Increase mosaic_domains to 100

[nksauter]

Currently two defaults are used, 100 domains for Bragg-only and 1 domain for Bragg+diffuse. Change this to one default only covering all cases.

[mewall]

I added the different Bragg+diffuse default as the run time is prohibitively slow using 100 domains.

…

________________________________ From: nksauter ***@***.***> Sent: Tuesday, May 9, 2023 2:39:07 AM To: dermen/sim_erice Cc: Subscribed Subject: [EXTERNAL] [dermen/sim_erice] Increase mosaic_domains to 100 (Issue #11) Currently two defaults are used, 100 domains for Bragg-only and 1 domain for Bragg+diffuse. Change this to one default only covering all cases. — Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/dermen/sim_erice/issues/11__;!!Bt8fGhp8LhKGRg!C5-9naeOcXAKOZLkVzwAc5TL64j5q6ZLKwCcuFv0qgxruChsoq6b-FumN6RcT1k_PSTFuqaj1r3OUxYa-OsZAiHZ$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AA67VELI4CLWBP3UOINCZDLXFH7CXANCNFSM6AAAAAAX26FVWA__;!!Bt8fGhp8LhKGRg!C5-9naeOcXAKOZLkVzwAc5TL64j5q6ZLKwCcuFv0qgxruChsoq6b-FumN6RcT1k_PSTFuqaj1r3OUxYa-NsYba59$>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[nksauter]

@mewall our development target is for a platform with GPU acceleration. As it happens, 100 domains is OK for the Bragg+diffuse case. Try this on Perlmutter:

DIFFBRAGG_USE_CUDA=1 libtbx.python sim_view.py

I get a response time of 0.018 seconds!

[mewall]

OK, that's fair. I was previously targeting laptops at the Erice workshop.

…

________________________________ From: nksauter ***@***.***> Sent: Tuesday, May 9, 2023 10:41:07 AM To: dermen/sim_erice Cc: Wall, Michael E; Mention Subject: [EXTERNAL] Re: [dermen/sim_erice] Increase mosaic_domains to 100 (Issue #11) @mewall<https://urldefense.com/v3/__https://github.com/mewall__;!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEG1xL5X10$> our development target is for a platform with GPU acceleration. As it happens, 100 domains is OK for the Bragg+diffuse case. Try this on Perlmutter: DIFFBRAGG_USE_CUDA=1 libtbx.python sim_view.py<https://urldefense.com/v3/__http://sim_view.py__;!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEG0iuQpa4$> I get a response time of 0.018 seconds! — Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/dermen/sim_erice/issues/11*issuecomment-1540514870__;Iw!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEG0eNx-nH$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AA67VEO53PUSEIHUMZJSNI3XFJXSHANCNFSM6AAAAAAX26FVWA__;!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEGxSVu4zC$>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[mewall]

@nksauter Do you want the default to be Monochromatic or Gaussian for the Bragg+diffuse case? Currently it's Monochromatic.

…

________________________________ From: Wall, Michael E Sent: Tuesday, May 9, 2023 11:06:26 AM To: dermen/sim_erice; dermen/sim_erice Cc: Mention Subject: Re: [EXTERNAL] Re: [dermen/sim_erice] Increase mosaic_domains to 100 (Issue #11) OK, that's fair. I was previously targeting laptops at the Erice workshop.

________________________________ From: nksauter ***@***.***> Sent: Tuesday, May 9, 2023 10:41:07 AM To: dermen/sim_erice Cc: Wall, Michael E; Mention Subject: [EXTERNAL] Re: [dermen/sim_erice] Increase mosaic_domains to 100 (Issue #11) @mewall<https://urldefense.com/v3/__https://github.com/mewall__;!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEG1xL5X10$> our development target is for a platform with GPU acceleration. As it happens, 100 domains is OK for the Bragg+diffuse case. Try this on Perlmutter: DIFFBRAGG_USE_CUDA=1 libtbx.python sim_view.py<https://urldefense.com/v3/__http://sim_view.py__;!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEG0iuQpa4$> I get a response time of 0.018 seconds! — Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/dermen/sim_erice/issues/11*issuecomment-1540514870__;Iw!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEG0eNx-nH$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AA67VEO53PUSEIHUMZJSNI3XFJXSHANCNFSM6AAAAAAX26FVWA__;!!Bt8fGhp8LhKGRg!AGJIt00wXxsLOQnScFGLt1BTUDggCwuXpdINLiwWnz3XTEZMbUcVEfdGHgSLboImnEs5JXCxYUjBQ_TEGxSVu4zC$>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[nksauter]

@mewall I get 0.4 seconds if set to Bragg+diffuse and Gaussian or SASE. 0.7 seconds in rotation mode. Do you think this is fast enough so that we can now remove the constraint on spectrum, and simply toggle the diffuse feature in whatever spectrum mode is currently set? I think we are ready for that now.

[mewall]

@nksauter I think that would be OK. Maybe it would be good to just make some recommendations to the user about investigating diffuse scattering in some documentation: (1) speed things up by using Monochromatic spectrum shape; (2) increase domain size to 10000 Angstroms to more easily see what happens when changing diffuse scattering parameters.

…

________________________________ From: nksauter ***@***.***> Sent: Tuesday, May 9, 2023 12:28:46 PM To: dermen/sim_erice Cc: Wall, Michael E; Mention Subject: [EXTERNAL] Re: [dermen/sim_erice] Increase mosaic_domains to 100 (Issue #11) @mewall<https://urldefense.com/v3/__https://github.com/mewall__;!!Bt8fGhp8LhKGRg!GK1IfJYTO6jzRyz8MviLNdMKwovXhjq6Fz3FdjuEV0__1T92avt03zgm2PapS0CVI09z763I04OZOwfkM1sT_2ar$> I get 0.4 seconds if set to Bragg+diffuse and Gaussian or SASE. 0.7 seconds in rotation mode. Do you think this is fast enough so that we can now remove the constraint on spectrum, and simply toggle the diffuse feature in whatever spectrum mode is currently set? I think we are ready for that now. — Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/dermen/sim_erice/issues/11*issuecomment-1540674060__;Iw!!Bt8fGhp8LhKGRg!GK1IfJYTO6jzRyz8MviLNdMKwovXhjq6Fz3FdjuEV0__1T92avt03zgm2PapS0CVI09z763I04OZOwfkM3KvvXDQ$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AA67VEJZIJBXMQQZ4HLZ323XFKEF5ANCNFSM6AAAAAAX26FVWA__;!!Bt8fGhp8LhKGRg!GK1IfJYTO6jzRyz8MviLNdMKwovXhjq6Fz3FdjuEV0__1T92avt03zgm2PapS0CVI09z763I04OZOwfkM2KedeWV$>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[mewall]

I would like to open discussion on this issue again. I'm concerned that when the user asks for a diffuse simulation using default parameters it takes an unreasonably long time to get results. With the new parameter for mosaic block size mentioned in issue #25 , for example, it should be possible to advise the user on selection of the diffuse simulation that the mosaic block size is being set to 1, and to allow the user to increase the block size as desired. This change in setting can be rationalized in the documentation as the mosaicity must anyway be sufficiently small to analyze the diffuse features. If this doesn't sound good I'm open to other ways to address the issue.