Expansion by regions seems to require a lot of time

[vsht]

Hi,

I'm considering a 2-loop 4-line integral with 2 scales, where I would like to obtain the expansion in mg up to 4th order.
The generation code reads as follows

#!/usr/bin/env python3
from pySecDec import sum_package, loop_package, loop_regions, LoopIntegralFromPropagators
import pySecDec as psd

li = LoopIntegralFromPropagators(
['-mg**2 + p1**2', '-mb**2 + (p3 + q1)**2', '-mb**2 + (p1 + q1)**2', '-mg**2 + (p1 - p3)**2'],
loop_momenta = ['p1', 'p3'],
powerlist = [1, 1, 1, 1],
dimensionality = '4 - 2*eps',
Feynman_parameters = 'x',
replacement_rules = [('q1**2','mb**2')],
regulators = ['eps']
)

regions_generator_args = loop_regions(
name = 'loopint',
loop_integral = li,
smallness_parameter = 'mg',
expansion_by_regions_order = 4,
contour_deformation = False,
additional_prefactor = '(1)*exp(2*EulerGamma*eps)',
decomposition_method = 'geometric'
)


sum_package(
name = 'loopint',
package_generators = regions_generator_args,
regulators = li.regulators,
requested_orders = [2],
real_parameters = ['mb', 'mg']
)

However, alone the preparation stage is running for almost 6 hours and is still far from being done.
output.txt

This makes me wonder whether it is because I'm doing something wrong or if this expansion
is really so tough that it requires hours just for the generation of the library.

[vsht]

I gave up after 27 hours on the cluster, since this case appears rather hopeless to me
and is unlikely to finish now. The z-method had the same runtime (I started both with a small delay).

The beta of FIESTA did it in in 1.5 hours, which is why I suppose that the problem is not pySecDec
per se, but rather my unfortunate config file.