From 43d2eff687a6f94a6a8538ff01ca4180c724a406 Mon Sep 17 00:00:00 2001
From: Lionel GUEZ <guez@lmd.ens.fr>
Date: Tue, 21 Dec 2021 19:54:12 +0100
Subject: [PATCH] Do not use a second, 0-based, date index
We are reverting commit e48b7f7e. I think I did not really appreciate
then that the number of days since 1950-1-1 is another date index. It
is clearer to use just one date index, which is in the dbf files. So,
in effect, we replace everywhere the old 0-based date index k by `k -
d_init`. Note that we no longer need argument `d_init` for procedure
`dispatch_snapshot` and we no longer need to write `d_init` to file
`node_id_param.json`. Note also that the value of every node index
changes by the amount `d_init * e_overestim`.
---
Common/read_snapshot.f90 | 13 +++++-----
Inst_eddies/Tests/test_nearby_extr.f90 | 2 +-
Overlap/Tests/Input/NSO_delta_in.txt | 2 +-
.../Input/successive_overlap_periodic_in.txt | 2 +-
Overlap/Tests/short_tests.json | 24 +++++++++----------
Overlap/Tests/test_get_dispatch_snap.f90 | 11 ++++-----
Overlap/Tests/test_overlap.f90 | 7 +++---
Overlap/Tests/test_read_snapshot.f90 | 2 +-
Overlap/Tests/test_send_recv.f90 | 6 ++---
Overlap/dispatch_snapshot.f90 | 13 ++++------
Overlap/eddy_graph.f90 | 17 +++++++------
Overlap/overlap.f90 | 2 +-
Overlap/report_graph.py | 22 +++++++----------
13 files changed, 56 insertions(+), 67 deletions(-)
diff --git a/Common/read_snapshot.f90 b/Common/read_snapshot.f90
index 9069e4b1..3239e8b0 100644
--- a/Common/read_snapshot.f90
+++ b/Common/read_snapshot.f90
@@ -41,24 +41,23 @@ contains
!---------------------------------------------------------------------
- if (k == 0) then
+ if (k == d_init) then
ishape_first = 0
else
- ! {k > 0}
- ishape_first = ishape_last(k) + 1
+ ! {k > d_init}
+ ishape_first = ishape_last(k - d_init) + 1
end if
- s%number_extr = ishape_last(k + 1) - ishape_first + 1
+ s%number_extr = ishape_last(k - d_init + 1) - ishape_first + 1
allocate(s%list(s%number_extr))
- do ishape = ishape_first, ishape_last(k + 1)
+ do ishape = ishape_first, ishape_last(k - d_init + 1)
call read_eddy(e, date, eddy_i, hshp, ishape)
! Check that all the eddies have the same date index:
- if (date - d_init /= k) then
+ if (date /= k) then
print *, "read_snapshot: bad date index"
print *, "date = ", date
- print *, "d_init = ", d_init
print *, "k = ", k
print *, "ishape = ", ishape
print *, "eddy_i = ", eddy_i
diff --git a/Inst_eddies/Tests/test_nearby_extr.f90 b/Inst_eddies/Tests/test_nearby_extr.f90
index f2d55824..280415c3 100644
--- a/Inst_eddies/Tests/test_nearby_extr.f90
+++ b/Inst_eddies/Tests/test_nearby_extr.f90
@@ -50,7 +50,7 @@ program test_nearby_extr
call shpc_open(hshp, trim(shpc_dir), rank = 0, pszaccess = "rb")
call dbf_read_attribute_03(d_init, hshp%extremum, hshp%extr_date, ishape = 0)
- call read_snapshot(s, hshp, nlon, nlat, d_init, k = 0, &
+ call read_snapshot(s, hshp, nlon, nlat, d_init, k = d_init, &
corner = corner_deg * deg_to_rad, step = step_deg * deg_to_rad, &
