You're reading the documentation for a development version. For the latest stable documentation, please have a look at v5.1.X.
Gaussian Process Regression¶
Group /input/model/unit_XXX/particle_type_ZZZ/adsorption – ADSORPTION_MODEL = GAUSSIAN_PROCESS_REGRESSION
For information on model equations, refer to Gaussian Process Regression.
IS_KINETICSelects kinetic or quasi-stationary adsorption mode: 1 = kinetic, 0 = quasi-stationary. If a single value is given, the mode is set for all bound states. Otherwise, the adsorption mode is set for each bound state separately.
Type: int |
Range: {0,1} |
Length: 1/NTOTALBND |
CP_VALSFlattened pore-phase concentration training inputs used by the Gaussian process regression model. The values represent the training input points \(X\) used to evaluate the kernel function. The array is interpreted according to
NDIM.
Unit: \(mol~m_{MP}^{-3}\)
Type: double |
Range: unrestricted |
Length: NTRAIN * NDIM |
CS_VALSSolid-phase training targets corresponding to
CP_VALS. These values form the training output vector used to compute the GPR coefficient vector \(\alpha = (K + \sigma_n^2 I)^{-1} y\).
Unit: \(mol~m_{SP}^{-3}\)
Type: double |
Range: unrestricted |
Length: NTRAIN |
TRAINED_PARAMSTrained kernel hyperparameters of the Gaussian process regression model. The parameters are expected in the following order:
MLP weight variance
MLP bias variance
MLP variance
Linear variance
RBF variance
RBF lengthscale
Gaussian noise variance
All entries must be provided, regardless of the selected kernel.
KERNELSelects the kernel function used by the Gaussian process regression model. Supported values are
MLP,RBF,RBF_Linear, andMLP_Linear.
Type: string |
Range: {MLP, RBF, RBF_Linear, MLP_Linear} |
Length: 1 |
NDIMNumber of input dimensions per training point used in
CP_VALS. Must be a positive integer.
Type: int |
Range: \(\geq 1\) |
Length: 1 |
GPR_KKINLinear-driving-force coefficients in component-major ordering.
Unit: \(s^{-1}\)
Type: double |
Range: \(\geq 0\) |
Length: NTOTALBND |