ELC core object can have uninitialized data members

[jngrad]

The ELC actor is set up by recycling the old struct and assigning new values to individual members, which is a potential source of bugs. For example, when the setter is called with const_pot=false, the core parameter elc_params.const_pot is not updated:

espresso/src/core/electrostatics_magnetostatics/elc.cpp

Lines 1111 to 1114 in 415f8d4

	if (const_pot) {
	elc_params.const_pot = true;
	elc_params.pot_diff = pot_diff;
	}

If the user deleted an older ELC actor that was set with const_pot=true, the new actor has constant potential enabled. MWE:

import espressomd.electrostatics

system = espressomd.System(box_l=[10, 10, 13], time_step=0.01)
system.cell_system.skin = 0.0
system.part.add(pos=[(0, 0, 1), (0, 0, 9)], q=[1, -1])

p3m = espressomd.electrostatics.P3M(prefactor=1, mesh=32, cao=5, r_cut=0.9570,
                                    alpha=2.272731, tune=False, accuracy=1e-3)
elc_params = {'p3m_actor': p3m, 'gap_size': 3., 'maxPWerror': 1e-3,
              'delta_mid_top': -1, 'delta_mid_bot': -1, 'pot_diff': -3.}

elc = espressomd.electrostatics.ELC(const_pot=True, **elc_params)
system.actors.add(elc)
print(elc.get_params()['const_pot'])  # prints True

system.actors.clear()

elc = espressomd.electrostatics.ELC(const_pot=False, **elc_params)
system.actors.add(elc)  # should fail: metallic boundaries
print(elc.get_params()['const_pot'])  # prints True instead of False

Output:

True
True

Expected:

True
Traceback (most recent call last):
  File "mwe.py", line 19, in <module>
    system.actors.add(elc)  # should fail: metallic boundaries
  File "actors.pyx", line 215, in espressomd.actors.Actors.add
  File "actors.pyx", line 70, in espressomd.actors.Actor._activate
  File "electrostatics.pyx", line 533, in espressomd.electrostatics.ELC._activate_method
  File "electrostatics.pyx", line 516, in espressomd.electrostatics.ELC._set_params_in_es_core
RuntimeError: ELC with two parallel metallic boundaries requires the const_pot option

Older versions of ESPResSo are not affected by this bug, because before 4.2 the ELC setter could only be called once.

In addition, when resizing the box in the z-direction, the gap parameter h is not updated, which leads to incorrect ELC energies being calculated and particles entering the gap region without triggering an exception:

import espressomd.electrostatics

system = espressomd.System(box_l=[10, 10, 13], time_step=0.01)
system.cell_system.skin = 0.0
system.part.add(pos=[(0, 0, 1), (0, 0, 9)], q=[1, -1])

p3m = espressomd.electrostatics.P3M(prefactor=1, mesh=32, cao=5, r_cut=0.9570,
                                    alpha=2.272731, tune=False, accuracy=1e-3)
elc_params = {'p3m_actor': p3m, 'gap_size': 3., 'maxPWerror': 1e-3,
              'delta_mid_top': -1, 'delta_mid_bot': -1, 'pot_diff': -3.}

elc = espressomd.electrostatics.ELC(const_pot=True, **elc_params)
system.actors.add(elc)
system.box_l = [10, 10, 10] # particle 2 has now entered the gap
system.analysis.energy()['coulomb']
system.part[1].pos = [0, 0, 6]
print('E =', system.analysis.energy()['coulomb'])

Output:

E = -1.6331453454189522

Expected:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "analyze.pyx", line 474, in espressomd.analyze.Analysis.energy
  File "utils.pyx", line 236, in espressomd.utils.handle_errors
  File "utils.pyx", line 255, in espressomd.utils.handle_errors
Exception: calc_long_range_energies failed: ERROR: Particle 1 entered ELC gap region by 2
E = -2.345896134603552