An improved statistical analysis for predicting the critical temperature and critical density with Gibbs ensemble...

Area of Science:

• Computational chemistry and physics
• Statistical mechanics
• Thermodynamics

Background:

• Traditional methods for estimating critical points from Gibbs ensemble Monte Carlo (GEMC) simulations often suffer from high uncertainty.
• Existing error propagation techniques inadequately account for the interdependence of vapor and liquid densities in regression analyses.
• Linearized regression approaches can lead to physically unrealistic confidence intervals for critical constants.

Purpose of the Study:

• To develop and present a rigorous statistical analysis for GEMC simulations to reduce uncertainty in critical point estimation.
• To introduce an improved error model for standard deviation estimation in regression.
• To compare a rigorous nonlinear regression algorithm against traditional linearization methods for determining critical constants.

Main Methods:

• Development of a novel error model to accurately predict standard deviations, addressing the interdependence of phase densities.
• Implementation and comparison of a rigorous nonlinear regression algorithm with traditional error propagation in linearized equations.
• Application of the enhanced methods to a case study on the n-alkane family to determine trends in critical temperature and density.

Main Results:

• The developed error model provides more accurate standard deviation estimates compared to traditional propagation of error methods.
• The rigorous nonlinear regression algorithm yields physically sensible confidence intervals for critical constants, unlike linearized approaches.
• The case study demonstrates enhanced reliability of molecular simulations for resolving trends in critical properties of n-alkanes.

Conclusions:

• The proposed statistical improvements significantly reduce uncertainty in critical point estimates from GEMC simulations.
• Rigorous nonlinear regression is essential for obtaining physically meaningful confidence intervals for critical constants.
• These advancements improve the reliability of molecular simulations for predicting thermodynamic properties across homologous series.