Benchmark showing the computational scaling of celerite using the Cholesky factorization routine for rank-﹩2\,J﹩ semiseparable matrices as described in the text. Left: cost of computing Equation (3) with a covariance matrix given by Equation (8) as a function of the number of data points N. The different lines show the cost for different numbers of terms J increasing from bottom to top. To guide the eye, the straight black line without points shows linear scaling in N. For comparison, the computational cost for the same model using a general Cholesky factorization routine implemented in the Intel MKL linear algebra package is shown as the dashed black line. Right: same information as in the left panel, plotted as a function of J for different values of N. Each line shows the scaling for a specific value of N increasing from bottom to top. The black line shows quadratic scaling in J.