add renaming of nn to ml

agentlib_mpc/data_structures/__init__.py

@@ -1,3 +1 @@
 
-
-

agentlib_mpc/data_structures/casadi_utils.py

@@ -247,7 +247,7 @@ def _create_proxqp_solver(self, nlp: dict, options: dict):
             "verbose": False,
             "print_time": False,
             "record_time": True,
-            "proxqp": {"max_iter": 200, "eps_abs": 1e-4, "backend": "sparse"}
+            "proxqp": {"max_iter": 200, "eps_abs": 1e-4, "backend": "sparse"},
         }
         opts = {**default_opts, **options}
         return ca.qpsol("mpc", "proxqp", nlp, opts)
@@ -257,7 +257,7 @@ def _create_osqp_solver(self, nlp: dict, options: dict):
             "verbose": False,
             "print_time": False,
             "record_time": True,
-            "osqp": {"max_iter": 200, "eps_abs": 1e-4, "verbose": False}
+            "osqp": {"max_iter": 200, "eps_abs": 1e-4, "verbose": False},
         }
         opts = {**default_opts, **options}
         return ca.qpsol("mpc", "osqp", nlp, opts)

agentlib_mpc/data_structures/mpc_datamodels.py

@@ -42,11 +42,9 @@ class DiscretizationOptions(pydantic.BaseModel):
 
 
 class Results(Protocol):
-
     df: pd.DataFrame
 
-    def __getitem__(self, item: str) -> Sequence[float]:
-        ...
+    def __getitem__(self, item: str) -> Sequence[float]: ...
 
 
 @dataclasses.dataclass

agentlib_mpc/models/__init__.py

@@ -1,6 +1,7 @@
 """
 Package containing models for agentlib_mpc.
 """
+
 from agentlib.utils.plugin_import import ModuleImport
 
 MODEL_TYPES = {

agentlib_mpc/models/casadi_ml_model.py

@@ -1,4 +1,5 @@
 """Holds the classes for CasADi variables and the CasADi model."""
+
 import itertools
 import logging
 from itertools import chain
@@ -379,9 +380,9 @@ def register_ml_models(
         for output in self.config.outputs + self.config.states:
             for serialized_output_names, ml_model in ml_model_sources_dict.items():
                 if output.name in serialized_output_names:
-                    output_to_ml_model[
-                        output.name
-                    ] = CasadiPredictor.from_serialized_model(ml_model)
+                    output_to_ml_model[output.name] = (
+                        CasadiPredictor.from_serialized_model(ml_model)
+                    )
                     ml_model_dict[output.name] = ml_model
         casadi_ml_model_dict: Dict[str, CasadiPredictor] = output_to_ml_model
         return ml_model_dict, casadi_ml_model_dict

agentlib_mpc/models/casadi_model.py

@@ -1,4 +1,5 @@
 """Holds the classes for CasADi variables and the CasADi model."""
+
 import json
 import logging
 import abc

agentlib_mpc/models/casadi_predictor.py

@@ -40,9 +40,9 @@ class Config:
     def __init__(self, serialized_model: SerializedMLModel) -> None:
         """Initialize Predictor class."""
         self.serialized_model: SerializedMLModel = serialized_model
-        self.predictor_model: Union[
-            Sequential, CustomGPR, LinearRegression
-        ] = serialized_model.deserialize()
+        self.predictor_model: Union[Sequential, CustomGPR, LinearRegression] = (
+            serialized_model.deserialize()
+        )
         self.sym_input: ca.MX = self._get_sym_input()
         self.prediction_function: ca.Function = self._build_prediction_function()
 

agentlib_mpc/modules/__init__.py

@@ -4,6 +4,7 @@
 
 It contains classes for local optimization and global coordination.
 """
+
 import importlib
 
 

agentlib_mpc/modules/data_source.py

@@ -71,7 +71,6 @@ def check_interpolation_method(cls, interpolation_method):
 
 
 class DataSource(BaseModule):
-
     config: DataSourceConfig
 
     def __init__(self, config: dict, agent: Agent):
@@ -81,8 +80,9 @@ def __init__(self, config: dict, agent: Agent):
 
