lca-algebraic 1.4.1.dev1__tar.gz → 1.4.1.dev3__tar.gz

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lca_algebraic
-Version: 1.4.1.dev1
+Version: 1.4.1.dev3
 Summary: This library provides a layer above Brightway for defining parametric models and running super fast LCA for monte carlo analysis.
 Author-email: OIE - Mines ParisTech <raphael.jolivet@mines-paristech.fr>
 License: BSD

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/lca_algebraic/lambda_expression.py

@@ -1,6 +1,6 @@
 from copy import copy
 from dataclasses import dataclass
-from typing import Dict, List
+from typing import Dict, List, Optional
 
 from sympy import Basic, Expr, lambdify, parse_expr
 from sympy.printing.numpy import NumPyPrinter
@@ -8,6 +8,7 @@ from sympy.printing.numpy import NumPyPrinter
 from lca_algebraic.axis_dict import AxisDict
 from lca_algebraic.base_utils import _user_functions
 from lca_algebraic.bw_wrapper import Activity
+from lca_algebraic.log import warn
 from lca_algebraic.params import (
     _complete_params,
     _expand_param_names,
@@ -39,7 +40,7 @@ class LambdaExpr:
 
         # List of background activities, indexed in the same order as impacts[i]
         self.background_activities: List[Activity] = list()
-        self.impacts: List[float] = None
+        self.impacts: Optional[List[float]] = None
 
         if isinstance(expr, dict) and not isinstance(expr, AxisDict):
             # Come from JSON serialization
@@ -123,12 +124,17 @@ class LambdaExpr:
     def with_impacts(self, impacts: Dict[Activity, float]):
         """Transforms a generic LambaExpr toa one with custom impact values"""
         if self.impacts is not None:
-            raise Exception("You are specializing a LambaExpression with impacts already set.")
+            warn("You are specializing a LambaExpression with impacts already set.")
+
         res = copy(self)
         res.impacts = [impacts[bg_act] for bg_act in self.background_activities]
 
         return res
 
+    def impacts_dict(self):
+        """Returns list of impacts as dict"""
+        return {bg_act: val for bg_act, val in zip(self.background_activities, self.impacts)}
+
 
 def _filter_param_values(params, expanded_param_names):
     return {key: val for key, val in params.items() if key in expanded_param_names}

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/lca_algebraic/lca.py

@@ -34,7 +34,7 @@ from lca_algebraic.bw_wrapper import (
     multi_lca,
 )
 from lca_algebraic.lambda_expression import LambdaExpr
-from lca_algebraic.settings import Settings
+from lca_algebraic.settings import UNITY_METHOD, Settings
 
 from . import atomic, copyActivity, getActByCode, resetDb
 from .cache import (
@@ -162,15 +162,29 @@ def _multiLCAWithProxies(acts: list[ActivityExtended], methods: list[MethodKey])
     return _multiLCA([{act: 1} for act in proxy_acts.values()], methods)
 
 
-def _multiLCAWithCache(all_acts, methods) -> Dict[Tuple[ActivityExtended, MethodKey], float]:
+def _multiLCAWithCache(all_acts, methods, scenario: str = None) -> Dict[Tuple[ActivityExtended, MethodKey], float]:
     res = dict()
 
+    # Possibly map bg activities to another target scenario db
+    # key = initial cat, value = target scenario act
+    scenario_mapping = _compute_scenario_mapping(all_acts, scenario=scenario)
+
+    all_acts = list(scenario_mapping.values())
+    reverse_scenario_mapping = {val: key for key, val in scenario_mapping.items()}
+
     # Split activities by db_name
     for db_name, acts in _group_acts_by_db(all_acts).items():
         with LCIACache(db_name) as cache:
             # List activities with at least one missing value
             remaining_acts = list(act for act in all_acts if any(method for method in methods if (act, method) not in cache.data))
 
