tests/testthat/test_compute_elastic_mean.R

In elasdics: Elastic Analysis of Sparse, Dense and Irregular Curves

test_that("initial param does not matter too much",{
  data_curve1 <- data.frame(x1 = sin(1:7/4*pi), x2 = cos(1:7/4*pi))
  data_curve2 <- data_curve <- data.frame(x1 = sin(1:15/8*pi), x2 = cos(1:15/8*pi))
  data_curves <- list(data_curve1, data_curve2)
  mean1 <- compute_elastic_mean(data_curves)
  mean2 <- compute_elastic_mean(data_curves, type = "poly", eps = 10^-5,
                                max_iter = 10)

  data_curves[[1]]$t <- c(0, 0.2, 0.3, 0.4, 0.6, 0.8, 1)
  mean3 <- compute_elastic_mean(data_curves, eps = 10^-5, max_iter = 10)
  mean4 <- compute_elastic_mean(data_curves, type = "poly")

  expect_equal(mean((mean1$coefs - mean3$coefs)^2), 0,  tolerance=1e-1)
  expect_equal(mean((mean2$coefs - mean4$coefs)^2), 0,  tolerance=1e-1)
})

test_that("0 and 1 iterations of open and closed means",{
  data_curve1 <- data.frame(x1 = sin(1:7/4*pi), x2 = cos(1:7/4*pi))
  data_curve2 <- data_curve <- data.frame(x1 = sin(1:15/8*pi), x2 = cos(1:15/8*pi))
  data_curves <- list(data_curve1, data_curve2)
  expect_warning(compute_elastic_mean(data_curves, max_iter = 1))
  expect_warning(compute_elastic_mean(data_curves, max_iter = 1, eps = 0.0001, closed = TRUE))
  expect_warning(compute_elastic_mean(data_curves, closed = TRUE, max_iter = 0),
                 regexp = NA)
})


test_that("polygon closed mean is closed",{
  data_curve1 <- data.frame(x1 = sin(0:7/4*pi), x2 = cos(0:7/4*pi))
  data_curve2 <- data_curve <- data.frame(x1 = sin(0:15/8*pi), x2 = cos(0:15/8*pi))
  data_curves <- list(data_curve1, data_curve2)
  elastic_mean <- compute_elastic_mean(data_curves, closed = TRUE,
                                       type = "polygon", knots = seq(0,1,0.1))
  points <- get_evals(elastic_mean)
  expect_equal(as.numeric(points[1,]), as.numeric(points[nrow(points),]), tolerance = 1e-2)
})

test_that("mean for 3d curves is 3d",{
  data_curve1 <- data.frame(x1 = sin(0:7/4*pi), x2 = cos(0:7/4*pi), x3 = cos(0:7/4*pi))
  data_curve2 <- data_curve <- data.frame(x1 = sin(0:15/8*pi), x2 = cos(0:15/8*pi), x3 = sin(0:15/8*pi))
  data_curves <- list(data_curve1, data_curve2)
  elastic_mean <- compute_elastic_mean(data_curves, type = "smooth", knots = seq(0,1,0.25))
  expect_equal(ncol(elastic_mean$coefs), 3)
})

test_that("mean for 1d curves",{
  data_curve1 <- data.frame(x1 = 1:20*sin(1:20)/20)
  data_curve2 <- data.frame(x1 = cos(1:15))
  data_curves <- list(data_curve1, data_curve2)
  expect_warning(elastic_mean <- compute_elastic_mean(data_curves, type = "smooth",
                                                      knots = seq(0,1,0.25)))
  expect_equal(ncol(elastic_mean$coefs), 1)
})

test_that("same dim for all curves", {
  data_curve1 <- data.frame(x1 = 1:20*sin(1:20), x2 = 1:10*cos(1:20), x3 = tan(1:20))
  data_curve2 <- data.frame(x1 = sin(1:20), x2 = cos(1:20/2))
  data_curves <- list(data_curve1, data_curve2)
  expect_error(compute_elastic_mean(data_curves), "All curves must have same number of dimensions!")
})

test_that("duplicated points are removed", {
  data_curve <- data.frame(x1 = 1:20*sin(1:20), x2 = 1:10*cos(1:20), x3 = tan(1:20))
  data_curve <- rbind(data_curve[1,], data_curve)
  expect_warning(compute_elastic_mean(list(data_curve)), "Duplicated points in data curves have been removed!")
})