copy = 0, ishape_last = [ishape_last])
CALL shpc_close(hshp)
diff --git a/Overlap/Tests/Input/NSO_delta_in.txt b/Overlap/Tests/Input/NSO_delta_in.txt
index d2089bdd..3a6620d9 100644
--- a/Overlap/Tests/Input/NSO_delta_in.txt
+++ b/Overlap/Tests/Input/NSO_delta_in.txt
@@ -1,3 +1,3 @@
- &MAIN_NML MAX_DELTA = 2 /
+ &MAIN_NML MAX_DELTA = 2, k_test_1 = 20454, k_test_2 = 20455 /
,,,,,,1
,,,,,,1
diff --git a/Overlap/Tests/Input/successive_overlap_periodic_in.txt b/Overlap/Tests/Input/successive_overlap_periodic_in.txt
index d04199ea..8fa2295e 100644
--- a/Overlap/Tests/Input/successive_overlap_periodic_in.txt
+++ b/Overlap/Tests/Input/successive_overlap_periodic_in.txt
@@ -1,3 +1,3 @@
- &MAIN_NML DIST_LIM = 4,k_test_2 = 0 /
+ &MAIN_NML DIST_LIM = 4,k_test_1 = 20454, k_test_2 = 20454 /
/
/
diff --git a/Overlap/Tests/short_tests.json b/Overlap/Tests/short_tests.json
index 023d6ccb..b52e2e5f 100644
--- a/Overlap/Tests/short_tests.json
+++ b/Overlap/Tests/short_tests.json
@@ -47,7 +47,7 @@
]
},
{
- "input": "&MAIN_NML k_test_1 = 1 /\n/\n/\n",
+ "input": "&MAIN_NML k_test_1 = 20455, k_test_2 = 20455 /\n/\n/\n",
"description": "Overlap of a snapshot with itself.",
"title": "Successive_overlap",
"command":
@@ -57,7 +57,7 @@
]
},
{
- "input": "&MAIN_NML k_test_1 = 1 /\n/\n/\n",
+ "input": "&MAIN_NML k_test_1 = 20455, k_test_2 = 20455 /\n/\n/\n",
"description": "Same as Successive_overlap, but with larger domain: region 5 instead of region 4.",
"title": "Successive_overlap_region_5_one_date",
"command":
@@ -77,7 +77,7 @@
]
},
{
- "input": "&MAIN_NML /\n/\n/\n",
+ "input": "&MAIN_NML k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description": "Overlap of different snapshots.",
"title": "Successive_overlap_different_snapshots",
"command":
@@ -87,7 +87,7 @@
]
},
{
- "input": "&MAIN_NML /\n/\n/\n",
+ "input": "&MAIN_NML k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description":
"Overlap of different snapshots. Same as Successive_overlap_different_snapshots, but with a larger region. The identifying numbers of the connected eddies are not the same for all edges.",
"title": "Successive_overlap_region_2",
@@ -98,7 +98,7 @@
]
},
{
- "input": "&MAIN_NML /\n/\n/\n",
+ "input": "&MAIN_NML k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description": "Same as Successive_overlap_region_2, but with a larger region. We get some merging and splitting.",
"title": "Successive_overlap_region_5",
"command":
@@ -108,7 +108,7 @@
]
},
{
- "input": "&MAIN_NML /\n/\n/\n",
+ "input": "&MAIN_NML k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description": "Global grid, normal 0.25° resolution.",
"title": "Successive_overlap_global",
"command":
@@ -118,7 +118,7 @@
]
},
{
- "input": "&MAIN_NML max_delta = 3, k_test_2 = 0 /\n/\n/\n",
+ "input": "&MAIN_NML max_delta = 3, k_test_1 = 20454, k_test_2 = 20454 /\n/\n/\n",
"description": "Overlap of a snapshot with itself. Same as Successive_overlap, except for max_delta.",
