linthresh

linthresh is a parameter used in Matplotlib’s SymLogNorm color normalization to control how the scale treats values near zero. It defines the width of the linear region around zero in a symmetric logarithmic scale, allowing both very small and very large values to be visualized in a single plot without the distortion caused by a pure log scale.

In practice, values with absolute value less than or equal to linthresh are mapped linearly, while values with larger magnitude are mapped logarithmically, preserving the sign of the data. The transition between the linear and logarithmic regions is governed by the linthresh value, and the relative size of the linear region is further influenced by the linscale parameter. This combination enables a visualization that spans several orders of magnitude while remaining interpretable near zero.

linthresh is primarily encountered in Matplotlib’s SymLogNorm, which is useful for data that contains both small and large positive and negative values, such as certain scientific measurements or residuals. It is not a universal mathematical constant, but a user-settable parameter within a plotting context.

Typical usage involves creating a normalization object with a chosen linthresh, for example: SymLogNorm(linthresh=0.01, linscale=1.0). The exact defaults depend on the library version, so users should consult current documentation when applying it. In summary, linthresh provides a practical way to blend linear and logarithmic scaling to improve the visualization of data that straddles zero and spans wide ranges.