test_likelihood_improvement.R
In deGradInfer: Parameter Inference for Systems of Differential Equation

# Tests that for a reasonable number of iterations, the log likelihood improves
library(deGradInfer)

set.seed(10)

context("Likelihood Improvement, 500 Iterations")

# Run deGradInfer on test dataset

dataTest <- LV_example_dataset$data
timeTest <- LV_example_dataset$time
noiseTest <- LV_example_dataset$noise

LV_func = function(t, X, params) {
  dxdt = cbind(
    X[,1]*(params[1] - params[2]*X[,2]),
    - X[,2]*(params[3] - params[4]*X[,1])
  )
  return(dxdt)
}

results = agm(data=dataTest,time=timeTest,noise.sd=0.31,ode.system=LV_func,
              numberOfParameters=4,temperMismatchParameter=TRUE,
              showPlot=FALSE,
              chainNum=5, maxIterations=500,originalSignalOnlyPositive=TRUE,
              logPrior="Gamma",defaultTemperingScheme="LB10")

latest = length(results$ll)

test_that("Likelihood is improved", {
  expect_gt(results$ll[latest], results$ll[2])
})