Compute gradient of weighted NLL w.r.t. all DCC parameters.
This is the main interface for use with optim().
Usage
dcc11_gradient(
params,
std_resid,
weights,
Qbar,
distribution = "mvn",
use_reparam = TRUE
)
Arguments
- params
Named vector of parameters to optimize
- std_resid
T x k matrix of standardized residuals
- weights
T-vector of observation weights
- Qbar
k x k unconditional covariance
- distribution
"mvn" or "mvt"
- use_reparam
Logical: use reparameterized (psi, phi) space?
Value
Vector of gradients (same length and order as params)
