We propose a new approach to circumvent the sign problem in which the integration path is optimized to control the sign problem. We give a trial function specifying the integration path in the complex plane and tune it to optimize the cost function which represents the seriousness of the sign problem. We call it the path optimization method. In this method, we do not need to solve the gradient flow required in the Lefschetz-thimble method and then the construction of the integration-path contour arrives at the optimization problem where several efficient methods can be applied. In a simple model with a serious sign problem, the path optimization method is demonstrated to work well; the residual sign problem is resolved and precise results can be obtained even in the region where the global sign problem is serious.
The Polyakov loop extended Nambu–Jona-Lasinio (PNJL) model with imaginary chemical potential is studied. The model possesses the extended symmetry that QCD does. Quantities invariant under the extended symmetry, such as the partition function, the chiral condensate, and the modified Polyakov loop, have Roberge-Weiss periodicity. The phase diagram of confinement/deconfinement transition derived with the PNJL model is consistent with the Roberge-Weiss prediction on it and the results of lattice QCD. The phase diagram of chiral transition is also presented by the PNJL model.
We introduce the feedforward neural network to attack the sign problem via the path optimization method. The variables of integration are complexified and the integration path is optimized in the complexified space by minimizing the cost function, which reflects the seriousness of the sign problem. For the preparation and optimization of the integral path in multi-dimensional systems, we utilize the feedforward neural network. We examine the validity and usefulness of the method in the 2D complex |$\lambda \phi^4$| theory at finite chemical potential as an example of the quantum field theory with the sign problem. We show that the average phase factor is significantly enhanced after the optimization and then we can safely perform the hybrid Monte Carlo method.
The fermion sign problem appearing in the mean-field approximation is considered, and the systematic computational scheme of the free energy is devised by using the Lefschetz-thimble method. We show that the Lefschetz-thimble method respects the reflection symmetry, which makes physical quantities manifestly real at any order of approximations using complex saddle points. The formula is demonstrated through the Airy integral as an example, and its application to the Polyakov-loop effective model of dense QCD is discussed in detail.