+            # Handle custom "unity" method separately
+            if UNITY_METHOD in methods:
+                for act in acts:
+                    for method in methods:
+                        if not (act, method) in cache.data:
+                            cache.data[(act, method)] = 1.0
+
             # list methods with at least one missing value
             remaining_methods = list(
                 method for method in methods if any(act for act in all_acts if (act, method) not in cache.data)
@@ -190,7 +204,7 @@ def _multiLCAWithCache(all_acts, methods) -> Dict[Tuple[ActivityExtended, Method
                         cache.data[(act, method)] = lca.iloc[imethod, iact]
 
             # Update res with a copy of the cache for selected impacts and activities
-            res.update({(act, method): cache.data[(act, method)] for act in acts for method in methods})
+            res.update({(reverse_scenario_mapping[act], method): cache.data[(act, method)] for act in acts for method in methods})
 
     return res
 
@@ -253,7 +267,7 @@ def lambdify_expr(expr):
 
 
 def _preMultiLCAAlgebric(
-    model: ActivityExtended, methods: List[MethodKey], alpha: ValueOrExpression = 1, axis: str = None, scenario: list[str] = None
+    model: ActivityExtended, methods: List[MethodKey], alpha: ValueOrExpression = 1, axis: str = None, scenarios: list[str] = None
 ) -> list[LambdaExpr]:
     """
     This method transforms an activity into a set of functions ready to compute LCA very fast on a set on methods.
@@ -262,26 +276,34 @@ def _preMultiLCAAlgebric(
     This method is used by multiLCAAlgebric
     """
 
-    with DbContext(model):
-        if isinstance(alpha, Quantity):
-            alpha = alpha.magnitude
+    if not isinstance(scenarios, list):
+        scenarios = [scenarios]
+
+    res: list[list[LambdaExpr]] = []
 
-        def _key(method):
-            key = (model, axis, method, alpha)
-            if scenario:
-                key += (scenario,)
-            return key
+    for scenario in scenarios:
+        with DbContext(model):
+            if isinstance(alpha, Quantity):
+                alpha = alpha.magnitude
 
-        with ExprCache(model["database"]) as cache:
-            missing_methods = [method for method in methods if not _key(method) in cache.data]
-            if len(missing_methods) > 0:
-                exprs = _modelToExpr(model, methods=missing_methods, axis=axis, alpha=alpha, scenario=scenario)
-                for method, expr in zip(missing_methods, exprs):
-                    cache.data[_key(method)] = expr
+            def _key(method):
+                key = (model, axis, method, alpha)
+                if scenario:
+                    key += (scenario,)
+                return key
 
-            # At this point, everything is in cache
-            # REturn the list in order
-            return list(cache.data[_key(method)] for method in methods)
+            with ExprCache(model["database"]) as cache:
+                missing_methods = [method for method in methods if not _key(method) in cache.data]
+                if len(missing_methods) > 0:
+                    exprs = _modelToExpr(model, methods=missing_methods, axis=axis, alpha=alpha, scenario=scenario)
+                    for method, expr in zip(missing_methods, exprs):
+                        cache.data[_key(method)] = expr
+
+                # At this point, everything is in cache
+                # REturn the list in order
+                res.append(list(cache.data[_key(method)] for method in methods))
+
+    return _merge_lambda_scenarios(res)
 
 
 def _build_mapping_key(act: Activity):
@@ -444,17 +466,13 @@ def _modelToExpr(
         zero.impacts = []
         return [zero] * len(methods)
 
-    # Possibly map bg activities to another target scenario db
-    act_mapping = _compute_scenario_mapping(generic_lambda_expr.background_activities, scenario=scenario)
-
     # Compute LCA for background activities
-    all_impacts = _multiLCAWithCache(act_mapping.values(), methods)
+    all_impacts = _multiLCAWithCache(all_acts=generic_lambda_expr.background_activities, methods=methods, scenario=scenario)
 
     res = list()
 
     for method in methods:
-        impacts = {bg_act: all_impacts[target_act, method] for bg_act, target_act in act_mapping.items()}
-
+        impacts = {bg_act: all_impacts[bg_act, method] for bg_act in generic_lambda_expr.background_activities}
         res.append(generic_lambda_expr.with_impacts(impacts))
 
     return res
@@ -487,8 +505,12 @@ def _postMultiLCAAlgebric(methods, lambdas: List[LambdaExpr], with_params=False,
 
     if len(lambdas[0].impacts) > 0:
         first_impact = lambdas[0].impacts[0]
+
+        # Vector impacts means several scenarios : should be same number as scenarios
         if isinstance(first_impact, (list, np.ndarray)):
             nb_scenarios = len(first_impact)
+            if nb_scenarios > 1 and param_length > 1 and nb_scenarios != param_length:
+                raise Exception("Number of scenarios and number of params should be the same")
             param_length = max(param_length, nb_scenarios)
 
