Fix parameter set logic and allow FittingProblem to take init_ocv rather than init_soc

CHANGELOG.md

@@ -16,6 +16,8 @@
 
 ## Breaking Changes
 
+- [#424](https://github.com/pybop-team/PyBOP/issues/424) - Replaces the `init_soc` input to `FittingProblem` with the option to pass an initial OCV value, updates `BaseModel` and fixes `multi_model_identification.ipynb` and `spm_electrode_design.ipynb`.
+
 # [v24.6.1](https://github.com/pybop-team/PyBOP/tree/v24.6.1) - 2024-07-31
 
 ## Features

examples/notebooks/multi_optimiser_identification.ipynb

@@ -116,7 +116,7 @@
    "source": [
     "t_eval = np.arange(0, 2000, 10)\n",
     "init_soc = 1.0\n",
-    "values = synth_model.predict(t_eval=t_eval, init_soc=init_soc)"
+    "values = synth_model.predict(t_eval=t_eval, initial_state=init_soc)"
    ]
   },
   {
@@ -448,7 +448,8 @@
    "source": [
     "optims = []\n",
     "xs = []\n",
-    "problem = pybop.FittingProblem(model, parameters, dataset, init_soc=init_soc)\n",
+    "model.set_initial_state(init_soc)\n",
+    "problem = pybop.FittingProblem(model, parameters, dataset)\n",
     "cost = pybop.SumSquaredError(problem)\n",
     "for optimiser in gradient_optimisers:\n",
     "    print(f\"Running {optimiser.__name__}\")\n",

examples/notebooks/optimiser_calibration.ipynb

@@ -110,7 +110,7 @@
     "    ]\n",
     "    * 2\n",
     ")\n",
-    "values = model.predict(init_soc=init_soc, experiment=experiment)"
+    "values = model.predict(initial_state=init_soc, experiment=experiment)"
    ]
   },
   {
@@ -240,7 +240,7 @@
     }
    ],
    "source": [
-    "problem = pybop.FittingProblem(model, parameters, dataset, init_soc=init_soc)\n",
+    "problem = pybop.FittingProblem(model, parameters, dataset)\n",
     "cost = pybop.SumSquaredError(problem)\n",
     "optim = pybop.GradientDescent(cost, sigma0=0.2, max_iterations=100)"
    ]
@@ -451,7 +451,7 @@
     "optims = []\n",
     "for sigma in sigmas:\n",
     "    print(sigma)\n",
-    "    problem = pybop.FittingProblem(model, parameters, dataset, init_soc=init_soc)\n",
+    "    problem = pybop.FittingProblem(model, parameters, dataset)\n",
     "    cost = pybop.SumSquaredError(problem)\n",
     "    optim = pybop.GradientDescent(cost, sigma0=sigma, max_iterations=100)\n",
     "    x, final_cost = optim.run()\n",

examples/scripts/cuckoo.py

@@ -4,6 +4,12 @@
 
 # Define model
 parameter_set = pybop.ParameterSet.pybamm("Chen2020")
+parameter_set.update(
+    {
+        "Negative electrode active material volume fraction": 0.7,
+        "Positive electrode active material volume fraction": 0.67,
+    }
+)
 model = pybop.lithium_ion.SPM(parameter_set=parameter_set)
 
 # Fitting parameters
@@ -13,14 +19,12 @@
         prior=pybop.Gaussian(0.6, 0.05),
         bounds=[0.4, 0.75],
         initial_value=0.41,
-        true_value=0.7,
     ),
     pybop.Parameter(
         "Positive electrode active material volume fraction",
         prior=pybop.Gaussian(0.48, 0.05),
         bounds=[0.4, 0.75],
         initial_value=0.41,
-        true_value=0.67,
     ),
 )
 init_soc = 0.7
@@ -32,9 +36,7 @@
         ),
     ]
 )
-values = model.predict(
-    init_soc=init_soc, experiment=experiment, inputs=parameters.as_dict("true")
-)
+values = model.predict(initial_state=init_soc, experiment=experiment)
 
 sigma = 0.002
 corrupt_values = values["Voltage [V]"].data + np.random.normal(
@@ -51,7 +53,7 @@
 )
 
 # Generate problem, cost function, and optimisation class
-problem = pybop.FittingProblem(model, parameters, dataset, init_soc=init_soc)
+problem = pybop.FittingProblem(model, parameters, dataset)
 cost = pybop.GaussianLogLikelihood(problem, sigma0=sigma * 4)
 optim = pybop.Optimisation(
     cost,

examples/scripts/exp_UKF.py

@@ -4,7 +4,7 @@
 from examples.standalone.model import ExponentialDecay
 
 # Parameter set and model definition
-parameter_set = {"k": "[input]", "y0": "[input]"}
+parameter_set = {"k": 0.1, "y0": 1.0}
 model = ExponentialDecay(parameter_set=parameter_set, n_states=1)
 
 # Fitting parameters
@@ -13,22 +13,21 @@
         "k",
         prior=pybop.Gaussian(0.1, 0.05),
         bounds=[0, 1],
-        true_value=0.1,
+        true_value=parameter_set["k"],
     ),
     pybop.Parameter(
         "y0",
         prior=pybop.Gaussian(1, 0.05),
         bounds=[0, 3],
-        true_value=1.0,
+        true_value=parameter_set["y0"],
     ),
 )
 
