The main reason why convex optimization isn’t a commonly used model in this setting is because interventions are discrete objects. You can’t run 1⁄4 of a malaria bednet project and 2⁄3 of a deworming project since there are scale effects.
When you treat projects as discrete and apply a budget constraint, the knapsack problem becomes a more natural model.
I might be misunderstanding what you mean by applying convex optimization. Say more?
The main reason why convex optimization isn’t a commonly used model in this setting is because interventions are discrete objects. You can’t run 1⁄4 of a malaria bednet project and 2⁄3 of a deworming project since there are scale effects.
When you treat projects as discrete and apply a budget constraint, the knapsack problem becomes a more natural model.
I might be misunderstanding what you mean by applying convex optimization. Say more?