Compute the gradient of log density (which is returned by monty_model_density) with respect to parameters. Not all models support this, and an error will be thrown if it is not possible.
Arguments
- model
A monty_model object
- parameters
A vector or matrix of parameters
- named
Logical, indicating if the output should be named using the parameter names.
See also
monty_model_density for log density, and monty_model_direct_sample to sample from a model
Examples
m <- monty_example("banana")
# Global maximum at (0, 0), and gradient is zero there:
monty_model_density(m, c(0, 0))
#> [1] -1.837877
monty_model_gradient(m, c(0, 0))
#> [1] 0 0
# Nonzero gradient away from the origin:
monty_model_gradient(m, c(0.4, 0.2))
#> [1] -1.440 0.376