Hi,
i hope these kind of questions are also welcome here:
yup. for as long as you can handle a long story/email... 
Can anyone tell me, what is the state of QM/MM-coupling of Lammps to PWSCF?
it is currently broken, due to recent changes in QE, which broke the
API of the QE library interface for no good reason. i have since sent
patches to paolo gianozzi that have been included into the development
source repo, that restore the API and allow to compile the code again,
but there seem to be other changes in QE 5.4 that break using the
mechanical coupling. i have not had time to do further investigations,
let along restore the available minimal functionality.
Is there some active development on the USER-QMMM package?
there is the desire, but the problem is lack of qualified and
motivated people. that has been plaguing QM/MM interfaces since day
one. to program these, you have to be very good at programming, in
this case in three different programming languages at the same time,
have a very good understanding of parallel programming, have good
knowledge in developing classical MD codes and quantum mechanical
codes, and have a good understanding of the coupling, which has many
challenges in itself. because of that, most QM/MM coupling are not in
a good shape, as they either haphazardly graft a classical MD code
onto a QM code (CPMD's QM/MM coupling to Gromos is an example), or
wedge QM calculations into a classical MD package (Amber-Mopac would
be an example). add to that, that QM/MM isn't the hot topic anymore
that it was a decade ago, and you have a mess. there also seems to be
the trend that people, who have spent significant time working on
developing/improving/maintaining a QM/MM interface, tend to grow less
enthusiastic over time. i fall into that category.
add this all up and you get a situation, where the people most
interested in using a feature are often the least qualified and the
ones that are the most qualified are the least interested. that said,
the quantum espresso community does have a significant interest to
complete the implementation.
I noticed that the recent QE release 5.4 'broke' (as in, reordered and
redistributed the used functions) the interface that this package uses.
Does anyone have experience with
http://qeforge.qe-forge.org/gf/project/qmmmw?
yes. that is what was used as inspiration for the current
implementation. i started that about 2 1/2 years as a favor to "un
veccio amico" in trieste using code from this project. however, it
turned out to be better doing a complete rewrite and in the process
carlo cavazzoni, paolo gianozzi and i ended up doing a massive
refactoring and cleanup of the communication layer in quantum
espresso, which then allowed to write the QE library interface, and
that in turn was the basis for the QM/MM library in LAMMPS. after
doing a proof of concept implementation of a simple mechanical QM/MM
coupling, somebody from the QE developer community was supposed to
take over and validate my code and then complete the implementation by
adding the electrostatic coupling, same as qmmmw.
since then rather little has happened until earlier this year, where i
was contacted by QE developers asking me for a couple more favors to
help completing and reviewing what was done since. so far, i've been
too busy and we're still lacking a suitable developer for the QE side
of things that cares enough and is a good enough programmer.
This seems more complete, feature-wise, but is an external project and only
guaranteed to work with specific versions of both software packages.
Additionally, development there seems stalled, too.
yes, it is "mostly dead" and nobody has yet made the trip to "miracle
max" to bring it back to life.
it may be easier to get it to work, yet, it also requires programming
to make it work with the current version of QE and decent knowledge of
QE internals or an excellent ability to read code.
axel.
