can we fastpath=FALSE our way out of irlba problems?

R/init.R

@@ -178,14 +178,19 @@ irlba_tsvd_normalized_laplacian_init <- function(A, ndim = 2, verbose = FALSE) {
 }
 
 irlba_spectral_tsvd <- function(L, n, iters = 1000) {
-  suppressWarnings(res <- irlba::irlba(L, nv = n, nu = 0, maxit = iters))
+  suppressWarnings(res <- tryCatch(
+    irlba::irlba(L, nv = n, nu = 0, maxit = iters),
+    error = function(c) {
+      irlba::irlba(L, nv = n, nu = 0, maxit = iters, fastpath = FALSE)
+    }
+  ))
   list(vectors = res$v, values = 2.0 - res$d, converged = res$iter != iters)
 }
 
 irlba_eigs_asym <- function(L, ndim) {
   suppressWarnings(res <- tryCatch(
     {
-      res <- irlba::partial_eigen(
+      irlba::partial_eigen(
         L,
         n = ndim + 1,
         symmetric = FALSE,
@@ -194,19 +199,34 @@ irlba_eigs_asym <- function(L, ndim) {
         maxit = 1000,
         verbose = TRUE
       )
-      res$values <- sqrt(res$values)
-      res
     },
     error = function(c) {
-      NULL
+      tryCatch(
+        irlba::partial_eigen(
+          L,
+          n = ndim + 1,
+          symmetric = FALSE,
+          smallest = TRUE,
+          tol = 1e-3,
+          maxit = 1000,
+          verbose = TRUE,
+          fastpath = FALSE
+        ),
+        error = function(c) {
+          NULL
+        }
+      )
     }
   ))
+  if (!is.null(res)) {
+    res$values <- sqrt(res$values)
+  }
   res
 }
 
 irlba_eigs_sym <- function(L, ndim, smallest = TRUE) {
   suppressWarnings(res <- tryCatch(
-    res <- irlba::partial_eigen(
+    irlba::partial_eigen(
       L,
       n = ndim + 1,
       symmetric = TRUE,
@@ -216,7 +236,21 @@ irlba_eigs_sym <- function(L, ndim, smallest = TRUE) {
       verbose = FALSE
     ),
     error = function(c) {
-      NULL
+      tryCatch(
+        irlba::partial_eigen(
+          L,
+          n = ndim + 1,
+          symmetric = TRUE,
+          smallest = smallest,
+          tol = 1e-3,
+          maxit = 1000,
+          verbose = FALSE,
+          fastpath = FALSE
+        ),
+        error = function(c) {
+          NULL
+        }
+      )
     }
   ))
   res
@@ -439,8 +473,7 @@ irlba_scores <- function(X, ncol, center = TRUE, ret_extra = FALSE, verbose = FA
     X <- X * 1
   }
   res <- irlba::prcomp_irlba(X,
-    n = ncol, retx = TRUE, center = center,
-    scale = FALSE
+    n = ncol, retx = TRUE, center = center, scale = FALSE
   )
   report_varex(res, verbose)
   if (ret_extra) {

tests/testthat/test_normlaplacian.R

@@ -7,38 +7,46 @@ context("normalized laplacian")
 # with sanitizers and valgrind (probably the extended compilation time with
 # eigen causes a preperror)
 
-test_that("normalized laplacian", {
-  # These numbers come from running UMAP Python code:
-  # spectral_layout(pairwise_distances(iris.data[0:10, :]))
-  # NB:
-  # 1. iris data in scikit-learn is currently from UCI repo, which has errors
-  #   (although this doesn't affect the first ten entries)
-  # 2. eigenvector calculation is not that converged and specifies a starting
-  #   vector that we can't supply with either RSpectra or eigen.
-  # 3. The eigenvectors are only identical up to a sign, so we take the absolute
-  #   values.
-  abs_expected_norm_lap <-
-    abs(
-      c2y(
-        0.7477, -0.1292, -0.03001, 0.02127, -0.563, -0.01149, 0.1402,
-        -0.2725, -0.01241, 0.1084, -0.106, -0.5723, 0.2024, -0.3082,
-        0.1642, -5.549e-05, -0.04843, -0.1747, 0.1684, 0.6611
-      )
+# These numbers come from running UMAP Python code:
+# spectral_layout(pairwise_distances(iris.data[0:10, :]))
+# NB:
+# 1. iris data in scikit-learn is currently from UCI repo, which has errors
+#   (although this doesn't affect the first ten entries)
+# 2. eigenvector calculation is not that converged and specifies a starting
+#   vector that we can't supply with either RSpectra or eigen.
+# 3. The eigenvectors are only identical up to a sign, so we take the absolute
+#   values.
+abs_expected_norm_lap <-
+  abs(
+    c2y(
+      0.7477, -0.1292, -0.03001, 0.02127, -0.563, -0.01149, 0.1402,
+      -0.2725, -0.01241, 0.1084, -0.106, -0.5723, 0.2024, -0.3082,
+      0.1642, -5.549e-05, -0.04843, -0.1747, 0.1684, 0.6611
     )
+  )
+sparse_m <- Matrix::drop0(x2d(iris[1:10, ]))
 
-  sparse_m <- Matrix::drop0(x2d(iris[1:10, ]))
+test_that("normalized laplacian", {
   res <- normalized_laplacian_init(sparse_m)
   expect_equal(abs(res), abs_expected_norm_lap, tolerance = 0.2)
+})
 
+test_that("irlba tsvd normalized", {
   # 115: ensure irlba code path gets tested if we can avoid Matrix ABI issue
-  if (exists("Matrix.Version", envir = asNamespace("Matrix")) &&
-    Matrix::Matrix.Version()$package >= "1.7.0") {
+  # if (exists("Matrix.Version", envir = asNamespace("Matrix")) &&
+  #     Matrix::Matrix.Version()$package >= "1.7.0") {
     res <- irlba_tsvd_normalized_laplacian_init(sparse_m)
     expect_equal(abs(res), abs_expected_norm_lap, tolerance = 0.2)
+  # }
+})
 
+test_that("irlba normalized", {
+  # 115: ensure irlba code path gets tested if we can avoid Matrix ABI issue
+  # if (exists("Matrix.Version", envir = asNamespace("Matrix")) &&
+  #     Matrix::Matrix.Version()$package >= "1.7.0") {
     res <- irlba_normalized_laplacian_init(sparse_m)
     expect_equal(abs(res), abs_expected_norm_lap, tolerance = 0.2)
-  }
+  # }
 })
 
 test_that("laplacian eigenmap", {

tests/testthat/test_output.R

@@ -310,14 +310,14 @@ expect_ok_matrix(res, nc = 1)
 
 # enforce irlba for spectral initialization even if RSpectra is present
 # 115: ensure irlba code path gets tested if we can avoid Matrix ABI issue
-if (exists("Matrix.Version", envir = asNamespace("Matrix")) &&
-  Matrix::Matrix.Version()$package >= "1.7.0") {
+# if (exists("Matrix.Version", envir = asNamespace("Matrix")) &&
+#   Matrix::Matrix.Version()$package >= "1.7.0") {
   res <- umap(iris10,
     n_components = 1, n_neighbors = 4, n_epochs = 2,
     n_threads = 1, verbose = FALSE, init = "irlba_spectral"
   )
   expect_ok_matrix(res, nc = 1)
-}
+# }
 
 # Supervised
 set.seed(1337)