Large Memory Use Resulting in Crash

[herkesg]

Hi,

I've been running into problems with large memory use. The code works well for a test case which I ran where 10 million particles within a unit cube and random masses were used:

Interestingly, the code crashes when I pass in particle positions and masses for approximately 8.6 million particles.

It produces the following error:

The server I am running on has approximately 129Gbs of ram. The thread memory use grows gradually until it has filled up the ram at which point it crashes. I am unsure why this is the case when the problem does not occur with the 10 million particle simulation. I initially thought the problem was to do with the particles not lying within a unit cube, and the particle masses not being between 0 and 1, but even after rescaling them so that this is the case, the error is still produced.

Any tips/suggestions would be greatly appreciated. If there is any important information I have missed don't hesitate to let me know!

Thanks in advance, Geoff

[mikegrudic]

Hi Geoff,

Thanks for reporting this. Are the particle positions in the failed test case unique? I have killed the code in the past with particle datasets that contain two at the same position, something I should eventually fix...

[herkesg]

Hi Mike,

That's an excellent point. I assumed that they were unique but upon checking found that there were actually quite a few (around 20,000) that were not, which took me quite by surprise! I imagine that is what is causing the problem. I shall try removing them and see if that fixes it. Thanks for the tip.

Geoff

[herkesg]

Hi Mike,

Just confirming that the cause of the problem was that there were multiple particles with the same position. Removing them fixed the problem. Thanks for your help :)

Geoff