Abrupt slope change at Z = 0.0220 in the DCO yield vs Z (i.e. metallicity) plot

[dumbscholar]

I'm getting a steep drop in the DCO count between metallicities 0.0215 and 0.0220, and 5-150 M_sol mass range. From 0.0220, the decline suddenly becomes gradual. The slope changes abruptly at 0.0220 (this looks out of place and concerning). I have checked multiple times. The trend appears every time.

For a given metallicity, this is my command:

./COMPAS --mode 'BSE' --eccentricity-distribution 'THERMAL' --number-of-systems 1000000 --detailed-output False --logfile-type 'CSV' --allow-rlof-at-birth False --metallicity 0.0001 --random-seed 0 | tee runlog.txt

In the attached figure, I show two options. One, by changing the random seed (advancing the random seed by 1 million for subsequent runs) for each metallicity so that no two ZAMS binaries have the same (other than metallicity) property. Second option, by keeping the random seed same, so that the same set of binaries are simulated each time with different metallicity. The term 'corrected' in the legend means each yield is multiplied by a correction factor of 0.02.

I have tried both thermal and flat eccentricity distributions. I have checked with v02.49.05 (older) and v02.50.01 (latest) versions. I have tried setting the option --common-envelope-lambda-nanjing-interpolate-in-metallicity to True (default is False). But it produced the exact same DCO yields as earlier when it was False.

I have checked with the options "--mass-loss-prescription 'BELCZYNSKI2010' --wolf-rayet-multiplier 0.1". This increased the DCO yield by 10% but the yield vs Z curve was similar as before with abrupt change of slope at Z=0.0220.

version: 2.49.05

[ilyamandel]

Thank you, this is very interesting, @dumbscholar !

Alas, I am traveling for the next month and won't be able to look into this for some time. If you have a chance, I recommend comparing individual binaries (same seed) just below/above the metallicity cut. Judging by the side of the jump, some 5% will suddenly go from becoming DCOs to not becoming DCOs. Can you understand why this is happening? Is it SN kicks disrupting the binary, or mergers inside a common envelope, or... ? Such an investigation will help understand what the culprit might be.

Thank you in advance for looking into this!

[dumbscholar]

One thing I noticed is that, going from right to left in the plot, as we go from 0.0220 to 0.0219, the NS production shoots up suddenly leading to an abundance of NS-NS and NS-BH binaries. The BH-BH binary population does not change abruptly and remains relatively level. I wonder if there is any metallicity-related hard threshold (in the code) such as the progenitor core mass that when exceeded results in NS formation.

[ilyamandel]

Hi @dumbscholar ,

Sorry, back from travels now.

I wonder if there is any metallicity-related hard threshold (in the code) such as the progenitor core mass that when exceeded results in NS formation.

Not that I am aware of, but there is an easy way to check. Just run two grids of single stars (SSE) of different masses at the two metallicities in question and compare outcomes. Is there an abrupt change? As mentioned earlier, to properly diagnose population-level trends, you'll need to compare individual binaries -- or even individual stars, for this particular hypothesis.

[ilyamandel]

See #1343 . Would be good to understand what's happening here. @themikelau ?

[themikelau]

There's nothing that comes to mind regarding hard-coded metallicity values at Z=0.0220. I agree with @ilyamandel's advice to check the detailed evolution of systems below and above the metallicity at which the drop occurs.

[ilyamandel]

@dumbscholar -- have you considered the suggestions above? Is there still an issue with the latest code version?

[dumbscholar]

@ilyamandel the abrupt change in the DCO yield at Z=0.0220 persists even with COMPAS version 3.23.00