TY - JOUR

T1 - Sensitivity of pair statistics on pair potentials in many-body systems

AU - Wang, Haina

AU - Stillinger, Frank H.

AU - Torquato, Salvatore

N1 - Funding Information:
This work was supported by the National Science Foundation under Award No. CBET-1701843.

PY - 2020/9/28

Y1 - 2020/9/28

N2 - We study the sensitivity and practicality of Henderson’s theorem in classical statistical mechanics, which states that the pair potential v(r) that gives rise to a given pair correlation function g2(r) [or equivalently, the structure factor S(k)] in a classical many-body system at number density ρ and temperature T is unique up to an additive constant. While widely invoked in inverse-problem studies, the utility of the theorem has not been quantitatively scrutinized to any large degree. We show that Henderson’s theorem has practical shortcomings for disordered and ordered phases for certain densities and temperatures. Using proposed sensitivity metrics, we identify illustrative cases in which distinctly different potential functions give very similar pair correlation functions and/or structure factors up to their corresponding correlation lengths. Our results reveal that due to a limited range and precision of pair information in either direct or reciprocal space, there is effective ambiguity of solutions to inverse problems that utilize pair information only, and more caution must be exercised when one claims the uniqueness of any resulting effective pair potential found in practice. We have also identified systems that possess virtually identical pair statistics but have distinctly different higher-order correlations. Such differences should be reflected in their individually distinct dynamics (e.g., glassy behaviors). Finally, we prove a more general version of Henderson’s theorem that extends the uniqueness statement to include potentials that involve two- and higher-body interactions.

AB - We study the sensitivity and practicality of Henderson’s theorem in classical statistical mechanics, which states that the pair potential v(r) that gives rise to a given pair correlation function g2(r) [or equivalently, the structure factor S(k)] in a classical many-body system at number density ρ and temperature T is unique up to an additive constant. While widely invoked in inverse-problem studies, the utility of the theorem has not been quantitatively scrutinized to any large degree. We show that Henderson’s theorem has practical shortcomings for disordered and ordered phases for certain densities and temperatures. Using proposed sensitivity metrics, we identify illustrative cases in which distinctly different potential functions give very similar pair correlation functions and/or structure factors up to their corresponding correlation lengths. Our results reveal that due to a limited range and precision of pair information in either direct or reciprocal space, there is effective ambiguity of solutions to inverse problems that utilize pair information only, and more caution must be exercised when one claims the uniqueness of any resulting effective pair potential found in practice. We have also identified systems that possess virtually identical pair statistics but have distinctly different higher-order correlations. Such differences should be reflected in their individually distinct dynamics (e.g., glassy behaviors). Finally, we prove a more general version of Henderson’s theorem that extends the uniqueness statement to include potentials that involve two- and higher-body interactions.

UR - http://www.scopus.com/inward/record.url?scp=85092514913&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85092514913&partnerID=8YFLogxK

U2 - 10.1063/5.0021475

DO - 10.1063/5.0021475

M3 - Article

C2 - 33003740

AN - SCOPUS:85092514913

VL - 153

JO - Journal of Chemical Physics

JF - Journal of Chemical Physics

SN - 0021-9606

IS - 12

M1 - 124106

ER -