Daniel&Dilan review changes

doc/example/hierarchical.ipynb

@@ -237,7 +237,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Parameters to be optimized in the inner problem are selected via the PEtab parameter table by setting a value in the non-standard column `parameterType` (`offset` for offset parameters, `scaling` for scaling parameters, and `sigma` for sigma parameters). When using hierarchical optimization, the nine overriding parameters {offset,scaling,sd}_{pSTAT5A_rel,pSTAT5B_rel,rSTAT5A_rel} are to estimated in the inner problem."
+    "Parameters to be optimized in the inner problem are specified via the PEtab parameter table by setting a value in the non-standard column `parameterType` (`offset` for offset parameters, `scaling` for scaling parameters, and `sigma` for sigma parameters). When using hierarchical optimization, the nine overriding parameters {offset,scaling,sd}_{pSTAT5A_rel,pSTAT5B_rel,rSTAT5A_rel} are to estimated in the inner problem."
    ]
   },
   {

pypesto/hierarchical/calculator.py

@@ -123,6 +123,7 @@ def __call__(
                 inner_result[HESS] = np.full(
                     shape=(dim, dim), fill_value=np.nan
                 )
+            inner_result[INNER_PARAMETERS] = None
             return inner_result
 
         inner_parameters = self.inner_solver.solve(

pypesto/hierarchical/inner_calculator_collector.py

@@ -418,7 +418,11 @@ def __call__(
         # only if the objective value improved.
         if ret[FVAL] < self.best_fval:
             ret[X_INNER_OPT] = all_inner_pars
-            ret[INNER_PARAMETERS] = interpretable_inner_pars
+            ret[INNER_PARAMETERS] = (
+                interpretable_inner_pars
+                if len(interpretable_inner_pars) > 0
+                else None
+            )
             self.best_fval = ret[FVAL]
 
         return filter_return_dict(ret)

pypesto/hierarchical/spline_approximation/problem.py

@@ -184,7 +184,7 @@ def get_inner_parameter_dictionary(self) -> Dict:
             inner_par_dict[x_id] = x.value
         return inner_par_dict
 
-    def get_inner_noise_parameters(self) -> Dict:
+    def get_inner_noise_parameters(self) -> list[float]:
         """Get a list with all noise parameter values."""
         return [
             x.value for x in self.get_xs_for_type(InnerParameterType.SIGMA)

pypesto/objective/amici/amici.py

@@ -3,7 +3,7 @@
 import os
 import tempfile
 from collections import OrderedDict
-from typing import TYPE_CHECKING, Dict, List, Optional, Sequence, Tuple, Union
+from typing import TYPE_CHECKING, Dict, Optional, Sequence, Tuple, Union
 
 import numpy as np
 
@@ -215,7 +215,7 @@ def __init__(
         self.custom_timepoints = None
 
         # Initialize the dictionary for saving of inner parameters.
-        self.inner_parameters: List[float] = []
+        self.inner_parameters: list[float] = []
 
     def get_config(self) -> dict:
         """Return basic information of the objective configuration."""
@@ -456,7 +456,7 @@ def call_unprocessed(
 
         nllh = ret[FVAL]
         rdatas = ret[RDATAS]
-        if INNER_PARAMETERS in ret and any(ret[INNER_PARAMETERS]):
+        if INNER_PARAMETERS in ret and ret[INNER_PARAMETERS] is not None:
             self.inner_parameters = ret[INNER_PARAMETERS]
 
         # check whether we should update data for preequilibration guesses

pypesto/optimize/optimizer.py

@@ -41,12 +41,13 @@ def __init__(self, optimizer: str):
         )
 
 
-def add_inner_parameters(
+def _add_inner_parameters(
     objective: Objective, optimizer_result: OptimizerResult
 ):
     """Add inner parameters from objective to the optimizer result."""
-    if hasattr(objective, INNER_PARAMETERS) and any(
-        objective.inner_parameters
+    if (
+        hasattr(objective, INNER_PARAMETERS)
+        and objective.inner_parameters is not None
     ):
         optimizer_result[INNER_PARAMETERS] = objective.inner_parameters
 
@@ -472,7 +473,7 @@ def fun(x):
             exitflag=res.status,
             message=res.message,
         )
-        add_inner_parameters(objective, optimizer_result)
+        _add_inner_parameters(objective, optimizer_result)
 
         return optimizer_result
 
@@ -618,7 +619,7 @@ def get_fval_vararg(*x):
 
         optimizer_result = OptimizerResult()
 
-        add_inner_parameters(objective, optimizer_result)
+        _add_inner_parameters(objective, optimizer_result)
 
         return optimizer_result
 
@@ -682,7 +683,7 @@ def minimize(
 
         optimizer_result = OptimizerResult(x=np.array(xopt), fval=fopt)
 
-        add_inner_parameters(problem.objective, optimizer_result)
+        _add_inner_parameters(problem.objective, optimizer_result)
 
         return optimizer_result
 
@@ -771,7 +772,7 @@ def minimize(
             x=np.array(result[0]), fval=result[1]
         )
 
