added tests to confirm that different sparse matrix subclasses give t…

…he same results

DESCRIPTION

@@ -44,4 +44,4 @@ LazyData: true
 LazyDataCompression: xz
 NeedsCompilation: yes
 VignetteBuilder: knitr
-RoxygenNote: 7.3.0
+RoxygenNote: 7.3.1

R/fit.R

@@ -108,7 +108,7 @@
 #'
 #' @param Y The n x m matrix of counts; all entries of \code{Y} should
 #'   be non-negative. It can be a sparse matrix (class
-#'   \code{"dgCMatrix"}) or dense matrix (class \code{"matrix"}).
+#'   \code{"dsparseMatrix"}) or dense matrix (class \code{"matrix"}).
 #'
 #' @param K Integer 1 or greater specifying the rank of the matrix
 #'   factorization. This should only be provided if the initial fit

tests/testthat/test_fit_glmpca_pois.R

@@ -282,3 +282,44 @@ test_that("Test fit works with input covariates",{
   expect_equivalent(crossprod(fit_quick$U),diag(3),scale = 1,tolerance = 1e-8)
   expect_equivalent(crossprod(fit_quick$V),diag(3),scale = 1,tolerance = 1e-8)
 })
+
+test_that("Results are the same with different classes of sparse matrices",{
+
+  # Simulate a 100 x 200 data set to factorize.
+  set.seed(1)
+  n <- 100
+  m <- 200
+  Y <- generate_glmpca_data_pois(n,m,K = 3)$Y
+
+  Y_dgC <- as(Y, "CsparseMatrix")
+  Y_dgT <- as(Y, "dgTMatrix")
+  Y_dgR <- as(Y, "dgRMatrix")
+
+  set.seed(1)
+  fit0_dgC <- init_glmpca_pois(Y_dgC,K = 3)
+  set.seed(1)
+  fit0_dgT <- init_glmpca_pois(Y_dgT,K = 3)
+  set.seed(1)
+  fit0_dgR <- init_glmpca_pois(Y_dgR,K = 3)
+
+  suppressWarnings(capture.output(
+    fit_quick_dgC <- fit_glmpca_pois(Y_dgC,fit0 = fit0_dgC,
+                                     control = list(maxiter = 20))))
+
+  suppressWarnings(capture.output(
+    fit_quick_dgT <- fit_glmpca_pois(Y_dgT,fit0 = fit0_dgT,
+                                     control = list(maxiter = 20))))
+
+  suppressWarnings(capture.output(
+    fit_quick_dgR <- fit_glmpca_pois(Y_dgR,fit0 = fit0_dgR,
+                                     control = list(maxiter = 20))))
+
+  expect_equal(fit_quick_dgC$progress$loglik, fit_quick_dgR$progress$loglik)
+  expect_equal(fit_quick_dgC$progress$loglik, fit_quick_dgT$progress$loglik)
+  expect_equal(fit_quick_dgC$U, fit_quick_dgR$U)
+  expect_equal(fit_quick_dgC$d, fit_quick_dgR$d)
+  expect_equal(fit_quick_dgC$V, fit_quick_dgR$V)
+  expect_equal(fit_quick_dgC$U, fit_quick_dgT$U)
+  expect_equal(fit_quick_dgC$d, fit_quick_dgT$d)
+  expect_equal(fit_quick_dgC$V, fit_quick_dgT$V)
+})