more on rupture_grouping

CHANGELOG.md

@@ -1,10 +1,24 @@
 # Changelog
 
-## [0.13.0] 2024-07-25
+## [0.13.0] 2024-09
 
-## Added
- - helper module to assist with participation rates
+## Changed
+ - drop_zero_rate differentiates between FSS and IS;
+ - faster testing (~3 times faster)
+ - documentation improvements
+ - abc427e fixing many deprecation warnings; updated geopandas, pandas, numpy, pyvista libs;
+ - read_csv dtype configuration improvements;
+ - many old functions are deprecated/migrated to new filter package
+ - refactor dataframe dtypes
+ - standardise participation functions API and return columns;
 
+## Added
+ - new filter package providing classes for filtering solotions
+ - support for 3d geometry (thanks @voj)
+ - c3ab6da simplify FSS participation using rate_weighted_mean
+ - participation performance testing;
+ - added participation methods to fault_system_solution
+ - a simple rupture grouping algorithm (can this be a different type of filter??);
 
 ## [0.12.3] 2024-07-04
 bump version to verify new pypi workflow

solvis/fault_system_solution_helper.py

@@ -1,4 +1,4 @@
-from typing import Dict, Iterator, List
+from typing import Dict, Iterator, List, Iterable
 
 from solvis.inversion_solution.typing import InversionSolutionProtocol
 
@@ -12,17 +12,20 @@
 """
 
 
-def build_rupture_groups(solution: InversionSolutionProtocol) -> Iterator[Dict]:
+def build_rupture_groups(
+    solution: InversionSolutionProtocol, rupture_ids: Iterable[int] = None, min_overlap: float = 0.8
+) -> Iterator[Dict]:
     dfrs = solution.rupture_sections
-    ruptures = dfrs['rupture'].unique().tolist()
-    print(f"there are {len(ruptures)} unique ruptures")
+    rupture_ids = rupture_ids or dfrs['rupture'].unique().tolist()
+    print(f"there are {len(rupture_ids)} unique ruptures")
     count = 0
     sample_sections = None
     sample_rupt = None
     sample_ruptures: List[int] = []
 
-    for rupt_id in ruptures:
+    for rupt_id in rupture_ids:
         sections = dfrs[dfrs.rupture == rupt_id]['section'].tolist()
+
         # first or reset
         if sample_rupt is None:
             sample_ruptures = []
@@ -34,19 +37,19 @@ def build_rupture_groups(solution: InversionSolutionProtocol) -> Iterator[Dict]:
         if rupt_id is not sample_rupt:
             sample_ruptures.append(rupt_id)
 
-        # otherwise compare overlap
-        diff = len(set(sections).symmetric_difference(sample_sections))
-        overlap = len(set(sections).intersection(sample_sections))
-
-        if overlap:  # and there must be some non-zero score
-            score = (sample_len - diff) / sample_len
+        # compare section overlap
+        section_overlap = len(set(sections).intersection(sample_sections))
 
-        # print(f'rupt_id {rupt_id} score {score} overlap {overlap} sample_len: {sample_len} diff {diff}')
-        if (score < 0.7) or not overlap:  # (overlap < 0.8 * sample_len):
-            yield {'rupture': sample_rupt, 'ruptures': sample_ruptures, 'sample_sections': sample_len}
+        if section_overlap:  # and there must be some non-zero score
+            section_diff_count = len(set(sections).symmetric_difference(sample_sections))
+            score = (sample_len - section_diff_count) / sample_len
+            if score >= min_overlap:
+                continue
+            else:
+                count += 1
+                yield {'rupture': sample_rupt, 'ruptures': sample_ruptures, 'sample_sections': sample_len}
 
-            # signal reset
-            sample_rupt = None
-            count += 1
+        # signal reset
+        sample_rupt = None
 
-    print(f"built {count} rupture groups")
+    print(f"built {count} rupture groups from {len(rupture_ids)} unique ruptures.")

solvis/solvis.py

@@ -3,7 +3,7 @@
 The original solvis API helper functions are defined in this module.
 
 NB please be aware that most functions in this module are deprecated and replaced
-in the 2nd generation Solvis API.
+in the 2nd generation Solvis API (version == 0.13.0)
 """
 # from functools import partial
 import warnings

test/filter/test_fault_system_solution_helper.py

@@ -1,3 +1,7 @@
+import pytest
+
+
+@pytest.mark.skip("This is not in doing much yet")
 def test_explore_zero_rates(crustal_solution_fixture):
     solution = crustal_solution_fixture
 

test/test_rupture_representatives.py

@@ -1,12 +1,54 @@
 import pytest
 
 from solvis.fault_system_solution_helper import build_rupture_groups
+from solvis.filter.rupture_id_filter import FilterRuptureIds
 
 
-@pytest.mark.skip("THis is not in use yet")
+@pytest.mark.slow
 def test_build_rupture_groups(composite_fixture):
     fss = composite_fixture._solutions['CRU']
 
-    for rep in build_rupture_groups(fss):
-        print(rep)
-    # assert 0
+    reps = list(build_rupture_groups(fss, min_overlap=0.5))
+    assert len(reps) == 406
+    assert reps[0] == {'rupture': 0, 'ruptures': [1, 2], 'sample_sections': 2}
+
+    reps = list(build_rupture_groups(fss, min_overlap=0.8))
+    assert len(reps) == 658
+    assert reps[0] == {'rupture': 0, 'ruptures': [1], 'sample_sections': 2}
+
+
+@pytest.mark.parametrize(
+    "min_overlap, expected_len, sample_zero",
+    [
+        (0.5, 2, {'rupture': 0, 'ruptures': [1, 2], 'sample_sections': 2}),
+        (0.8, 4, {'rupture': 0, 'ruptures': [1], 'sample_sections': 2}),
+        (0.99, 5, {'rupture': 0, 'ruptures': [1], 'sample_sections': 2}),
+    ],
+)
+def test_build_rupture_groups_fast(small_composite_fixture, min_overlap, expected_len, sample_zero):
+    fss = small_composite_fixture._solutions['CRU']
+
+    reps = list(build_rupture_groups(fss, min_overlap=min_overlap))
+    assert len(reps) == expected_len
+    assert reps[0] == sample_zero
+
+
+@pytest.mark.parametrize(
+    "min_overlap, expected_len, sample_zero",
+    [
+        (0.5, 1, {'rupture': 5, 'ruptures': [6, 7, 8, 9], 'sample_sections': 7}),
+        (0.8, 1, {'rupture': 5, 'ruptures': [6, 7], 'sample_sections': 7}),
+        (0.99, 2, {'rupture': 5, 'ruptures': [6], 'sample_sections': 7}),
+    ],
+)
+def test_build_rupture_groups_filtered(small_composite_fixture, min_overlap, expected_len, sample_zero):
+    fss = small_composite_fixture._solutions['CRU']
+
+    m7less = list(FilterRuptureIds(fss).for_magnitude(max_mag=7.5))
+
+    reps = list(build_rupture_groups(fss, rupture_ids=m7less, min_overlap=min_overlap))
+
+    print(reps)
+
+    assert len(reps) == expected_len
+    assert reps[0] == sample_zero