gtDesign: Optimal designs for group testing experiments

April 9, 2026

• Overview
• Installation
• Statistical model (Huang et al. 2020; Sec. 2 of arXiv:2508.08445)
• Package contents (exported)
• Example: D-optimal design (Table 1, M = 61, q = 0)
• Example: A-optimal design (same theta, q = 0)
• Example: Cost depending on pool size (q > 0)
• Example: c-optimality
• Example: E-optimality via compute_design_SO
• Example: Equivalence theorem check (D-opt)
• Maximin multi-objective designs
• Tables and Figures in the paper
• References
• License
• TODO

Authors

Chi-Kuang Yeh (Georgia State University)

Weng Kee Wong (University of California, Los Angeles)

Julie Zhou (University of Victoria)

Overview

gtDesign is an R package for locally optimal approximate designs on a finite candidate set for group testing (pooled testing). Designs are found by convex optimization via CVXR (typically with the CLARABEL solver).

The main application is the Huang et al. (2020) model for prevalence, sensitivity, and specificity with optional cost that depends on pool size, as used in Yeh, Wong, and Zhou (2025); see arXiv:2508.08445. The same interface also supports generic nonlinear design whenever the (approximate) information matrix is a sum of rank-one terms