Locally Adapted Reference Frame Fields using Moving Least Squares

Area of Science:

• Fluid mechanics
• Flow visualization
• Vector field analysis

Background:

• Analyzing fluid flow features is crucial in fluid mechanics.
• Current methods for computing optimal reference frames are either locally limited or globally expensive.
• Existing techniques may not effectively capture the full extent of flow features.

Purpose of the Study:

• To develop a novel objective method for computing optimal reference frames that adapt locally to flow fields.
• To overcome the limitations of fixed neighborhoods and costly global optimization in existing methods.
• To enable adaptive computation of reference frames without prior neighborhood selection.

Main Methods:

• Formulation of the problem as a moving least squares approximation.
• Determination of a continuous field of reference frames.
• Introduction of a scalar guidance field to incorporate flow features into the moving least squares approximation.
• Utilizing the guidance field to define a curved manifold for input vector field sampling.

Main Results:

• The proposed method generates a continuous field of reference frames that adapt locally to the flow.
• Using a finite-time Lyapunov exponent (FTLE) field as guidance improves adaptation to local flow features compared to prior work.
• The moving least squares framework is general and allows for future use of other guidance fields.

Conclusions:

• The novel method provides an adaptive and efficient approach to computing optimal reference frames for fluid flow analysis.
• The use of a guidance field, particularly FTLE, enhances the ability to capture local flow dynamics.
• The generalized framework offers potential for future advancements in adapting to diverse fluid features.