Beware of density dependent pair potentials

Density (or state) dependent pair potentials arise naturally from coarse-graining procedures in many areas of condensed matter science. However, correctly using them to calculate physical properties of interest is subtle and cannot be uncoupled from the route by which they were derived. Furthermore, there is usually no unique way to coarse-grain to an effective pair potential. Even for simple systems like liquid argon. the pair potential that correctly reproduces the pair structure will not generate the right virial pressure. Ignoring these issues in naive applications of density dependent pair potentials can lead to an apparent dependence of thermodynamic properties on the ensemble within which they are calculated. as well as other inconsistencies. These concepts are illustrated by several pedagogical examples. including effective pair potentials for systems with many-body interactions, and the mapping of charged (Debye-Huckel) and uncharged (Asakura-Oosawa) two-component systems onto effective one-component ones. The differences between the problems of transferability and representability for effective potentials are also discussed.