Contour deformation for Hubbard Model

Workflow

First define a physical model with a text file. For example, with improved Gaussian action, the following is saved as model.dat:

hubbard.ImprovedGaussianModel(
L=4,
nt=16,
Kappa=1,
U=8,
Mu=1.5,
dt=1/16
)

Then use contour.py to find a contour to maximize the average sign. After one finds a contour, generate configurations with the given contour using sample.py. Finally, one can estimate observables with errors through bootstrap.py.

Here is an example:

./contour.py model.dat c.pickle -l 1 -w 1 -lr 1e-3 # Terminate with CTRL-C
./sample.py model.dat c.pickle -N 1000 > sam.dat &
./bootstrap.py < sam.dat

To compare it with the real-plane Monte Carlo, generate the real plane contour and sample it:

./contour.py model.dat c_real.pickle -R
./sample.py model.dat c_real.pickle -N 1000 > sam_real.dat &
./bootstrap < sam_real.dat