"title": "Non_successive_overlap",
"command":
@@ -128,7 +128,7 @@
]
},
{
- "input": "&MAIN_NML max_delta = 3 /\n/\n/\n",
+ "input": "&MAIN_NML max_delta = 3, k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description":
"Overlap of different snapshots. Same as Successive_overlap_different_snapshots, except for max_delta.",
"title": "NSO_different_snapshots",
@@ -139,7 +139,7 @@
]
},
{
- "input": "&MAIN_NML max_delta = 3 /\n/\n/\n",
+ "input": "&MAIN_NML max_delta = 3, k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description":
"Same as Successive_overlap_region_5, except for max_delta.",
"title": "NSO_region_5",
@@ -150,7 +150,7 @@
]
},
{
- "input": "&MAIN_NML max_delta = 3 /\n/\n/\n",
+ "input": "&MAIN_NML max_delta = 3, k_test_1 = 20454, k_test_2 = 20455 /\n/\n/\n",
"description":
"Same as Successive_overlap_global, except for max_delta.",
"title": "NSO_global",
@@ -191,7 +191,7 @@
},
{
"input":
- "&MAIN_NML CORNER = 5.125, -36.375, NLON = 29, NLAT = 17 /\n",
+ "&MAIN_NML CORNER = 5.125, -36.375, NLON = 29, NLAT = 17, k = 20454 /\n",
"title": "Get_dispatch_snap",
"stdout": "test_get_dispatch_snap_stdout.txt",
"commands": [
@@ -209,7 +209,7 @@
]
},
{
- "input": "&MAIN_NML CORNER = 5.125, -36.375, NLON = 29, NLAT = 17, k = 2 /\n",
+ "input": "&MAIN_NML CORNER = 5.125, -36.375, NLON = 29, NLAT = 17, k = 20456 /\n",
"description": "Receive case for one of the processes in procedure get_snapshot.",
"title": "Get_dispatch_snap_recv",
"stdout": "test_get_dispatch_snap_stdout.txt",
diff --git a/Overlap/Tests/test_get_dispatch_snap.f90 b/Overlap/Tests/test_get_dispatch_snap.f90
index 656f32cc..3a0a1c3c 100644
--- a/Overlap/Tests/test_get_dispatch_snap.f90
+++ b/Overlap/Tests/test_get_dispatch_snap.f90
@@ -96,23 +96,22 @@ program test_get_dispatch_snap
if (rank == 0) call dbf_read_attribute_03(d_init, hshp%extremum, &
hshp%extr_date, ishape = 0)
call ezmpi_bcast(d_init, root = 0)
- k_begin = (rank * n_dates) / n_proc
+ k_begin = d_init + (rank * n_dates) / n_proc
if (rank < n_proc - 1) then
- k_end = ((rank + 1) * n_dates) / n_proc
+ k_end = d_init + ((rank + 1) * n_dates) / n_proc
else
- k_end = n_dates - 1
+ k_end = d_init + n_dates - 1
end if
call get_snapshot(s, nlon, nlat, ishape_last, corner * deg_to_rad, &
step * deg_to_rad, copy, hshp, d_init, k, k_end, rank, n_proc, &
max_delta = 1)
CALL shpc_close(hshp)
- call dispatch_snapshot(s, unit_isolated, rank, k_begin, max_delta = 1, &
- d_init = d_init, k = k)
+ call dispatch_snapshot(s, unit_isolated, rank, k_begin, max_delta = 1, k = k)
if (rank == 0) then
- call write_snapshot(s, corner, step, nlon, nlat, copy, d = d_init + k)
+ call write_snapshot(s, corner, step, nlon, nlat, copy, d = k)
close(unit_isolated)
end if
diff --git a/Overlap/Tests/test_overlap.f90 b/Overlap/Tests/test_overlap.f90
index 4a6e2810..d48ff4f7 100644
--- a/Overlap/Tests/test_overlap.f90
+++ b/Overlap/Tests/test_overlap.f90
@@ -68,15 +68,14 @@ program test_overlap