 # Make a prediction with measurement noise
 sigma = 1e-2
 t_eval = np.linspace(0, 20, 10)
-model.parameters = parameters
 true_inputs = parameters.as_dict("true")
-values = model.predict(t_eval=t_eval, inputs=true_inputs)
+values = model.predict(t_eval=t_eval)
 values = values["2y"].data
 corrupt_values = values + np.random.normal(0, sigma, len(t_eval))
 

examples/scripts/spm_AdamW.py

@@ -30,7 +30,7 @@
     ]
     * 2
 )
-values = model.predict(init_soc=init_soc, experiment=experiment)
+values = model.predict(initial_state=init_soc, experiment=experiment)
 
 
 def noise(sigma):
@@ -50,9 +50,7 @@ def noise(sigma):
 
 signal = ["Voltage [V]", "Bulk open-circuit voltage [V]"]
 # Generate problem, cost function, and optimisation class
-problem = pybop.FittingProblem(
-    model, parameters, dataset, signal=signal, init_soc=init_soc
-)
+problem = pybop.FittingProblem(model, parameters, dataset, signal=signal)
 cost = pybop.Minkowski(problem, p=2)
 optim = pybop.AdamW(
     cost,

examples/scripts/spm_MAP.py

@@ -46,7 +46,7 @@
     ]
 )
 values = model.predict(
-    init_soc=init_soc, experiment=experiment, inputs=parameters.true_value()
+    initial_state=init_soc, experiment=experiment, inputs=parameters.true_value()
 )
 corrupt_values = values["Voltage [V]"].data + np.random.normal(
     0, sigma, len(values["Voltage [V]"].data)
@@ -62,7 +62,7 @@
 )
 
 # Generate problem, cost function, and optimisation class
-problem = pybop.FittingProblem(model, parameters, dataset, init_soc=init_soc)
+problem = pybop.FittingProblem(model, parameters, dataset)
 cost = pybop.MAP(problem, pybop.GaussianLogLikelihoodKnownSigma, sigma0=sigma)
 optim = pybop.AdamW(
     cost,

examples/scripts/spm_NelderMead.py

@@ -30,7 +30,7 @@
     ]
     * 2
 )
-values = model.predict(init_soc=init_soc, experiment=experiment)
+values = model.predict(initial_state=init_soc, experiment=experiment)
 
 
 def noise(sigma):
@@ -48,11 +48,9 @@ def noise(sigma):
     }
 )
 
-signal = ["Voltage [V]", "Bulk open-circuit voltage [V]"]
 # Generate problem, cost function, and optimisation class
-problem = pybop.FittingProblem(
-    model, parameters, dataset, signal=signal, init_soc=init_soc
-)
+signal = ["Voltage [V]", "Bulk open-circuit voltage [V]"]
+problem = pybop.FittingProblem(model, parameters, dataset, signal=signal)
 cost = pybop.RootMeanSquaredError(problem)
 optim = pybop.NelderMead(
     cost,

examples/scripts/spm_UKF.py

@@ -36,12 +36,9 @@
 )
 
 # Build the model to get the number of states
-model.build(dataset=dataset.data, parameters=parameters)
+model.build(dataset=dataset, parameters=parameters)
 
 # Define the UKF observer, setting the particle boundaries as uncertain states
-signal = ["Voltage [V]"]
-n_states = model.n_states
-n_signals = len(signal)
 covariance = np.diag([0] * 20 + [sigma**2] + [0] * 20 + [sigma**2])
 process_noise = np.diag([0] * 20 + [1e-6] + [0] * 20 + [1e-6])
 measurement_noise = np.diag([sigma**2])
@@ -52,7 +49,6 @@
     process_noise,
     measurement_noise,
     dataset,
-    signal=signal,
 )
 
 # Generate problem, cost function, and optimisation class

examples/scripts/spm_scipymin.py

@@ -34,7 +34,7 @@
 
 # Define the cost to optimise
 signal = ["Voltage [V]"]
-problem = pybop.FittingProblem(model, parameters, dataset, signal=signal, init_soc=0.98)
+problem = pybop.FittingProblem(model, parameters, dataset, signal=signal)
 cost = pybop.RootMeanSquaredError(problem)
 
 # Build the optimisation problem

examples/scripts/spm_weighted_cost.py

@@ -34,7 +34,7 @@
     ]
     * 2
 )
-values = model.predict(experiment=experiment, init_soc=init_soc)
+values = model.predict(experiment=experiment, initial_state=init_soc)
 
 
 def noise(sigma):
@@ -51,7 +51,7 @@ def noise(sigma):
 )
 
 # Generate problem, cost function, and optimisation class
-problem = pybop.FittingProblem(model, parameters, dataset, init_soc=init_soc)
+problem = pybop.FittingProblem(model, parameters, dataset)
 cost1 = pybop.SumSquaredError(problem)
 cost2 = pybop.RootMeanSquaredError(problem)
 weighted_cost = pybop.WeightedCost(cost1, cost2, weights=[0.1, 1])

examples/standalone/problem.py

@@ -18,9 +18,7 @@ def __init__(
         additional_variables=None,
         init_soc=None,
     ):
-        super().__init__(
-            parameters, model, check_model, signal, additional_variables, init_soc
-        )
+        super().__init__(parameters, model, check_model, signal, additional_variables)
         self._dataset = dataset.data
 
         # Check that the dataset contains time and current

pybop/costs/_weighted_cost.py

@@ -24,10 +24,10 @@ class WeightedCost(BaseCost):
         If True, the shared problem will be evaluated once and saved before the
         self._evaluate() method of each cost is called (default: False).
     _has_separable_problem: bool
-        If True, the shared problem is seperable from the cost function and
-        will be evaluated for each problem before the cost evaluation is
-        called (default: False). This attribute is used for sub-cost objects;
-        however, the top-level WeightedCost attribute is not used (i.e. == False).
+        This attribute must be set to False for WeightedCost objects. If the
+        corresponding attribute of an individual cost is True, the problem is
+        separable from the cost function and will be evaluated before the
+        individual cost evaluation is called.
     """
 
     def __init__(self, *costs, weights: Optional[list[float]] = None):