     # lambda are SymDict ?
@@ -635,6 +657,8 @@ def compute_inventory(
     as_dict=False,
     impact_method=None,
     fields=["database", "name", "location", "unit"],
+    scenario: Union[str, list[str], None] = None,
+    filter_zeros=True,
     **params,
 ):
     """
@@ -667,41 +691,55 @@ def compute_inventory(
     Dataframe or Dict of act => value
 
     """
+    if not isinstance(scenario, list):
+        scenario = [scenario]
+
+    if impact_method is None:
+        impact_method = UNITY_METHOD
+
+    zeros = np.full(len(scenario), 0.0)
 
     with temp_settings(factorize_static_bg=False):
-        lambda_expr = _actToLambdaExpr(model, alpha=1 / functional_unit)
+        # Compute once the generic lambda expression
+        lambda_exprs = _preMultiLCAAlgebric(model=model, methods=[impact_method], alpha=1 / functional_unit, scenarios=scenario)
+
+        lambda_expr = lambda_exprs[0]
 
         # Transform to dict of act => value
-        val_by_act = dict()
+        val_by_act: Dict[Activity, DataFrame] = dict()
         for bg_act in lambda_expr.background_activities:
-            # Dummy impact set to 1 only for current bg act, zero for others
-            impacts = {act: 1.0 if act == bg_act else 0.0 for act in lambda_expr.background_activities}
+            impacts = lambda_expr.impacts_dict()
+
+            # Build temp lambda with only a single act activated, the other impacts being set to Zero
+            filtered_impacts = {act: impacts[act] if act == bg_act else zeros for act in lambda_expr.background_activities}
 
-            bg_expr = lambda_expr.with_impacts(impacts)
-            val = bg_expr.compute(**params).value
-            if isinstance(val, dict):
-                val = val["_all_"]
-            val_by_act[bg_act] = val
+            single_bg_expr = lambda_expr.with_impacts(filtered_impacts)
 
-        if impact_method is not None:
-            # Compute LCA of background activities
-            impact_by_act = _multiLCAWithCache(val_by_act.keys(), [impact_method])
+            res = _postMultiLCAAlgebric(methods=[impact_method], lambdas=[single_bg_expr], **params)
 
-            val_by_act: {act: value * impact_by_act[(act, impact_method)] for act, value in val_by_act.items()}
+            val_by_act[bg_act] = res.values.flatten()
 
         if as_dict:
             return val_by_act
 
         # Transform to dataframe
         items = []
-        for act, value in val_by_act.items():
+        for act, values in val_by_act.items():
+            if filter_zeros and np.all(values == 0.0):
+                continue
+
             item = dict()
 
             for field in fields:
                 if field in act:
                     item[field] = act[field]
 
-            item["value"] = value
+            # Flatten multiple values
+            if len(values) == 1:
+                item["value"] = values[0]
+            else:
+                for i in range(0, len(values)):
+                    item[scenario[i]] = values[i]
             items.append(item)
 
     return DataFrame(items)
@@ -857,9 +895,7 @@ def compute_impacts(
 
             scenarios = scenario if isinstance(scenario, list) else [scenario]
 
-            lambdas = _merge_lambda_scenarios(
-                _preMultiLCAAlgebric(model, methods, scenario=scenario, alpha=alpha, axis=axis) for scenario in scenarios
-            )
+            lambdas = _preMultiLCAAlgebric(model, methods, scenarios=scenarios, alpha=alpha, axis=axis)
 
             unit: Optional[Unit] = functional_unit.units if isinstance(functional_unit, Quantity) else None
 

