CV = nan in gmx dos?

GROMACS version: 2018
GROMACS modification: No

Dear all,

I’m having some issue trying to use gmx dos (which is a bit new for me, so, sorry if the question is too basic).

In the very end, our objective will be to use in a similar way as Caleman et al., JCTC 2012, 61-74. Our systems however differs from the ones in the above paper by two main characteristics: they are mixtures of different species (rather than single specie systems), and there are free charged species (cations and anions) involved.

Before we dig into the 2PT model, we decided to initially calculate both the gmx energy (classical) and gmx dos (properties): among them, we calculated the constant volume heat capacity (CV). It ran as well as possible for the classical calculation, but the gmx dos, specifically and only in this property, yield a nan result.

However, from our understanding, this two complicating factors should not, specially since they do not affect in that way the calculation of any other properties reported by the same program over the same testing trajectories (25ps saved at each 1fs), lead the calculation of this property to this error (we also understand that gmx dos should not give directly equivalent results to the above paper). We don’t really see that there is any wrong assumption on this.

If so, have anybody had such a problem before? Does anyone have any idea on the reason why this would happen, something specific I could focus on?

Thanks a lot in advance for any possible suggestion,

Best regards,