Sylvain Barde

Comments on the paper. "Reconstructing Aggregate Dynamics in Heterogeneous Agents Models" by D. Delli Gatti et al

The main aim of the paper is to apply the image processing interpretation of the Maximum Entropy (MaxEnt) method to the Kirman (1993) model and the Abrams and Strogatz (20003) voter model as implemented by Stauffer et al. (2007). This follows the initial work in Barde 2012 which showed that the Schelling (1969) model of segregation can be predicted with the methodology. The discussant does point out some of the major issues that are associated with the methodology, many of which I agree with. The most important comment is probably the fact that more exploratory work is needed to establish a taxonomy of valid assumptions for corresponding statistical properties. Having said this, I feel that two important clarifications are needed.