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/lca_algebraic/methods.py

@@ -4,6 +4,7 @@ from typing import Dict, Tuple
 from pint import Unit
 
 from lca_algebraic.bw_wrapper import Method, methods
+from lca_algebraic.settings import UNITY_METHOD
 
 
 def _impact_labels():
@@ -56,7 +57,11 @@ def findMethods(search=None, mainCat=None):
 def method_unit(method: Tuple, fu_unit: Unit = None):
     """Get the unit of an impact method"""
 
-    res = Method(method).metadata["unit"]
+    if method == UNITY_METHOD:
+        res = "unit"
+    else:
+        res = Method(method).metadata["unit"]
+
     if fu_unit is not None:
         res += f" / {fu_unit}"
 

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/lca_algebraic/settings.py

@@ -11,6 +11,10 @@ except ImportError:
         pass
 
 
+# Custom method that returns "1.0" used for inventories
+UNITY_METHOD = ("UNITY", "UNITY", "UNITY")
+
+
 class CSE(StrEnum):
     NONE = "none"
     DEFAULT = "default"  # Default sympy one

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/lca_algebraic.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lca_algebraic
-Version: 1.4.1.dev1
+Version: 1.4.1.dev3
 Summary: This library provides a layer above Brightway for defining parametric models and running super fast LCA for monte carlo analysis.
 Author-email: OIE - Mines ParisTech <raphael.jolivet@mines-paristech.fr>
 License: BSD

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "lca_algebraic"
-version = "1.4.1.dev1"
+version = "1.4.1.dev3"
 
 description = "This library provides a layer above Brightway for defining parametric models and running super fast LCA for monte carlo analysis."
 readme = "README.md"

{lca_algebraic-1.4.1.dev1 → lca_algebraic-1.4.1.dev3}/test/test_premise.py

@@ -1,7 +1,9 @@
 from numpy.ma.testutils import assert_array_equal
+from pandas import DataFrame
 
-from lca_algebraic import resetDb, newActivity, newFloatParam, compute_impacts, temp_settings
+from lca_algebraic import resetDb, newActivity, newFloatParam, compute_impacts, temp_settings, compute_inventory
 from test.conftest import DataFixture, USER_DB, assert_impacts
+from pandas.testing import assert_frame_equal
 
 
 def test_compute_impact_with_scenario(data: DataFixture):
@@ -73,6 +75,38 @@ def test_compute_impact_with_several_scenarios(data: DataFixture):
     print(res)
 
 
+def test_compute_inventories_with_several_scenarios(data: DataFixture):
+    # Scnerio database are separated with '#'
+    BG_DB1 = "bg#scen1"
+    BG_DB2 = "bg#scen2"
+
+    # Build two databases
+    resetDb(BG_DB1, foreground=False)
+    resetDb(BG_DB2, foreground=False)
+
+    # Add twin BG activities with two different values
+    bga = newActivity(BG_DB1, name="bg_a", unit="kg", exchanges={data.bio1: 1.0})
+    bga_ = newActivity(BG_DB2, name="bg_a", unit="kg", exchanges={data.bio1: 2.0})
+
+    # Add twin BG activities with two different values ibio2
+    bgb = newActivity(BG_DB1, name="bg_b", unit="kg", exchanges={data.bio2: 1.0})
+    bgb_ = newActivity(BG_DB2, name="bg_b", unit="kg", exchanges={data.bio2: 2.0})
+
+    # Build parametrized model on BG1
+    p1 = newFloatParam("p1", default=1, min=0, max=1)
+
+    fg_act = newActivity(USER_DB, "fg", "kg", {bga: p1, bgb: 1.0})
+
+    # Raw impacts without scenario : targeting scen1
+    res = compute_inventory(model=fg_act, impact_method=data.ibio1, p1=[1.0, 2.0], scenario=["scen1", "scen2"])
+
+    assert_frame_equal(
+        rtol=1e-03,
+        left=res,
+        right=DataFrame([{"database": "bg#scen1", "name": "bg_a", "location": "GLO", "unit": "kg", "scen1": 1.0, "scen2": 4.0}]),
+    )
+
+
 def test_compute_impact_with_background_proxy(data: DataFixture):
     """Test the same with background proxy factorization activated"""