         # Filter columns if specified
         if self.config.columns:
-            columns_to_keep = [col for col in self.config.columns if
-                               col in data.columns]
+            columns_to_keep = [
+                col for col in self.config.columns if col in data.columns
+            ]
             if not columns_to_keep:
                 raise ValueError("None of the specified columns exist in the dataframe")
             data = data[columns_to_keep]
@@ -112,7 +112,6 @@ def transform_index(self, data: pd.DataFrame) -> pd.DataFrame:
                 except ValueError:
                     raise ValueError("Unable to convert index to numeric format")
 
-
         data.index = data.index.astype(float) - offset
         return data
 

agentlib_mpc/modules/dmpc/admm/admm.py

@@ -1,4 +1,5 @@
 """Holds functionality for ADMM modules."""
+
 import time
 import threading
 from typing import List, Dict, Tuple, Iterable, Optional, TypeVar, Union
@@ -23,6 +24,7 @@
 import agentlib_mpc.data_structures.admm_datatypes as adt
 from agentlib_mpc.data_structures.mpc_datamodels import Results
 
+
 # noinspection PyArgumentList
 class ModuleStatus(Enum):
     not_started = auto()

agentlib_mpc/modules/dmpc/admm/admm_coordinated.py

@@ -1,5 +1,6 @@
 """Module implementing the coordinated ADMM module, which works together
 with a coordinator."""
+
 from collections import namedtuple
 from typing import Dict, Optional
 import pandas as pd
@@ -38,7 +39,6 @@ def __init__(self, *, config: dict, agent: Agent):
         self._result: Optional[pd.DataFrame] = None
 
     def process(self):
-
         # send registration request to coordinator
         timeout = self.config.registration_interval
 

agentlib_mpc/modules/dmpc/admm/admm_coordinator.py

@@ -1,6 +1,7 @@
 """
 Defines classes that coordinate an ADMM process.
 """
+
 import os
 import time
 from ast import literal_eval
@@ -129,11 +130,9 @@ def default_sampling_time(cls, samp_time, info: FieldValidationInfo):
 
 
 class ADMMCoordinator(Coordinator):
-
     config: ADMMCoordinatorConfig
 
     def __init__(self, *, config: dict, agent: Agent):
-
         if agent.env.config.rt:
             self.process = self._realtime_process
             self.registration_callback = self._real_time_registration_callback
@@ -414,7 +413,8 @@ def _check_convergence(self, iteration) -> bool:
         if self.config.use_relative_tolerances:
             # scaling factors for relative criterion
             primal_scaling = max(
-                np.linalg.norm(flat_locals), np.linalg.norm(flat_means)  # Ax  # Bz
+                np.linalg.norm(flat_locals),
+                np.linalg.norm(flat_means),  # Ax  # Bz
             )
             dual_scaling = np.linalg.norm(flat_multipliers)
             # compute tolerances for this iteration
@@ -494,7 +494,6 @@ def trigger_optimizations(self):
 
         # aggregate and send trajectories per agent
         for source, agent in active_agents:
-
             # collect mean and multiplier per coupling variable
             mean_trajectories = {}
             multipliers = {}

agentlib_mpc/modules/dmpc/coordinator.py

@@ -25,7 +25,6 @@
 
 
 class CoordinatorConfig(BaseModuleConfig):
-
     maxIter: int = Field(default=10, description="Maximum number of iterations")
     time_out_non_responders: float = Field(
         default=1, description="Maximum wait time for subsystems in seconds"

agentlib_mpc/modules/dmpc/employee.py

@@ -43,7 +43,6 @@ class MiniEmployeeConfig(BaseModuleConfig):
 
 
 class MiniEmployee(BaseModule):
-
     config: MiniEmployeeConfig
 
     def __init__(self, *, config: dict, agent: Agent):

agentlib_mpc/modules/ml_model_simulator.py

@@ -10,7 +10,6 @@
 
 
 class MLModelSimulatorConfig(SimulatorConfig):
-
     serialized_ml_models: AgentVariables = []
 
