Moving particle routines from analysis members to shared #1468

davelee2804 · 2017-12-11T16:53:51Z

Particle routines previously only available to light analysis members directory, now moved to general location

adding new particles pool to the registry and moving arrays from lagrangian particle pools in analysis members to this new pool
modifying routines in geometric_utils to remove the pointer attribute from incoming mesh pools
calling shift_time_levels on contents of the particels pool, which differ slightly from the previous lagrangian particles pool in analysis members
have not correspondingly updated mpas_particle_list_update_particle_block() for new pool
have added particle helpers module to shared and moved routines from analysis members to here

davelee2804 · 2017-12-11T16:57:05Z

Branched off an earlier version of ocean develop for compatibility of CDG and FCT tracer advection prior to operator split version of FCT being merged in.

…ion routine 2) moving wachspressAreaB array to the ocean/shared/Registry 3) deleting the particles Pool and reverting the contents back to the lagrangian particles pools

pwolfram

Please see some comments and thanks for chatting about this @davelee2804. The key thing is to have distinct velocity namespace between LIGHT and CDG.

pwolfram · 2017-12-13T20:42:49Z

src/core_ocean/analysis_members/Registry_lagrangian_particle_tracking.xml

- />
- <var name="ucReconstructZ" type="real" dimensions="nVertLevels nCells" units="m s^{-1}"
- description="reconstructed cell center velocity- z component"
- />


@davelee2804, I think these have to be left here, correct?

We need them to advect particles.

pwolfram · 2017-12-13T20:43:09Z

src/core_ocean/analysis_members/Registry_lagrangian_particle_tracking.xml

- />
- <var name="ucZ" type="real" dimensions="nVertLevels nCells" units="m s^{-1}"
- description="cell velocity- z component"
- />


Same here, please leave.

pwolfram · 2017-12-13T20:43:39Z

src/core_ocean/analysis_members/Registry_lagrangian_particle_tracking.xml

- />
- <var name="boundaryVertexGlobal" type="integer" dimensions="nVertLevels nVertices" units="unitless"
- description="Mask for determining boundary vertices, but global. A boundary vertex has at least one inactive cell neighboring it."
- />


Same here, please leave.

pwolfram · 2017-12-13T20:44:27Z

src/core_ocean/analysis_members/mpas_ocn_lagrangian_particle_tracking.F

@@ -398,7 +400,7 @@ subroutine ocn_compute_lagrangian_particle_tracking(domain, timeLevel, err)!{{{
 call mpas_timer_start("reconst_filter_LPT")
 #endif
 ! need to handle periodicity within functions below for vertex reconstruction
- call ocn_vertex_reconstruction(filterNum, meshPool, lagrPartTrackScratchPool, lagrPartTrackCellsPool, &


I think we want to leave this line as-is because vertex reconstruction can be filtered.

pwolfram · 2017-12-13T20:44:58Z

src/core_ocean/analysis_members/mpas_ocn_lagrangian_particle_tracking.F

@@ -1327,7 +1331,7 @@ subroutine initalize_fields(domain, err)!{{{
 #ifdef MPAS_DEBUG
 call mpas_timer_start("init_reconst_filter_LPT")
 #endif


Same here, we don't want these changes because we are keeping separate fields for CDG and LIGHT.

pwolfram · 2017-12-13T20:45:45Z

src/core_ocean/analysis_members/mpas_ocn_lagrangian_particle_tracking_interpolations.F

@@ -412,173 +128,6 @@ subroutine ocn_vector_vertex_to_cell_center(meshPool, & !{{{

 end subroutine ocn_vector_vertex_to_cell_center !}}}



Why didn't you bring over this function too?

pwolfram · 2017-12-13T20:49:04Z

src/core_ocean/shared/mpas_ocn_particle_helpers.F

+ ! output variables
+ real (kind=RKIND), dimension(:,:), pointer, intent(out) :: uvReconstructX !< Output: Vertex Reconstructed X Velocity Component
+ real (kind=RKIND), dimension(:,:), pointer, intent(out) :: uvReconstructY !< Output: Vertex Reconstructed Y Velocity Component
+ real (kind=RKIND), dimension(:,:), pointer, intent(out) :: uvReconstructZ !< Output: Vertex Reconstructed Z Velocity Component


