Skip to contents

Build a Vector Field Interpolation Function from a Grid Data Frame

build_field_fun.Rd

Converts the output of grid_vectors into a callable function f(c(x, y)) -> c(fx, fy) suitable for use as the fun argument of ggvfields::geom_stream_field(). Interpolation is performed using inverse-distance weighting (IDW) over the four nearest grid points.

Usage

build_field_fun(grid_df, emb = NULL, density_thresh = 0)

Arguments

grid_df

A data frame produced by grid_vectors, containing columns start.emb_1, start.emb_2 (grid point positions) and end.xd, end.yd (grid point endpoints).

emb

A cells-by-2 matrix of embedding coordinates. Required when density_thresh > 0 to compute the KDE.

density_thresh

Numeric in [0, 1]. Normalised KDE density below which the field returns zero velocity. Default 0 disables the mask.

Value

A function that accepts a numeric vector c(x, y) and returns the IDW-interpolated vector components c(fx, fy), or c(0, 0) when the query point falls in a low-density region.

Details

When density_thresh > 0, a 2D kernel density estimate (KDE) of the cell positions is computed via MASS::kde2d() and normalised to [0, 1]. Any query point whose normalised density falls below density_thresh returns a zero vector c(0, 0), causing geom_stream_field() to terminate the streamline at that point.