err_min[]

    • err_rms[] stores the root mean square error averaged over the mesh.

    • err_max[] stores the largest positive local relative difference.

    • err_min[] stores the largest negative local relative difference