allocate(flow(max_delta + 1))
call read_column(ishape_last, file = trim(shpc_dir) // "/ishape_last.txt")
n_dates = size(ishape_last)
- call assert(0 <= [k_test_1, k_test_2] .and. [k_test_1, k_test_2] < n_dates, &
- "test_overlap k_test_1, k_test_2")
+ call assert(d_init <= [k_test_1, k_test_2] .and. [k_test_1, k_test_2] &
+ < d_init + n_dates, "test_overlap k_test_1, k_test_2")
e_overestim = maxval([ishape_last(1) + 1, ediff1d(ishape_last)])
call shpc_open(hshp, trim(shpc_dir), rank = 0, pszaccess = "rb")
call dbf_read_attribute_03(d_init, hshp%extremum, hshp%extr_date, ishape = 0)
open(unit, file = "node_id_param.json", status = "replace", &
action = "write")
- write(unit, fmt = *) '{"e_overestim": ', e_overestim, ', "d_init": ', &
- d_init, '}'
+ write(unit, fmt = *) '{"e_overestim": ', e_overestim, '}'
close(unit)
call read_snapshot(flow(1), hshp, nlon, nlat, d_init, k_test_1, corner, &
step, copy, ishape_last)
diff --git a/Overlap/Tests/test_read_snapshot.f90 b/Overlap/Tests/test_read_snapshot.f90
index efc9ef12..0dc040a2 100644
--- a/Overlap/Tests/test_read_snapshot.f90
+++ b/Overlap/Tests/test_read_snapshot.f90
@@ -68,7 +68,7 @@ program test_read_snapshot
call shpc_open(hshp, trim(shpc_dir), rank = 0, pszaccess = "rb")
call dbf_read_attribute_03(d_init, hshp%extremum, hshp%extr_date, ishape = 0)
- call read_snapshot(s, hshp, nlon, nlat, d_init, k = 0, &
+ call read_snapshot(s, hshp, nlon, nlat, d_init, k = d_init, &
corner = corner_deg * deg_to_rad, step = step_deg * deg_to_rad, &
copy = copy, ishape_last = [ishape_last])
CALL shpc_close(hshp)
diff --git a/Overlap/Tests/test_send_recv.f90 b/Overlap/Tests/test_send_recv.f90
index ec179173..658e63f2 100644
--- a/Overlap/Tests/test_send_recv.f90
+++ b/Overlap/Tests/test_send_recv.f90
@@ -102,19 +102,19 @@ program test_send_recv
if (rank == 0) call dbf_read_attribute_03(d_init, hshp%extremum, &
hshp%extr_date, ishape = 0)
call ezmpi_bcast(d_init, root = 0)
- if (rank == 1) call read_snapshot(s, hshp, nlon, nlat, d_init, k = 0, &
+ if (rank == 1) call read_snapshot(s, hshp, nlon, nlat, d_init, k = d_init, &
corner = corner_deg * deg_to_rad, step = step_deg * deg_to_rad, &
copy = copy, ishape_last = ishape_last)
CALL shpc_close(hshp)
if (rank == 1) then
- call send_snapshot(s, dest = 0, tag = 0)
+ call send_snapshot(s, dest = 0, tag = d_init)
else
! rank == 0
call MPI_Comm_get_attr(MPI_Comm_world, MPI_TAG_UB, attribute_val, flag)
call assert(flag, "test_send_recv MPI_Comm_get_attr MPI_TAG_UB")
print *, "MPI_TAG_UB = ", attribute_val
- call recv_snapshot(s, nlon, nlat, copy, source = 1, tag = 0)
+ call recv_snapshot(s, nlon, nlat, copy, source = 1, tag = d_init)
call write_snapshot(s, corner_deg, step_deg, nlon, nlat, copy, d_init)
end if
diff --git a/Overlap/dispatch_snapshot.f90 b/Overlap/dispatch_snapshot.f90
index f3662094..daeb1449 100644
--- a/Overlap/dispatch_snapshot.f90
+++ b/Overlap/dispatch_snapshot.f90
@@ -4,8 +4,7 @@ module dispatch_snapshot_m
contains