-        add_inner_parameters(problem.objective, optimizer_result)
+        _add_inner_parameters(problem.objective, optimizer_result)
 
         return optimizer_result
 
@@ -842,7 +843,7 @@ def minimize(
             x=np.array(result.x), fval=result.fun
         )
 
-        add_inner_parameters(problem.objective, optimizer_result)
+        _add_inner_parameters(problem.objective, optimizer_result)
 
         return optimizer_result
 
@@ -963,7 +964,7 @@ def successively_working_fval(swarm: np.ndarray) -> np.ndarray:
             fval=float(cost),
         )
 
-        add_inner_parameters(problem.objective, optimizer_result)
+        _add_inner_parameters(problem.objective, optimizer_result)
 
         return optimizer_result
 
@@ -1197,7 +1198,7 @@ def nlopt_objective(x, grad):
             exitflag=opt.last_optimize_result(),
         )
 
-        add_inner_parameters(problem.objective, optimizer_result)
+        _add_inner_parameters(problem.objective, optimizer_result)
 
         return optimizer_result
 
@@ -1385,7 +1386,7 @@ def minimize(
             exitflag=opt.exitflag,
         )
 
-        add_inner_parameters(problem.objective, optimizer_result)
+        _add_inner_parameters(problem.objective, optimizer_result)
 
         return optimizer_result
 

pypesto/visualize/parameters.py

@@ -376,33 +376,9 @@ def handle_inputs(
     # retrieve results
     fvals = result.optimize_result.fval
     xs = result.optimize_result.x
-    # retrieve inner parameters if available
-    inner_xs = [
-        res.get(INNER_PARAMETERS, None) for res in result.optimize_result.list
-    ]
-
-    from ..hierarchical.calculator import HierarchicalAmiciCalculator
 
-    if (
-        any(inner_x is not None for inner_x in inner_xs)
-        and hasattr(result.problem.objective, 'calculator')
-        and isinstance(
-            inner_calculator := result.problem.objective.calculator,
-            HierarchicalAmiciCalculator,
-        )
-    ):
-        inner_xs_names = inner_calculator.inner_problem.get_x_ids()
-        # replace None with a list of nans
-        inner_xs = [
-            [np.nan for i in range(len(inner_xs_names))]
-            if inner_xs_idx is None
-            else np.asarray(inner_xs_idx)
-            for inner_xs_idx in inner_xs
-        ]
-        # set bounds for inner parameters
-        inner_lb, inner_ub = inner_calculator.inner_problem.get_bounds()
-    else:
-        inner_xs = None
+    # retrieve inner parameters if available
+    inner_xs, inner_xs_names, inner_lb, inner_ub = handle_inner_inputs(result)
 
     # parse indices which should be plotted
     if start_indices is not None:
@@ -454,6 +430,57 @@ def handle_inputs(
     return lb, ub, x_labels, fvals_out, xs_out
 
 
+def handle_inner_inputs(
+    result: Result,
+):
+    """Handle inner parameters if available.
+
+    Parameters
+    ----------
+    result:
+        Optimization result obtained by 'optimize.py'.
+
+    Returns
+    -------
+    inner_xs:
+        Inner parameter values which will be appended to xs.
+    inner_xs_names:
+        Inner parameter names.
+    inner_lb:
+        Inner parameter lower bounds.
+    inner_ub:
+        Inner parameter upper bounds.
+    """
+    inner_xs = [
+        res.get(INNER_PARAMETERS, None) for res in result.optimize_result.list
+    ]
+
+    if any(inner_x is not None for inner_x in inner_xs):
+        from ..hierarchical.calculator import HierarchicalAmiciCalculator
+
+        if hasattr(result.problem.objective, 'calculator') and isinstance(
+            inner_calculator := result.problem.objective.calculator,
+            HierarchicalAmiciCalculator,
+        ):
+            inner_xs_names = inner_calculator.inner_problem.get_x_ids()
+            # replace None with a list of nans
+            inner_xs = [
+                [np.nan for i in range(len(inner_xs_names))]
+                if inner_xs_idx is None
+                else np.asarray(inner_xs_idx)
+                for inner_xs_idx in inner_xs
+            ]
+            # set bounds for inner parameters
+            inner_lb, inner_ub = inner_calculator.inner_problem.get_bounds()
+        else:
+            inner_xs = None
+            inner_xs_names = None
+            inner_lb = None
+            inner_ub = None
+
+    return inner_xs, inner_xs_names, inner_lb, inner_ub
+
+
 def parameters_correlation_matrix(
     result: Result,
     parameter_indices: Union[str, Sequence[int]] = 'free_only',

test/visualize/test_visualize.py

@@ -459,7 +459,7 @@ def test_parameters_hist():
 @pytest.mark.parametrize("scale_to_interval", [None, (0, 1)])
 @close_fig
 def test_parameters_hierarchical(scale_to_interval):
-    # bbtain a petab problem with hierarchical parameters
+    # obtain a petab problem with hierarchical parameters
     petab_problem = (
         get_Boehm_JProteomeRes2014_hierarchical_petab_corrected_bounds()
     )