correct load unfinished atlas expansion accounting for key errors wit…

…h atlas history

src/examples/scratch/outputs/default_atlas_dir/README.md

@@ -0,0 +1,5 @@
+# README
+
+This atlas dir is a good place to develop. A commannd one can use is
+
+python main.py --bibtex_fp data/Imeletal2022a.bib --atlas_dir outputs/default_atlas_dir --target_size 100 --max_pubs_per_expand 20

src/sciterra/mapping/atlas.py

@@ -33,7 +33,7 @@ def __init__(
         publications: list[Publication],
         projection: Projection = None,
         bad_ids: set[str] = set(),
-        history: dict[str, Any] = dict(),
+        history: dict[str, Any] = None,
         center: Publication = None,
     ) -> None:
         if not isinstance(publications, list):
@@ -104,6 +104,8 @@ def save(
                 fp = os.path.join(atlas_dirpath, fn)
                 if os.path.isfile(fp):
                     warnings.warn(f"Overwriting existing file at {fp}.")
+                else:
+                    warnings.warn(f"Writing to {fp}.")
                 write_pickle(fp, attributes[attribute])
             else:
                 warnings.warn(f"No {attribute} to save, skipping.")

src/sciterra/mapping/tracing.py

@@ -55,11 +55,14 @@ def iterate_expand(
 
     # Expansion loop
     failures = 0
-    its = 0
+    # Count previous iterations from loaded atlas as part of total
+    its = len(atl.history['pubs_per_update']) if atl.history is not None else 0
     while not converged:
         its += 1
         len_prev = len(atl)
 
+        print(f"\nExpansion {its}\n-------------------------------")
+
         # Retrieve up to n_pubs_max citations and references.
         atl = crt.expand(
             atl,
@@ -153,7 +156,9 @@ def __init__(
                 f"Loaded atlas has {len(atl)} publications and {len(atl.projection) if atl.projection is not None else 'None'} embeddings.\n"
             )
             # Crucial step: align the history of crt with atl
-            self.cartographer.pubs_per_update = atl.history["pubs_per_update"]
+            if atl.history is not None:
+                self.cartographer.pubs_per_update = atl.history["pubs_per_update"]
+                print(f"Loaded atlas at expansion iteration {len(atl.history['pubs_per_update'])}.")
         else:
             print(f"Initializing atlas.")
 
@@ -162,6 +167,7 @@ def __init__(
 
             # Get center from file
             atl_center = self.cartographer.bibtex_to_atlas(bibtex_fp)
+            atl_center = self.cartographer.project(atl_center)
 
             num_entries = len(atl_center.publications.values())
             if num_entries > 1:

src/tests/test_cartography.py

@@ -157,6 +157,7 @@ def test_dummy_projection(self):
                     "identifier": f"id_{i}",
                     "abstract": "blah blah blah",
                     "publication_date": datetime(2023, 1, 1),
+                    "fields_of_study": ["dummy_field"],
                 }
             )
             for i in range(10)
@@ -180,6 +181,7 @@ def test_dummy_projection_partial(self):
                     "identifier": f"id_{0}",
                     "abstract": "We use cosmological hydrodynamic simulations with stellar feedback from the FIRE (Feedback In Realistic Environments) project to study the physical nature of Lyman limit systems (LLSs) at z ≤ 1.",  # everything here should be in the Word2Vec default vocab, since it trains on this abstract.
                     "publication_date": datetime(2023, 1, 1),
+                    "fields_of_study": ["dummy_field"],
                 }
             ),
             Publication(
@@ -194,6 +196,7 @@ def test_dummy_projection_partial(self):
                     "identifier": f"id_{2}",
                     "abstract": "We use cosmological hydrodynamic simulations with stellar feedback from the FIRE (Feedback In Realistic Environments) project to study the physical nature of Lyman limit systems (LLSs) at z ≤ 1.",
                     "publication_date": datetime(2023, 1, 1),
+                    "fields_of_study": ["dummy_field"],                    
                 }
             ),
         ]