gradient {umbridge}

Evaluate gradient of target functional depending on model.

Description

Evaluate gradient of target functional depending on model.

Usage

gradient(
  url,
  name,
  out_wrt,
  in_wrt,
  parameters,
  sens,
  config = jsonlite::fromJSON("{}")
)

Arguments

url

URL the model is running at.

name

Name of the desired model.

out_wrt

Output variable to take gradient with respect to.

in_wrt

Input variable to take gradient with respect to.

parameters

Model input parameter (a list of vectors).

sens

Sensitivity of target functional with respect to model output.

config

Model-specific configuration options.

Value

Gradient of target functional.