
Generate synthetic calibration data with correlated discrepancy
Source:R/func_calibrate.R
gen_calibrate_data.RdSimulates observed calibration data \(z_t\) by adding a spatially correlated random discrepancy \(u_t\) and an independent noise term to the computer model output \(y_t\). The variance \(\tau^2_t\) decays over time via the discount factor \(b\).
Arguments
- ycal_mat
Numeric matrix of dimension
c(t_cal, s_cal). Computer model output at calibration locations and time steps.- para_gen_cal
Named list with elements:
- b
Numeric scalar. Discount factor in \((0, 1)\).
- n0
Numeric scalar. Prior shape of \(\tau^2_0\).
- d0
Numeric scalar. Prior scale of \(\tau^2_0\).
- u0
Numeric vector of length
s_cal. Initial discrepancy.
- U_gen
Numeric matrix. Spatial covariance matrix for the discrepancy (
s_cal x s_cal).
Value
A named list with:
- u
Matrix of dimension
c(t_cal, s_cal)of discrepancy realisations.- z
Matrix of dimension
c(t_cal, s_cal)of simulated observations.- tau2
Numeric vector of length
t_calwith sampled variances.