In this paper we study the SU(3)-gluodynamics shear viscosity temperature dependence on the lattice. To do so, we measure the correlation functions of the energymomentum tensor in the range of temperatures T /T c ∈ [0.9, 1.5]. To extract the shear viscosity we used two approaches. The first one is to fit the lattice data with a physically motivated ansatz for the spectral function with unknown parameters and then determine the shear viscosity. The second approach is to apply the Backus-Gilbert method allowing to extract the shear viscosity from the lattice data nonparametrically. The results obtained within both approaches agree with each other. Our results allow us to conclude that within the range T /T c ∈ [0.9, 1.5] the SU(3)-gluodynamics reveals the properties of a strongly interacting system, which cannot be described perturbatively, and has the ratio η/s close to the value 1/4π of the N = 4 Supersymmetric Yang-Mills theory.

Neural quantum states (NQS) attract a lot of attention due to their potential to serve as a very expressive variational ansatz for quantum many-body systems. Here we study the main factors governing the applicability of NQS to frustrated magnets by training neural networks to approximate ground states of several moderately-sized Hamiltonians using the corresponding wave function structure on a small subset of the Hilbert space basis as training dataset. We notice that generalization quality, i.e. the ability to learn from a limited number of samples and correctly approximate the target state on the rest of the space, drops abruptly when frustration is increased. We also show that learning the sign structure is considerably more difficult than learning amplitudes. Finally, we conclude that the main issue to be addressed at this stage, in order to use the method of NQS for simulating realistic models, is that of generalization rather than expressibility.