Reconstructing QCD spectral functions with Gaussian processes

We reconstruct ghost and gluon spectral functions in $2+1$ flavor QCD with Gaussian process regression. This framework allows us to largely suppress spurious oscillations other common reconstruction artifacts by specifying generic magnitude length scale parameters the kernel function. The Euclidean propagator data are taken from lattice simulations domain wall fermions at physical point. For infrared ultraviolet extensions of propagators as well low-frequency asymptotics function, we utilize results functional computations Yang-Mills theory QCD. further reduces systematic error significantly. Our numerical compared against a direct real-time computation an earlier theory. approach presented this work offers promising route toward unveiling properties