- subroutine dispatch_snapshot(s, unit_isolated, rank, k_begin, max_delta, &
- d_init, k)
+ subroutine dispatch_snapshot(s, unit_isolated, rank, k_begin, max_delta, k)
use derived_types, only: snapshot
use send_snapshot_m, only: send_snapshot
@@ -15,21 +14,19 @@ contains
integer, intent(in):: unit_isolated
! logical unit for file isolated_nodes_$rank.csv
- integer, intent(in):: rank, k_begin, max_delta, d_init
- integer, intent(in):: k ! date index (0-based)
+ integer, intent(in):: rank, k_begin, max_delta
+ integer, intent(in):: k ! date index
! Local:
- integer i, d
+ integer i
!------------------------------------------------------------------
if (rank == 0 .or. k >= k_begin + max_delta) then
- d = k + d_init
-
do i = 1, s%number_extr
if (s%list(i)%valid .and. s%list(i)%delta_in == huge(0) &
.and. s%list(i)%delta_out == huge(0)) &
- write(unit_isolated, fmt = *) d, i
+ write(unit_isolated, fmt = *) k, i
end do
else
call send_snapshot(s, tag = k, dest = rank - 1)
diff --git a/Overlap/eddy_graph.f90 b/Overlap/eddy_graph.f90
index 028b3751..e75db8ac 100644
--- a/Overlap/eddy_graph.f90
+++ b/Overlap/eddy_graph.f90
@@ -22,7 +22,7 @@ program eddy_graph
integer rank, n_proc, copy, delta, j
integer:: n_dates = huge(0) ! number of dates to read
- integer k ! date index ((0-based)
+ integer k ! date index
integer d_init, k_begin, k_end, k_end_main_loop
character(len = :), allocatable:: shpc_dir
integer:: max_delta = 1
@@ -124,8 +124,7 @@ program eddy_graph
if (rank == 0) then
open(unit, file = "node_id_param.json", status = "replace", &
action = "write")
- write(unit, fmt = *) '{"e_overestim": ', e_overestim, ', "d_init": ', &
- d_init, '}'
+ write(unit, fmt = *) '{"e_overestim": ', e_overestim, '}'
close(unit)
end if
@@ -138,13 +137,13 @@ program eddy_graph
call open_edge_file(rank)
! We do not write the title line. That will be handled by eddy_graph.sh.
- k_begin = (rank * n_dates) / n_proc
+ k_begin = d_init + (rank * n_dates) / n_proc
if (rank < n_proc - 1) then
- k_end = ((rank + 1) * n_dates) / n_proc + max_delta - 1
+ k_end = d_init + ((rank + 1) * n_dates) / n_proc + max_delta - 1
k_end_main_loop = k_end - max_delta + 1
else
- k_end = n_dates - 1
+ k_end = d_init + n_dates - 1
k_end_main_loop = k_end
end if
@@ -168,7 +167,7 @@ program eddy_graph
do k = k_begin + max_delta + 1, k_end_main_loop
call dispatch_snapshot(flow(1), unit_isolated, rank, k_begin, max_delta, &
- d_init, k = k - max_delta - 1)
+ k = k - max_delta - 1)
flow(:max_delta) = flow(2:)
call get_snapshot(flow(max_delta + 1), nlon, nlat, ishape_last, corner, &
step, copy, hshp, d_init, k, k_end, rank, n_proc, max_delta)
@@ -184,7 +183,7 @@ program eddy_graph
do k = k_end_main_loop + 1, k_end
! {rank < n_proc - 1 and k >= k_end - max_delta + 2}
call dispatch_snapshot(flow(1), unit_isolated, rank, k_begin, max_delta, &
- d_init, k = k - max_delta - 1)
+ k = k - max_delta - 1)
flow(:max_delta) = flow(2:)
call get_snapshot(flow(max_delta + 1), nlon, nlat, ishape_last, corner, &
