WHAM boundaries bug?

GROMACS version:2019.3
GROMACS modification:No

Hello everyone,

I’m doing an umbrella sampling simulation where I’m pulling a solute from an alkane phase through a water phase and back into the alkane phase. I have around 70 windows either side of the COM of the water phase and for evaluation I tried evaluating them separately and combined.
I’m getting problems like this:

Evaluating only 554 of 14000 expressions.

   1) Maximum change 6.810421e-02

Switched to exact iteration in iteration 2
Converged in 3 iterations. Final maximum change -1e+20

Weirdly I didn’t get those all the time and narrowed it down to setting the -min -max boundaries. The auto setting gave me -min 0.000002 -max 6.641400, which produces results. If I go below 6.4 with the upper bound or above 1 with the lower one I get the above error (this varies depending on the windows I include).

As far as I’m aware of the wham algorithm there is no reason for this to happen. Histograms are fine.
However, I’m happy to be wrong on this and if someone can point out how I can make it work for custom boundaries I would be very grateful.
The full command:

gmx wham -temp 303.15 -it lhs-tpr-files.dat -if lhs-force-files.dat -min 0.1 -max 6.6

If people agree that this is probably a bug I’m happy to report it.

While I’m here I also wanted to ask if someone could elucidate me on the meaning of this line: “Evaluating only 554 of 14000 expressions.”

I have the vage feeling this is not a good thing, but I have absolutely no clue what expressions are talked about and therefore cannot say how to improve the number of evaluated expression (and if that’s necessary).

Thank you for your help,