The paper introduces a new probability distribution built from two mathematical tools chained together. The first tool, the Lehmer transform, takes a set of positive measurements and produces a smooth curve by computing ratios of power sums at successive orders. This curve always runs from the smallest observation to the largest and captures how the data's "center of gravity" shifts as you change the exponent used to average the values. The second tool maps that curve onto the entire real line using the Lambert W function, which is a well-known mathematical inverse function. The result is a new probability distribution called the Lehmer-Lambert distribution.
The distribution has several attractive mathematical properties. Its key formulas, including the cumulative distribution, the density, and the quantile function, all have clean closed-form expressions, meaning you can write them out explicitly rather than relying on numerical approximation. Random samples from the distribution can be generated exactly. The distribution is also invariant under rescaling the signal or raising it to a power, which makes it robust to certain common transformations of data. The peaks of the distribution occur at points where the underlying signal changes scale, and the tails of the distribution decay exponentially at rates determined by the gaps between the largest and smallest values in the sample. The authors also derive moments and other standard statistical quantities, and they prove limit theorems describing the distribution's behavior in large samples.
To demonstrate the distribution's practical value, the authors apply it to electroencephalogram (EEG) recordings from patients with major depressive disorder. Brain signals recorded by EEGs are positive-valued time series that span multiple scales, making them a natural candidate for this kind of analysis. The Lehmer-Lambert framework provides a way to characterize the statistical structure of these signals and potentially distinguish patterns associated with depression from those seen in healthy individuals. The authors suggest that the distribution's sensitivity to scale changes in the signal could make it a useful tool for identifying neurological signatures of psychiatric conditions.