Figure 6.

The estimated probability densities from the large dataset (in red) consisting of 5000 data-points and the 1000 small datasets (in blue) each consisting of 40 data-points. The datasets were generated from the probability density function, $p(x)={\mathrm{e}}^{a{x}^2-b{x}^4}/Z$, where Z is the normalizing constant based on the free energy landscape, F(x) = −ax² + bx⁴, from Landau’s second order phase transition formalism.