context("Gaussian Random Field Classifier") library(kernlab) data <- generateTwoCircles(400,0.1) data[-sample(1:nrow(data),10),]$Class <- NA y <- na.omit(data$Class) X <- as.matrix(data[!is.na(data$Class),1:2]) X_u <- as.matrix(data[is.na(data$Class),1:2]) Xin <- rbind(X,X_u) Y <- model.matrix(~y-1,data.frame(y))[1:2,1:2,drop=FALSE] W <- exp(-as.matrix(dist(Xin))^2/0.1) test_that("harmonic_function does not give error",{ RSSL:::harmonic_function(W,Y) expect_equal(RSSL:::harmonic_function(W,Y)$fu, RSSL:::harmonic_function_cpp(W,Y), tolerance=0.01,check.attributes=FALSE) # Small benchmark: #library(microbenchmark) #microbenchmark(RSSL:::harmonic_function(W,Y),RSSL:::harmonic_function_cpp(W,Y)) })