Searching for local order parameters to classify water structures at triple points.

Area of Science:

• Condensed-matter physics
• Materials science
• Planetary science

Background:

• The diversity of ice polymorphs is crucial for understanding phase transitions.
• Distinguishing between ice polymorphs' molecular structures is challenging.
• Order parameters are vital for quantifying structural ordering and classifying phases.

Purpose of the Study:

• To systematically investigate the capabilities of order parameters in distinguishing ice polymorphs.
• To identify optimal sets of order parameters for classifying specific ice triple points.
• To develop a robust method for differentiating ice structures.

Main Methods:

• Consideration of 493 order parameters and their combinations.
• Application of supervised machine learning for systematic parameter searching.
• Analysis focused on two triple points: ice III-V-liquid and ice V-VI-liquid.

Main Results:

• Identification of the best set of two order parameters for distinguishing three ice structures at each triple point with high accuracy.
• Demonstration of machine learning's effectiveness in automating the search for distinguishing parameters.
• Suggestion of a set of three order parameters for enhanced accuracy in structure classification.

Conclusions:

• A systematic approach using machine learning and order parameters can effectively distinguish ice polymorphs.
• Specific sets of order parameters provide high accuracy in classifying ice structures at critical triple points.
• This methodology offers a valuable tool for research in condensed-matter physics, engineering, and climate science.