Softening

[dipto4]

Hi,
Is there any possible mechanism to include force softening inside NBODY6++GPU?

Thank you.

[lwang-astro]

You can try to modify the force integration part in the code if you want to add softening. However, any regularzation should be complelely switched off to make it works.

[dipto4]

So the softening parameter can be included in regint.f which calculates the regular force? Would you recommend including it somewhere else?

[lwang-astro]

The neighbor force is more important than regular force (regint.f). But if the GPU and AVX are used, the AVX and GPU force calculation functions also need to be modified, which is more complicated. Besides, the energy calculation also should have consistent potential. Windows 10 版のメ`ル<https://go.microsoft.com/fwlink/?LinkId=550986>から送信 差出人: Diptajyoti Mukherjee<mailto:[email protected]> 送信日�r: 2021年1月28日 15:28 宛先: nbodyx/Nbody6ppGPU<mailto:[email protected]> CC: lwang-astro<mailto:[email protected]>; Comment<mailto:[email protected]> 件名: Re: [nbodyx/Nbody6ppGPU] Softening (#32) So the softening parameter can be included in regint.f which calculates the regular force? Would you recommend including it somewhere else? ― You are receiving this because you commented. Reply to this email directly, view it on GitHub<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnbodyx%2FNbody6ppGPU%2Fissues%2F32%23issuecomment-768831231&data=04%7C01%7C%7C05ee11b3315945c0d55708d8c355f6e6%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637474121314510387%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=3PiYF%2B8manzXUPBXd3G5yCr4NPiOcQCpV5v4AlxBq3I%3D&reserved=0>, or unsubscribe<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACB2PKX2O2I3ZMNZZAE6QHDS4D7ZZANCNFSM4WWNP5IA&data=04%7C01%7C%7C05ee11b3315945c0d55708d8c355f6e6%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637474121314510387%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=v6E%2B6pMU2JFTTy%2FFCENfgeQ0%2Bz7aND5Op1Nok%2BHLv%2FU%3D&reserved=0>.

[dipto4]

I am running the code without GPU and AVX. Is the neighbor force calculated in nbint.f in that case?

[lwang-astro]

Yes. Windows 10 版のメ`ル<https://go.microsoft.com/fwlink/?LinkId=550986>から送信 差出人: Diptajyoti Mukherjee<mailto:[email protected]> 送信日�r: 2021年1月28日 15:37 宛先: nbodyx/Nbody6ppGPU<mailto:[email protected]> CC: lwang-astro<mailto:[email protected]>; Comment<mailto:[email protected]> 件名: Re: [nbodyx/Nbody6ppGPU] Softening (#32) I am running the code without GPU and AVX. Is the neighbor force calculated in nbint.f in that case? ― You are receiving this because you commented. Reply to this email directly, view it on GitHub<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnbodyx%2FNbody6ppGPU%2Fissues%2F32%23issuecomment-768834366&data=04%7C01%7C%7C146db65a9c8e47384ea508d8c3572387%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637474126304251562%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=I%2BSueSUeX51HlwwxhYNQmUmsHJgFf4A68dj5iQy256Q%3D&reserved=0>, or unsubscribe<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACB2PKRZ4SET5K3DXOCA5T3S4EAZLANCNFSM4WWNP5IA&data=04%7C01%7C%7C146db65a9c8e47384ea508d8c3572387%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637474126304261553%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=qieA%2BIPtYAHC4tT0HgfmIOs0bvqQcE7JS%2FN0BWUPBZw%3D&reserved=0>.

[dipto4]

Hi,
You mentioned that if we manually insert softening, we would need to correct the energy. Where would we factor in this correction for a softening parameter \epsilon ?

Thank you.

[lwang-astro]

In energy.F and energy_mpi.F. 差出人: Diptajyoti Mukherjee<mailto:[email protected]> 送信日時: 2021年2月19日 12:27 宛先: nbodyx/Nbody6ppGPU<mailto:[email protected]> CC: lwang-astro<mailto:[email protected]>; Comment<mailto:[email protected]> 件名: Re: [nbodyx/Nbody6ppGPU] Softening (#32) Hi, You mentioned that if we manually insert softening, we would need to correct the energy. Where would we factor in this correction for a softening parameter \epsilon ? Thank you. — You are receiving this because you commented. Reply to this email directly, view it on GitHub<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnbodyx%2FNbody6ppGPU%2Fissues%2F32%23issuecomment-781789479&data=04%7C01%7C%7C7fedcdcaafdc4788347708d8d4864470%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637493020424526137%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=yts%2FhT8qZpUjt3oKubR6XFTovMVQyx48ABFjnL8GWwc%3D&reserved=0>, or unsubscribe<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FACB2PKTXEJYXQYKF6T34XVDS7XLBRANCNFSM4WWNP5IA&data=04%7C01%7C%7C7fedcdcaafdc4788347708d8d4864470%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637493020424526137%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=64gyJGXb3IyrNukFXXY44Zv0eY7CjUl4QKI5eB3hnY0%3D&reserved=0>.