strip down extra code, try to debug

psimulate is finding same dalys in all scenarios, but I don't know why

README.rst

@@ -21,27 +21,7 @@ your operating system at the following places:
 Once you have all three installed, you should open up your normal shell
 (if you're on linux or OSX) or the ``git bash`` shell if you're on windows.
 You'll then make an environment, clone this repository, then install
-all necessary requirements as follows::
-
-  :~$ conda create --name=vivarium_sodium_reduction python=3.11
-  ...conda will download python and base dependencies...
-  :~$ conda activate vivarium_sodium_reduction
-  (vivarium_sodium_reduction) :~$ git clone https://github.com/ihmeuw/vivarium_sodium_reduction.git
-  ...git will copy the repository from github and place it in your current directory...
-  (vivarium_sodium_reduction) :~$ cd vivarium_sodium_reduction
-  (vivarium_sodium_reduction) :~$ pip install -e .
-  ...pip will install vivarium and other requirements...
-
-Supported Python versions: 3.9, 3.10, 3.11
-
-Note the ``-e`` flag that follows pip install. This will install the python
-package in-place, which is important for making the model specifications later.
-
-To install requirements from a provided requirements.txt (e.g. installing an
-archived repository with the exact same requirements it was run with), replace
-`pip install -e .` with the following::
-
-  (vivarium_sodium_reduction) :~$ pip install -r requirements.txt
+all necessary requirements with `source evironment.sh`.
 
 Cloning the repository should take a fair bit of time as git must fetch
 the data artifact associated with the demo (several GB of data) from the
@@ -50,16 +30,7 @@ you are likely only retrieving the checksum file that github holds onto,
 and your simulations will fail.** If you are only retrieving checksum
 files you can explicitly pull the data by executing ``git-lfs pull``.
 
-Vivarium uses the Hierarchical Data Format (HDF) as the backing storage
-for the data artifacts that supply data to the simulation. You may not have
-the needed libraries on your system to interact with these files, and this is
-not something that can be specified and installed with the rest of the package's
-dependencies via ``pip``. If you encounter HDF5-related errors, you should
-install hdf tooling from within your environment like so::
-
-  (vivarium_sodium_reduction) :~$ conda install hdf5
-
-The ``(vivarium_sodium_reduction)`` that precedes your shell prompt will probably show
+The ``(vivarium_sodium_reduction_simulation)`` that precedes your shell prompt will probably show
 up by default, though it may not.  It's just a visual reminder that you
 are installing and running things in an isolated programming environment
 so it doesn't conflict with other source code and libraries on your
@@ -123,12 +94,7 @@ step. For more ways to run simulations, see the tutorials at
 https://vivarium.readthedocs.io/en/latest/tutorials/running_a_simulation/index.html
 and https://vivarium.readthedocs.io/en/latest/tutorials/exploration.html
 
-To run multiple simulations in parallel, you can use the `psimulate` command, which you can install with::
-
-   (vivarium_sodium_reduction) :~$ pip install vivarium_cluster_tools
-   (vivarium_sodium_reduction) :~$ conda install redis
-
-Then you can run simulations in parallel on IHME's cluster::
+To run multiple simulations in parallel, you can use the `psimulate` command::
 
-   (vivarium_sodium_reduction) :~$ 
-   psimulate run -o /mnt/share/homes/abie/vivarium_results/sodium_usa/ -P proj_simscience src/vivarium_sodium_reduction/model_specifications/model_spec.yaml src/vivarium_sodium_reduction/model_specifications/branches/scenarios.yaml
+  psimulate run -vvv -m 10 -P proj_simscience src/vivarium_sodium_reduction/model_specifications/model_spec.yaml sr
+c/vivarium_sodium_reduction/model_specifications/branches/scenarios.yaml

src/vivarium_sodium_reduction/components/interventions.py

@@ -37,7 +37,7 @@ def setup(self, builder: Builder) -> None:
         self.population_view = builder.population.get_view(["age"])
 
         builder.value.register_value_modifier(
-            self.target, modifier=self.adjust_exposure, requires_columns=["age"]
+            f'{self.target}.exposure', modifier=self.adjust_exposure, requires_columns=["age"]
         )
 
