function EstimateDiagonal(A::Matrix{Float64},Algorithm::Symbol, StoppingCriterion::Symbol, distribution::Symbol, normalizationParam::Bool=true, parallelizationParam::Bool=false ;maxqueries::Int=0,O=nothing,var_est::Float64=1/eps(),epsilon::Float64=1.0, delta::Float64=1.0,con::Float64=1.0)

Main function to compute a randomized estimate of the diagonal of a matrix A

Input:

• A: Matrix to estimate the diagonal of
• Algorithm: Choose which diagonal estimator is used, options are
  • :GirardHutchinson
  • DiagPP
  • :NysDiagPP
  • :XDiag
  • :GirardHutchinsonShift
  • :RSVD
  • :AdaptiveDiagPP
• StoppingCriterion: How to terminate, possible options
  • doubling: Use doubling strategy and terminate when the relative error estimate is below a threshold eps, required parameter
    • queries_start: number of queries to start with
    • eps: bound for the relative error estimate
    • Note: has only been implemented with :GirardHutchinson and :DiagPP, other methods will be added with future releases
  • queries: terminate when the maximum number of queries to A is reacher
    • maxqueries: maximum number of queries to A
  • adaptive: Using an epsilon delta estimator with :AdaptiveDiagPP
    • epsdelta: parameters (epsilon,delta) for an epsilon-delta estimator
    • maxiter: maximum iterations
• distribution: Select the distribution from which the random vectors are drawn, inbuilt options are
  • :Rademacher
  • :Gaussion
  • :custom, in this case provide a matrix with the test vectors as columns
    • O: matrix with the test vectors as columns