Regrow later years when a caller sets an input for an earlier year

changelog.d/2774.fixed.md

@@ -0,0 +1 @@
+When a caller sets an input for an earlier year after a calculation, recalculate the later years the default branch had grown, carried over or calculated from earlier inputs, as a named branch and a new simulation do.

docs/book/usage/scenarios.md

@@ -186,13 +186,24 @@ calculations retain them too. To undo an override, call
 variable's formula or default runs again. A reset alone no longer undoes an
 input supplied after construction.
 
-Every top-level `set_input` on a built simulation queues dependent-cache cleanup
-before its next outermost read. This includes simulations with no modifier: a
+Every top-level `Simulation.set_input` on a built simulation queues
+dependent-cache cleanup before its next outermost read. This includes simulations with no modifier: a
 caller can read Child Benefit, add an amount and then obtain fresh income and
 response values. Inputs stored only on another clone or branch do not become
 the parent's inputs. Formula and framework writes during a calculation keep
 their existing selective invalidation.
 
+In that cleanup, as in `reset_calculations`, the simulation keeps only the
+years of each input variable that the data, a modifier or a caller supplied.
+It calculates the other years again, whether it grew them from an earlier
+input by the variable's uprating index, carried an earlier input into them,
+ran the variable's formula for them or, for a stored household total, grew
+them from an earlier total. So when a caller sets an input for an earlier year
+after a calculation, the later years the simulation had already filled follow
+it, as they do on a named branch and in a new simulation that gets the same
+inputs. The simulation keeps a later year that the data or `set_input`
+supplied.
+
 Direct edits of the parameter tree require a separate array-cache boundary.
 After calling a parameter's `update` method on a built simulation, call
 `simulation.drop_computed_arrays()` before reading its results. This public

policyengine_uk/simulation.py

@@ -85,8 +85,8 @@
 from policyengine_uk.utils.data_source import built_from_data
 from policyengine_uk.utils.input_dataframe_export import input_dataframe
 from policyengine_uk.utils.measurement_branches import (
-    drop_calculated_totals,
     drop_calculated_values,
+    drop_calculated_years,
     retained_input_variables,
 )
 from policyengine_uk.utils.supplied_inputs import (
@@ -815,10 +815,9 @@ def _stored_household_total(self, variable, period):
                 continue
             native = self.tax_benefit_system.variables[part].definition_period
             if part_source.start > source.start or part_source.unit != native:
-                # Person balances stored for a later year than the total are
-                # the newer observation, and Core carries a balance stored in
-                # another unit flat rather than uprating it, so their sum
-                # stands.
+                # Person balances stored for a later year than the total give
+                # the newer observation, and Core doesn't uprate a balance
+                # stored in another unit, so their sum stands.
                 return None
         try:
             index_name = stored_total_uprating(self.tax_benefit_system, variable.name)
@@ -1119,8 +1118,9 @@ def reset_calculations(self):
         for variable in self.tax_benefit_system.variables:
             if variable not in retained:
                 self.delete_arrays(variable)
-            elif variable in STORED_HOUSEHOLD_TOTALS:
-                drop_calculated_totals(self, variable)
+        drop_calculated_years(
+            self, retained.intersection(self.tax_benefit_system.variables)
+        )
 
     def get_known_variables(self):
         variables = []