     def adjust_exposure(self, index: pd.Index, exposure: pd.Series) -> pd.Series:

src/vivarium_sodium_reduction/components/risks.py

@@ -16,7 +16,6 @@
 )
 from vivarium_public_health.utilities import EntityString
 
-
 class DropValueRisk(Risk):
     def __init__(self, risk: str):
         super().__init__(risk)
@@ -60,133 +59,6 @@ def post_processor(exposure, _):
 
         return post_processor
 
-
-class CorrelatedRisk(DropValueRisk):
-    """A risk that can be correlated with another risk.
-
-    TODO: document strategy used in this component in more detail,
-    Abie had an AI adapt it from https://github.com/ihmeuw/vivarium_nih_us_cvd"""
-
-    @property
-    def columns_created(self) -> List[str]:
-        return []
-
-    @property
-    def columns_required(self) -> Optional[List[str]]:
-        return [self.propensity_column_name]
-
-    @property
-    def initialization_requirements(self) -> Dict[str, List[str]]:
-        return {
-            "requires_columns": [],
-            "requires_values": [],
-            "requires_streams": [],
-        }
-
-    def on_initialize_simulants(self, pop_data: SimulantData) -> None:
-        pass
-
-    def on_time_step_prepare(self, event: Event) -> None:
-        pass
-
-
-class ThresholdRisk(Component):
-    """A component that generates a risk based on a threshold of another risk."""
-
-    def __init__(self, risk: str, threshold: str):
-        super().__init__()
-        self.risk = EntityString(risk)
-        self.threshold = float(threshold)
-        self.exposure_pipeline_name = f"{self.risk.name}.threshold_exposure"
-
-    def setup(self, builder: Builder) -> None:
-        self.continuous_exposure = builder.value.get_value(self.risk.exposure_pipeline_name)
-        self.threshold_exposure = builder.value.register_value_producer(
-            self.exposure_pipeline_name,
-            source=self.continuous_exposure
-        )
-
-    def get_exposure_threshold_value(self, value):
-
-        return np.where(value <= self.threshold, 'cat1', 'cat2')
-
-
-
-class RiskCorrelation(Component):
-    """A component that generates a specified correlation between two risk exposures."""
-
-    @property
-    def columns_created(self) -> List[str]:
-        return self.propensity_column_names + self.exposure_column_names
-
-    @property
-    def columns_required(self) -> Optional[List[str]]:
-        return ["age"]
-
-    @property
-    def initialization_requirements(self) -> Dict[str, List[str]]:
-        return {"requires_columns": ["age"] + self.ensemble_propensities}
-
-    def __init__(self, risk1: str, risk2: str, correlation: str):
-        super().__init__()
-        correlated_risks = [risk1, risk2]
-        correlation_matrix = np.array([[1, float(correlation)], [float(correlation), 1]])
-        self.correlated_risks = [EntityString(risk) for risk in correlated_risks]
-        self.correlation_matrix = correlation_matrix
-        self.propensity_column_names = [
-            f"{risk.name}_propensity" for risk in self.correlated_risks
-        ]
-        self.exposure_column_names = [
-            f"{risk.name}_exposure" for risk in self.correlated_risks
-        ]
-        self.ensemble_propensities = [
-            f"ensemble_propensity_" + risk
-            for risk in self.correlated_risks
-            if risk_factors[risk.name].distribution == "ensemble"
-        ]
-
-    def setup(self, builder: Builder) -> None:
-        self.distributions = {
-            risk: builder.components.get_component(risk).exposure_distribution
-            for risk in self.correlated_risks
-        }
-        self.exposures = {
-            risk: builder.value.get_value(f"{risk.name}.exposure")
-            for risk in self.correlated_risks
-        }
-        self.input_draw = builder.configuration.input_data.input_draw_number
-        self.random_seed = builder.configuration.randomness.random_seed
-
-    def on_initialize_simulants(self, pop_data: SimulantData) -> None:
-        pop = self.population_view.subview(["age"]).get(pop_data.index)
-        propensities = pd.DataFrame(index=pop.index)
-
-        np.random.seed(get_hash(f"{self.input_draw}_{self.random_seed}"))
-        probit_propensity = np.random.multivariate_normal(
-            mean=[0] * len(self.correlated_risks), cov=self.correlation_matrix, size=len(pop)
-        )
-        correlated_propensities = scipy.stats.norm().cdf(probit_propensity)
-        propensities[self.propensity_column_names] = correlated_propensities
-
-        def get_exposure_from_propensity(propensity_col: pd.Series) -> pd.Series:
-            risk = propensity_col.name.replace("_propensity", "")
-            exposure_values = self.distributions["risk_factor." + risk].ppf(propensity_col)
-            return pd.Series(exposure_values)
-
-        exposures = propensities.apply(get_exposure_from_propensity)
-        exposures.columns = [
-            col.replace("_propensity", "_exposure") for col in propensities.columns
-        ]
-
-        self.population_view.update(pd.concat([propensities, exposures], axis=1))
-
-    def on_time_step_prepare(self, event: Event) -> None:
-        for risk in self.exposures:
-            exposure_values = self.exposures[risk](event.index)
-            exposure_col = pd.Series(exposure_values, name=f"{risk.name}_exposure")
-            self.population_view.update(exposure_col)
-
-
 class SodiumSBPEffect(Component):
     @property
     def name(self):
@@ -198,22 +70,21 @@ def setup(self, builder: Builder):
 
