Discussion

We know that heterogeneity causes scattering, and we have discovered that this scattering is not apparent in the transmissivity at least up to and including the second-order term in $i\omega$. A straight-forward consequence is that no visco-elastic model can simulate scattering loss - it has to be a less lossy system. Nor can the equivalent dynamic model be a Kjartannsson model since this would have the wrong static (non-elastic) properties. Is there necessarily any dynamically equivalent model? No. In the end we could always replace our complex heterogeneities with simpler ones. As an example, we could substitute bimodal models for multi-modal data.