##  [Optimal Experimental Design](/optimal-experimental-design-0) 

 Definition

The process of selecting experimental conditions, inputs, or data-collection strategies to maximize information gain or estimation efficiency with respect to specified modeling objectives, subject to resource and practical constraints.

 

 

 

 

 

 





## Principle

Principle

Formalize an objective (maximize Fisher information, minimize parameter variance, maximize expected utility or posterior precision) and search or optimize over design variables (sampling locations, experimental settings, allocation of replications), often using Bayesian or decision-theoretic criteria that incorporate cost and prior uncertainty.

 

 

 

 

 





## Demonstration

Demonstration

Choose sensor placements along a river network to minimize the posterior variance of pollutant transport parameters (D-optimality), or select time points for pharmacokinetic sampling to maximize information about absorption and clearance under a budget of blood draws.

 

 

 

 

## Misapplication

Misapplication

Optimize a design using an incorrect or oversimplified model without checking robustness, or ignore logistical constraints (measurement noise, setup time), resulting in designs that are theoretically optimal but impractical or misleading in practice.

 

 

 

 

 





## Consequence

Consequence

Properly applied, optimal designs reduce the number of experiments needed, increase estimator precision, and focus resources where they most improve inference or prediction.

 

 

 

 

## Reversal

Reversal

Randomized or convenience sampling without optimization trades optimal information efficiency for simplicity or fairness; adaptive designs flip the static planning by updating choices with incoming data.

 

 

 

 

 





## Boundary

Boundary

Pertains to model- or objective-driven experimental planning; does not replace the need for validation, ethical review, or adaptive re-planning when model assumptions fail or when sequential learning occurs.

 

 

 

 

 





## Semantic Tension

Semantic Tension

Tension exists between classical optimality criteria (D-, A-, E-optimality) that target parameter estimation precision and decision-theoretic or utility-based designs that directly optimize expected decision outcomes; choice depends on whether inference or decisions are primary.

 

 

 

 

 





## Synthesis

Synthesis

Optimal experimental design formalizes information goals and constraints into an optimization over design variables, producing experiments that maximize learning or decision-relevant precision while requiring careful modeling of objectives, priors, and practical limits.