     @field_validator("t_sample")
@@ -27,7 +26,6 @@ def check_t_sample(cls, t_sample, info: FieldValidationInfo):
 
 
 class MLModelSimulator(Simulator):
-
     config: MLModelSimulatorConfig
     model: CasadiMLModel
 

agentlib_mpc/modules/ml_model_training/data_reduction.py

@@ -53,7 +53,6 @@ def reduce(
 
 
 if __name__ == "__main__":
-
     from sklearn.gaussian_process import GaussianProcessRegressor
     import matplotlib.pyplot as plt
 
@@ -65,7 +64,6 @@ def f(x):
         return x[:, 0] * 2 + x[:, 1] * x[:, 0] + x[:, 1] * 3 + 1 / x[:, 1] * x[:, 1]
 
     def get_score(x, y, x_test, y_test, message):
-
         gpr = GaussianProcessRegressor(RBF(), normalize_y=True)
         gpr.fit(x, y)
         s = gpr.score(x_test, y_test)

agentlib_mpc/modules/ml_model_training/setpoint_generator.py

@@ -1,5 +1,6 @@
 """Module which generates random set points within a comfort zone. Code heavily stolen
- from Max Berktold"""
+from Max Berktold"""
+
 import datetime
 import random
 
@@ -38,8 +39,7 @@ def __init__(self, config: dict, agent: Agent):
         lb, ub = self._bounds()
         self.current_target = random.uniform(lb, ub)
 
-    def register_callbacks(self):
-        ...
+    def register_callbacks(self): ...
 
     def process(self):
         while True:

agentlib_mpc/modules/mpc.py

@@ -90,6 +90,7 @@ def default_sampling_time(cls, samp_time, info: FieldValidationInfo):
             samp_time = info.data["time_step"]
         return samp_time
 
+
 def _create_optimization_backend(optimization_backend, agent_id):
     """Set up the optimization_backend"""
     optimization_backend = optimization_backend.copy()
@@ -278,9 +279,9 @@ def _init_optimization(self):
 
     def re_init_optimization(self, parameter: AgentVariable):
         """Re-initializes the optimization backend with new parameters."""
-        self.optimization_backend.discretization_options[
-            parameter.name
-        ] = parameter.value
+        self.optimization_backend.discretization_options[parameter.name] = (
+            parameter.value
+        )
         self._init_optimization()
 
     @property

agentlib_mpc/modules/mpc_full.py

@@ -1,4 +1,5 @@
 """Holds the class for full featured MPCs."""
+
 import numpy as np
 import pandas as pd
 from agentlib.core import AgentVariable

agentlib_mpc/optimization_backends/casadi_/admm.py

@@ -29,7 +29,6 @@ class CasadiADMMSystem(FullSystem):
     penalty_factor: OptimizationParameter
 
     def initialize(self, model: CasadiModel, var_ref: admm_datatypes.VariableReference):
-
         super().initialize(model=model, var_ref=var_ref)
 
         coup_names = [c.name for c in var_ref.couplings]
@@ -265,7 +264,7 @@ def _discretize(self, sys: CasadiADMMSystem):
             self.pred_time = ts * self.k
             xk = self.add_opt_var(sys.states)
             vars_dict[sys.states.name][self.k] = xk
-            self.add_constraint(xk-xk_end, gap_closing=True)
+            self.add_constraint(xk - xk_end, gap_closing=True)
 
