Some strange behavior of maddm near s-channel resonant pole

[WenbinZhao]

Basically I was investigating a dirac dark matter model whose relic density is governed by a s-channel process and the mediator is a gauge boson whose mass is around 3 times dark matter mass. So the process is dominated by pole behavior.
And if I define r=m_dm/m_boson, I would like to see as r grows, the relic density should drop as we are closer to the pole.
In general it works in maddm, but there are some fluctuations in the curve. I tried to identify what's going wrong here and found that the <sigmav> calculation might be the reason. I tried to change the following parameters ngrid_init and nres_points to 500 and 1000. But I didn't see any improvement. (Also iter_wij)
Since in this case the low velocity is favored by Maxwell Distribution and large velocity is favored by cross-section, there might be something hard to handle by the code.
Is there any way to improve the results? Or I simply missed some other important parameters which might also play a role here?

[J. Heisig (heisig)]

Hi WenbinZhao,

Thanks for the bug report. Could you attach (or send me) your model and param_card you used (plus a short explanation on what mass you fixed when varying r), so I can try to reproduce your results?

Best,
Jan

[WenbinZhao (s6wnzhao)]

Dear Jan,

Please find the model in the attachment. This is a simple L_\mu - L_\tau
model.
I only have 4 BSM parameters here{*qpsi,mx,ra,gX*} : one is dm charge *qpsi*,
which is not relevant with our problem. Another is *mx*, which is the mass
of the mediator gauge boson. Then the *ra*, defines the ratio between dark
matter mass and gauge boson mass. And *gX* defines the gauge coupling. They
are in *L23input* section.
when varying r (ra) I fixed all the other parameters. I think their value
is the same as in launch.txt.
I also put the output file in the attachment.
The plot <sigmav> vs r is somehow gotten by hand. I used the function
*taacs_check
*for each point and picked a certain value.
Hope everything will work out well.

Bests,
Wenbin

On Wed, May 12, 2021 at 2:25 PM J. Heisig <email address hidden>
wrote:

> Hi WenbinZhao,
>
> Thanks for the bug report. Could you attach (or send me) your model and
> param_card you used (plus a short explanation on what mass you fixed
> when varying r), so I can try to reproduce your results?
>
> Best,
> Jan
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1928212
>
> Title:
> Some strange behavior of maddm near s-channel resonant pole
>
> Status in MadDM:
> New
>
> Bug description:
> Basically I was investigating a dirac dark matter model whose relic
> density is governed by a s-channel process and the mediator is a gauge
> boson whose mass is around 3 times dark matter mass. So the process is
> dominated by pole behavior.
> And if I define r=m_dm/m_boson, I would like to see as r grows, the
> relic density should drop as we are closer to the pole.
> In general it works in maddm, but there are some fluctuations in the
> curve. I tried to identify what's going wrong here and found that the
> <sigmav> calculation might be the reason. I tried to change the following
> parameters ngrid_init and nres_points to 500 and 1000. But I didn't see any
> improvement. (Also iter_wij)
> Since in this case the low velocity is favored by Maxwell Distribution
> and large velocity is favored by cross-section, there might be something
> hard to handle by the code.
> Is there any way to improve the results? Or I simply missed some other
> important parameters which might also play a role here?
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/maddm/+bug/1928212/+subscriptions
>