Replace the diagonal of a square matrix

Writes a new diagonal onto a square numeric matrix, leaving the off-diagonal values and the dimnames unchanged.

Usage

pd_diagonal(data, replace = "median")

Arguments

data

A square numeric matrix, data frame, or tibble.

replace

One of the strings "min", "max", "mean", "median" (the default), "mode", or "geometric", or a finite numeric value, or a finite numeric vector as long as one side of the matrix.

Value

A square numeric matrix with the new diagonal.

Details

A character replace is a summary computed over every value in the matrix, including the current diagonal. "min", "max", "mean", and "median" propagate NA, so a matrix containing NA yields an NA diagonal. "mode" returns the most frequent value. "geometric" is the exponential of the mean log of the positive, non-missing values and errors when there are none.

A numeric replace is written directly. It may be one finite value or a finite vector as long as one side of the matrix, in which case each diagonal position takes its own value.

Non-square input, non-numeric input, and a replace that is neither character nor numeric are errors.

See also

precise_transform(), whose diagonal argument performs the same operation on a collection and records it in history.

Author

Brian Muchmore

Examples

data(data_cell_cycle, package = "PreciseDist")

cells <- c(1:4, 60:63, 118:121)
cell_cycle <- as.matrix(data_cell_cycle[cells, 2:41])
rownames(cell_cycle) <- paste0(data_cell_cycle$Cell_cycle[cells], "_", cells)

euclidean <- as.matrix(stats::dist(cell_cycle))
diag(euclidean)[1:3]
#> G1_1 G1_2 G1_3 
#>    0    0    0 

# The default writes the median of every value in the matrix.
diag(pd_diagonal(euclidean))[1:3]
#>     G1_1     G1_2     G1_3 
#> 7.251484 7.251484 7.251484 

# A numeric replacement is written directly.
diag(pd_diagonal(euclidean, replace = 1))[1:3]
#> G1_1 G1_2 G1_3 
#>    1    1    1