             # add model constraints last due to fatrop
             self.add_constraint(
@@ -294,7 +293,9 @@ def _create_ode(
         integrator_ode = {"x": x, "p": p, "ode": ode}
         if integrator == Integrators.euler:
             xk_end = x + ode * opts["tf"]
-            opt_integrator = ca.Function("system", [x, p], [xk_end], ["x0", "p"], ["xf"])
+            opt_integrator = ca.Function(
+                "system", [x, p], [xk_end], ["x0", "p"], ["xf"]
+            )
         else:  # rk, cvodes
             opt_integrator = ca.integrator("system", integrator, integrator_ode, opts)
         return opt_integrator

agentlib_mpc/optimization_backends/casadi_/basic.py

@@ -43,7 +43,6 @@ class BaseSystem(System):
     ode: ca.MX
 
     def initialize(self, model: CasadiModel, var_ref: VariableReference):
-
         # define variables
         self.states = OptimizationVariable.declare(
             denotation="state",
@@ -171,7 +170,6 @@ def _discretize(self, sys: BaseSystem):
             for constraint in constraints:
                 self.add_constraint(*constraint)
 
-
     def _construct_stage_function(self, system: BaseSystem):
         """
         Combine information from the model and the var_ref to create CasADi
@@ -322,11 +320,13 @@ def _collocation_inner_loop(
             )
 
             constraints.append((ts * stage["ode"] - xp,))
-            constraints.append((
-                stage["model_constraints"],
-                stage["lb_model_constraints"],
-                stage["ub_model_constraints"],
-            ))
+            constraints.append(
+                (
+                    stage["model_constraints"],
+                    stage["lb_model_constraints"],
+                    stage["ub_model_constraints"],
+                )
+            )
 
             # Add contribution to the end state
             state_k_end = state_k_end + collocation.D[j] * state_collocation[j - 1]
@@ -458,7 +458,9 @@ def _create_ode(self, sys: BaseSystem, opts: dict, integrator: Integrators):
 
         if integrator == Integrators.euler:
             xk_end = x + ode * opts["tf"]
-            opt_integrator = ca.Function("system", [x, p], [xk_end], ["x0", "p"], ["xf"])
+            opt_integrator = ca.Function(
+                "system", [x, p], [xk_end], ["x0", "p"], ["xf"]
+            )
         else:  # rk, cvodes
             opt_integrator = ca.integrator("system", integrator, integrator_ode, opts)
 

agentlib_mpc/optimization_backends/casadi_/casadi_ml.py

@@ -73,14 +73,16 @@ def initialize(self, model: CasadiMLModel, var_ref: FullVariableReference):
             ref_list=var_ref.parameters,
         )
         self.initial_state = OptimizationParameter.declare(
-            denotation="initial_state",  # append the 0 as a convention to get initial guess
+            denotation="initial_state",
+            # append the 0 as a convention to get initial guess
             variables=model.get_states(var_ref.states),
             ref_list=var_ref.states,
             use_in_stage_function=False,
             assert_complete=True,
         )
         self.last_control = OptimizationParameter.declare(
-            denotation="initial_control",  # append the 0 as a convention to get initial guess
+            denotation="initial_control",
+            # append the 0 as a convention to get initial guess
             variables=model.get_inputs(var_ref.controls),
             ref_list=var_ref.controls,
             use_in_stage_function=False,
@@ -147,6 +149,11 @@ def _discretize(self, sys: CasadiMLSystem):
                 sys.states, lb=x_past, ub=x_past, guess=x_past
             )
             mx_dict[time][sys.initial_state.name] = x_past
+            z_past = self.add_opt_par(sys.algebraics)
+            mx_dict[time][sys.algebraics.name] = self.add_opt_var(
+                sys.algebraics, lb=z_past, ub=z_past, guess=z_past
+            )
+            mx_dict[time][sys.algebraics.name] = z_past
 
         # add past inputs
         for time in pre_grid_inputs:

agentlib_mpc/optimization_backends/casadi_/core/casadi_backend.py

@@ -118,7 +118,7 @@ def setup_optimization(self, var_ref: mpc_datamodels.VariableReference):
             bat_file=self.config.build_batch_bat,
             name=self.config.name,
             options=self.config.solver,
-            logger=self.logger
+            logger=self.logger,
         )
         self.discretization.initialize(
             system=self.system, solver_factory=solver_factory