gridSearchPrior: Find the optimal design for a given network.

In bmp22/networkDesign: Design of Experiments for Networks

Find the optimal design for a given network.

Description

Finds the optimal design for a given network. Various algorithms are implemented to search a network to find the optimal design for. estimating treatment effects on that network. It can also be used to find optimal designs for experiments that contain blocking. Can be slow for high A>20 or high p>2.

Usage

gridSearchPrior(
  A,
  p,
  isoSearch = FALSE,
  blockList = NULL,
  ignoreLastNode = FALSE,
  algorithm = "sequential",
  networkEffects = FALSE,
  weightPrior = NULL,
  viralOpt = TRUE,
  weights
)

Arguments

A	An adjacency matrix
p	The number of treatments in the experiment
isoSearch	Whether to ignore designs that are isomorphic to designs evaluated earlier
blockList	A list. each element of which is a collection of nodes which form a block.
ignoreLastNode	Whether we set the last node as zero. Set to true to satisfy constraints in double blocking structures
algorithm	Which algorithm to use for finding which designs to evaluate; currently 'sequential', 'random' and 'ce' and exchange are implemented
networkEffects	Set to true if the optimal design for the network effects be found (default is to find direct effects)
viralOpt	If we have a viral parameter, do we want to estimate it (TRUE) or just estimate the difference in treatment effects

bmp22/networkDesign documentation built on June 25, 2022, 11:45 a.m.