"insert segment end time could not be inserted." for large number of sweeps

[steffenbackes]

Dear developers,

when I run the solver doing a large number of sweeps the program crashes every time
the number of sweeps exceeds the range of an unsigned integer.

For example, if I run a 3-orbital calculation, beta=40 on 20 cores,
after about 36 minutes I get still a valid result:

total (effective) number of sweeps, normalized by N_meas: 4.20709e+09
Simulation details after 210624900 sweeps...

But after 38 minutes, which then exceeds 4.294e+09 sweeps, I get the error message:
"insert segment end time could not be inserted."
and the code exits.

best regards,
Steffen

[egull]

Hi Steffen,

that sounds plausible. For now please ramp up the number of measurements per sweep, and get the number of sweeps below the max unsigned int. Will that help?

[egull]

weird though. On first glance, all the count types are int64_t, so it should be able to run through. Could you please attach an example input?

[steffenbackes]

Hi Emanuel,

thanks for the quick response!
Ok, that sounds reasonable, but is there a parameter to increase the number of measurements per sweep? As I understand a sweep involves exactly one measurement after N_MEAS updates.
I ran into this problem because in this case the acceptance rate is quite low, so even for a large number of sweeps statistics are not that good.

Here is an input which shows the mentioned behaviour.
Depending on your hardware maybe you need to increase the MAX_TIME a bit. For me on 20 cores it takes about 38min until it crashes.

cthyb_input.tar.gz

[egull]

Thanks. I'll try to find the problem over the weekend.
If your updates are frequently rejected you will have long autocorrelation times and measuring less frequently (i.e. increasing N_MEAS) won't make much of a difference. But we should then also go over the reason why you get stuck and think of updates to avoid getting stuck in the first place.