step, copy, hshp, d_init, k, k_end, rank, n_proc, max_delta)
@@ -199,7 +198,7 @@ program eddy_graph
do j = 1, max_delta + 1
call dispatch_snapshot(flow(j), unit_isolated, rank, k_begin, max_delta, &
- d_init, k = k_end - max_delta - 1 + j)
+ k = k_end - max_delta - 1 + j)
end do
! 5. Back matter
diff --git a/Overlap/overlap.f90 b/Overlap/overlap.f90
index b0579f6d..af0658b7 100644
--- a/Overlap/overlap.f90
+++ b/Overlap/overlap.f90
@@ -35,7 +35,7 @@ contains
integer, intent(in):: e_overestim
! over-estimation of the number of eddies at each date
- integer, intent(in):: k ! date index (0-based)
+ integer, intent(in):: k ! date index
integer, intent(in):: delta ! between 1 and max_delta
integer, intent(in):: j
diff --git a/Overlap/report_graph.py b/Overlap/report_graph.py
index 4003d52a..7d7a1927 100755
--- a/Overlap/report_graph.py
+++ b/Overlap/report_graph.py
@@ -12,22 +12,20 @@ from networkx import nx_agraph
import os
import json
-def node_to_date_eddy(node_index, e_overestim, d_init):
+def node_to_date_eddy(node_index, e_overestim):
"""Convert from node identification in graph to eddy identification in
shapefiles.
"""
k = (node_index - 1) // e_overestim
- date = d_init + k
- return {"eddy_index": node_index - k * e_overestim, "date_index": k,
- "date": date}
+ return {"eddy_index": node_index - k * e_overestim, "date_index": k}
-def date_eddy_to_node(date, eddy_index, e_overestim, d_init):
+def date_eddy_to_node(date, eddy_index, e_overestim):
"""Convert from eddy identification in shapefiles to node
identification in graph.
"""
- return (date - d_init) * e_overestim + eddy_index
+ return date * e_overestim + eddy_index
def read_eddy_graph(edgelist, shpc_dir = None):
if os.access(edgelist, os.R_OK):
@@ -37,7 +35,6 @@ def read_eddy_graph(edgelist, shpc_dir = None):
node_id_param_file = path.join(dir_edgelist, "node_id_param.json")
with open(node_id_param_file) as f: node_id_param = json.load(f)
G.graph["e_overestim"] = node_id_param["e_overestim"]
- G.graph["d_init"] = node_id_param["d_init"]
if shpc_dir is not None:
# Read and set attributes of eddies:
@@ -54,7 +51,7 @@ def set_attribute(G, extr_file):
for shape_rec in s_read:
n = date_eddy_to_node(shape_rec.record.date,
shape_rec.record.eddy_index,
- G.graph["e_overestim"], G.graph["d_init"])
+ G.graph["e_overestim"])
if shape_rec.record.valid == 1 and n in G:
G.add_node(n, coordinates = tuple(shape_rec.shape.points[0]),
@@ -65,13 +62,12 @@ def to_eddy_agraph(G):
my_subgraphs = {}
for n in G:
- date_eddy = node_to_date_eddy(n, G.graph["e_overestim"],
- G.graph["d_init"])
+ date_eddy = node_to_date_eddy(n, G.graph["e_overestim"])
- if date_eddy["date"] not in my_subgraphs:
- my_subgraphs[date_eddy["date"]] = []
+ if date_eddy["date_index"] not in my_subgraphs:
+ my_subgraphs[date_eddy["date_index"]] = []
- my_subgraphs[date_eddy["date"]].append(n)
+ my_subgraphs[date_eddy["date_index"]].append(n)
A = nx_agraph.to_agraph(G)
--
GitLab