| 1 | #TODO: toy dataset, check that indices returned are correct + colors |
| 2 | |
| 3 | context("Check that getParamsDirs behaves as expected") |
| 4 | |
| 5 | test_that("on input of sufficient size, beta is estimated accurately enough", { |
| 6 | n = 100000 |
| 7 | d = 2 |
| 8 | K = 2 |
| 9 | Pr = c(0.5, 0.5) |
| 10 | |
| 11 | betas_ref = array( c(1,0,0,1 , 1,-2,3,1), dim=c(2,2,2) ) |
| 12 | for (i in 1:(dim(betas_ref)[3])) |
| 13 | { |
| 14 | beta_ref = betas_ref[,,i] |
| 15 | #all parameters are supposed to be of norm 1: thus, normalize beta_ref |
| 16 | norm2 = sqrt(colSums(beta_ref^2)) |
| 17 | beta_ref = beta_ref / norm2[col(beta_ref)] |
| 18 | |
| 19 | io = generateSampleIO(n, d, K, Pr, beta_ref) |
| 20 | beta = getParamsDirs(io$X, io$Y, K) |
| 21 | betas = .labelSwitchingAlign( |
| 22 | array( c(beta_ref,beta), dim=c(d,K,2) ), compare_to="first", ls_mode="exact") |
| 23 | |
| 24 | #Some traces: 0 is not well estimated, but others are OK |
| 25 | cat("\n\nReference parameter matrix:\n") |
| 26 | print(beta_ref) |
| 27 | cat("Estimated parameter matrix:\n") |
| 28 | print(betas[,,2]) |
| 29 | cat("Difference norm (Matrix norm ||.||_1, max. abs. sum on a column)\n") |
| 30 | diff_norm = norm(beta_ref - betas[,,2]) |
| 31 | cat(diff_norm,"\n") |
| 32 | |
| 33 | #NOTE: 0.5 is loose threshold, but values around 0.3 are expected... |
| 34 | expect_that( diff_norm, is_less_than(0.5) ) |
| 35 | } |
| 36 | }) |