I think the intent here needs to be inout to prevent issues for Theta.

pwolfram · 2017-12-13T20:51:00Z

src/core_ocean/shared/mpas_ocn_particle_helpers.F

+
+ end subroutine mpas_particle_horizontal_movement!}}}
+
+ real (kind=RKIND) function old_ocn_spherical_area(ax, ay, az, bx, by, bz, cx, cy, cz, radius)!{{{


Header needed

pwolfram · 2017-12-13T20:53:56Z

src/core_ocean/Registry.xml

+ <var name="ucReconstructZonal" type="real" dimensions="nVertLevels nCells" units="m s^{-1}"
+ description="reconstructed cell center velocity- zonal component"
+ />
+ <var name="boundaryVertexGlobal" type="integer" dimensions="nVertLevels nVertices" units="unitless"


boundaryVertexGlobal can stay here, rest should be duplicated.

Only add fields you need for CDG

Should have your own scratchpool for CDG.

pwolfram · 2017-12-13T21:04:27Z

src/core_ocean/shared/mpas_ocn_particle_helpers.F

+ ! outputs
+ type (field2DReal), pointer, intent(out) :: uvReconstructX !< Output: X Component of velocity reconstructed to vertices
+ type (field2DReal), pointer, intent(out) :: uvReconstructY !< Output: Y Component of velocity reconstructed to vertices
+ type (field2DReal), pointer, intent(out) :: uvReconstructZ !< Output: Z Component of velocity reconstructed to vertices


May need intents inout.

…lagrangian particles scratch pool, and also adding a meshInterpolation package (on if cdg or light is active)

pwolfram · 2017-12-18T19:31:36Z

src/core_ocean/shared/mpas_ocn_particle_helpers.F

+
+ !AB = old_ocn_arc_length(ax,ay,az , bx,by,bz, radius) / radius
+ !BC = old_ocn_arc_length(bx,by,bz , cx,cy,cz, radius) / radius
+ !CA = old_ocn_arc_length(cx,cy,cz , ax,ay,az, radius) / radius


Remove commented lines

David Robert Lee added 2 commits December 8, 2017 10:49

initial commit of the new particle reorg branch

d2bef5a

adding particle helpers module to the makefile

e43d4e7

davelee2804 requested review from pwolfram and mark-petersen December 11, 2017 16:53

davelee2804 added clean-up Ocean labels Dec 11, 2017

1) deleting lagrangian particles version of the wachspress initialzat…

19e5bc3

…ion routine 2) moving wachspressAreaB array to the ocean/shared/Registry 3) deleting the particles Pool and reverting the contents back to the lagrangian particles pools

pwolfram suggested changes Dec 13, 2017

View reviewed changes

additional changes to particle pull request to facilitate removal of …

d320b8e

…lagrangian particles scratch pool, and also adding a meshInterpolation package (on if cdg or light is active)

mark-petersen added the Don't Merge label Dec 14, 2017

removing old version of the particle_horiztonal_movement routine

0413eb1

pwolfram reviewed Dec 18, 2017

View reviewed changes

cleaning up the spherical triangle area routine

6be3c95

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Moving particle routines from analysis members to shared #1468

Moving particle routines from analysis members to shared #1468

davelee2804 commented Dec 11, 2017

davelee2804 commented Dec 11, 2017

pwolfram left a comment

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 13, 2017

pwolfram Dec 18, 2017

		@@ -412,173 +128,6 @@ subroutine ocn_vector_vertex_to_cell_center(meshPool, & !{{{

		end subroutine ocn_vector_vertex_to_cell_center !}}}


		end subroutine mpas_particle_horizontal_movement!}}}

		real (kind=RKIND) function old_ocn_spherical_area(ax, ay, az, bx, by, bz, cx, cy, cz, radius)!{{{

Moving particle routines from analysis members to shared #1468

Are you sure you want to change the base?

Moving particle routines from analysis members to shared #1468

Conversation

davelee2804 commented Dec 11, 2017

davelee2804 commented Dec 11, 2017

pwolfram left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment