trimap: Dimensionality Reduction Using Triplet Constraints
In scfurl/m3addon: This package adds to the popular "monocle3"

Find a low-dimensional repersentation of the data by satisfying the sampled triplet constraints from the high-dimensional features.

trimap(
  cds,
  preprocess_method = "PCA",
  python_home = system("which python", intern = TRUE),
  num_dims = NULL,
  module_file = paste(system.file(package = "m3addon"), "trimap.py", sep = "/"),
  n_dims = 2,
  n_inliers = 10,
  n_outliers = 5,
  n_random = 5,
  distance = c("euclidean", "manhattan", "angular", "hamming"),
  lr = 1000,
  n_iters = 400,
  knn_tuple = NULL,
  apply_pca_trimap = FALSE,
  opt_method = c("dbd", "sd", "momentum"),
  verbose = TRUE,
  weight_adj = 500,
  return_seq = FALSE
)

`n_dims`	Number of dimensions of the embedding (default = 2)
`data`	input data samples (rows), features (columns)
`n_inliers:`	Number of inlier points for triplet constraints (default = 10)
`n_outliers:`	Number of outlier points for triplet constraints (default = 5)
`n_random:`	Number of random triplet constraints per point (default = 5)
`distance:`	Distance measure ('euclidean' (default), 'manhattan', 'angular', 'hamming')
`lr:`	Learning rate (default = 1000.0)
`n_iters:`	Number of iterations (default = 400)
`knn_tuple:`	Use the pre-computed nearest-neighbors information in form of a tuple (knn_nbrs, knn_distances) (default = None)
`apply_pca:`	Apply PCA to reduce the dimensions to 100 if necessary before the nearest-neighbor calculation (default = True)
`opt_method:`	Optimization method ('sd': steepest descent, 'momentum': GD with momentum, 'dbd': GD with momentum delta-bar-delta (default))
`verbose:`	Print the progress report (default = True)
`weight_adj:`	Adjusting the weights using a non-linear transformation (default = 500.0)
`return_seq:`	Return the sequence of maps recorded every 10 iterations (default = False)