Build a constraint system

Build an inequation system from constraints provided by the user.

Usage

buildConstraints(
  constraint_types,
  constraint_vars,
  num_elements,
  rho = 1,
  block_list = NULL,
  block_matrix = NULL
)

Arguments

constraint_types

a vector of strings denoting the type of constraint to be added; options: `max_size`, `must_link`, `cannot_link`

constraint_vars

a list of parameters defining the constraints; in case of max-size constraints, the list element must contain an integer denoting the maximum size of the feature set, in case of max-link or cannot link, the list element must be a vector of feature indices to be linked

num_elements

the total number of features (feature-wise constraints) or blocks (block-wise constraints) in the dataset

rho

a positive parameter denoting the level of relaxation; `Inf` denotes a hard constraint, i.e. no relaxation

block_list

the list of feature indices for each block; only required, if block-wise constraints are built and `block_matrix` is `NULL`

block_matrix

the matrix containing affiliations of features to each block; only required, if block-wise constraints are built and `block_list` is `NULL`

Value

a `UBayconstraint` containing a matrix `A` and a vector `b` representing the inequality system `Ax<=b`, and a vector `rho` representing the penalty shape

Details

The function transforms user information about relations between features (must-link or cannot-link constraints) and maximum feature set size (max-size) into a linear inequation system. In addition, the relaxation parameter `rho` can be specified to achieve soft constraints.

Examples

# given a dataset with 10 features, we create a max-size constraint limiting
# the set to 5 features and a cannot-link constraint between features 1 and 2
buildConstraints(constraint_types = c("max_size","cannot_link"),
                 constraint_vars = list(5, c(1,2)),
                 num_elements = 10,
                 rho = 1)
#>  A
#>      [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10]
#> [1,]    1    1    1    1    1    1    1    1    1     1
#> [2,]    1    1    0    0    0    0    0    0    0     0
#>  b
#> [1] 5 1
#>  rho
#> [1] 1 1
#>  block_matrix
#>       [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10]
#>  [1,]    1    0    0    0    0    0    0    0    0     0
#>  [2,]    0    1    0    0    0    0    0    0    0     0
#>  [3,]    0    0    1    0    0    0    0    0    0     0
#>  [4,]    0    0    0    1    0    0    0    0    0     0
#>  [5,]    0    0    0    0    1    0    0    0    0     0
#>  [6,]    0    0    0    0    0    1    0    0    0     0
#>  [7,]    0    0    0    0    0    0    1    0    0     0
#>  [8,]    0    0    0    0    0    0    0    1    0     0
#>  [9,]    0    0    0    0    0    0    0    0    1     0
#> [10,]    0    0    0    0    0    0    0    0    0     1