## Frequently Asked Questions

### Q: How do I fit a GLE matrix for my system?

A: Fitting GLE parameters from scratch is not for the faint hearted. However, parameters for a given application
can typically be transfered easily from a system to another. The input page
provides a convenient interface to fetch pre-computed parameters from a library and adapt them to the application
at hand.

### Q: The conserved quantity has a noticeable drift when I use a GLE. Is there a problem?

A: Having small drift of the conserved quantity in a constant-temperature simulation is a
sufficient but not necessary condition for accurately sampling the canonical ensemble. Optimal-sampling
GLEs - just as aggressive white-noise thermostatting - interferes with the accuracy of integration
and increases the drift of the conserved quantity. In general, however, this does not lead
to measurable degradation of the accuracy of computed observables.

### Q: How do I choose the value of ℏω/kT in a QT/PI+GLE/PIGLET calculation?

A: You just need to know an order-of-magnitude estimate of the maximum frequency in your system.
Pick the value of ℏω/kT that approaches more closely this frequency: the colored noise
will try to reproduce quantum effects for modes up to this cutoff. Choosing a value which is too
high is not inherently problematic, but might require the use of a smaller time step, since the
noise contains then very high frequency components.

### Q: The temperature in my quantum thermostat/PI+GLE/PIGLET run is not what I indicated. What's wrong?

A: Nothing is wrong. The "temperature" output by MD codes is typically just proportional to the instantaneous
value of the classical kinetic energy, relying on the relation
\[
\left\langle T\right\rangle=\frac{2}{3N_fk_B} \left\langle K\right\rangle.
\]
Clearly this relation only holds in a classical context. In a quantum-thermostat simulation
the mean kinetic energy is (an approximation to) the quantum expectation value, that is
not related in a simple way to the physical (ensemble) temperature. In a PI+GLE or PIGLET
simulation, the particle momenta have no physical meaning, and there is no obvious meaning
to the value of temperature printed by the code.