         builder.value.register_value_modifier(
             "high_systolic_blood_pressure.drop_value",
-            modifier=self.sodium_effect_on_sbp,
+            modifier=self.sodium_effect_on_sbp_drop,
             requires_columns=["age", "sex"],
             requires_values=[
                 "diet_high_in_sodium.exposure",
                 "diet_high_in_sodium.raw_exposure",
             ],
         )
 
-    def sodium_effect_on_sbp(self, index, sbp_drop_value):
+    def sodium_effect_on_sbp_drop(self, index, sbp_drop_value):
         sodium_exposure = self.sodium_exposure(index)
         sodium_exposure_raw = self.sodium_exposure_raw(index)
 
         # FIXME: this should go in the constants.py file
         mmHg_per_g_sodium = 5.8/6.0  # 5.8 (2.5, 9.2) mmHg decrease per 6g/day sodium decrease
 
-        sbp_increase = pd.Series(0, index=index)
         sodium_drop = sodium_exposure_raw - sodium_exposure
 
         sbp_drop_due_to_sodium_drop = sodium_drop * mmHg_per_g_sodium

src/vivarium_sodium_reduction/model_specifications/branches/scenarios.yaml

@@ -1,5 +1,5 @@
-input_draw_count: 5
-random_seed_count: 30
+input_draw_count: 1
+random_seed_count: 1
 
 branches:
     - relative_shift_intervention_on_diet_high_in_sodium:

src/vivarium_sodium_reduction/model_specifications/model_spec.yaml

@@ -32,9 +32,8 @@ components:
                                      'cause.stomach_cancer.incidence_rate')
     vivarium_sodium_reduction.components:
         risks:
-            - CorrelatedRisk('risk_factor.high_systolic_blood_pressure')
-            - CorrelatedRisk('risk_factor.diet_high_in_sodium')
-            - RiskCorrelation('risk_factor.high_systolic_blood_pressure', 'risk_factor.diet_high_in_sodium', '0.0')
+            - DropValueRisk('risk_factor.high_systolic_blood_pressure')
+            - DropValueRisk('risk_factor.diet_high_in_sodium')
             - SodiumSBPEffect()
         interventions:
             - RelativeShiftIntervention('diet_high_in_sodium')
@@ -61,12 +60,12 @@ configuration:
             day: 31
         step_size: 168 # 28*6 (approximately 6 months, in days)
     population:
-        population_size: 100_000
+        population_size: 10_000
         initialization_age_min: 0
         initialization_age_max: 100
 
     relative_shift_intervention_on_diet_high_in_sodium:
-        shift_factor: 0.85
+        shift_factor: 2.0
         age_start: 0
         age_end: 125