Skip to content

Latest commit

 

History

History
1158 lines (1099 loc) · 81.9 KB

CHANGELOG.rst

File metadata and controls

1158 lines (1099 loc) · 81.9 KB
Raw

Change Log

[TODO]

  • [???] closer integration with gym especially the "register env", being able to create an env from a string etc.
  • [???] clean the notebook on RL
  • [???] use the typing module for type annotation.
  • [???] use some kind of "env.get_state()" when simulating instead of recoding everything "by hand"
  • [???] use "backend.get_action_to_set()" in simulate
  • [???] model better the voltage, include voltage constraints
  • [???] use the prod_p_forecasted and co in the "next_chronics" of simulate
  • [???] add a "_cst" or something in the const member of all the classes
  • [???] in deepcopy of env, make tests that the "pointers" are properly propagated in the attributes (for example envcpy._game_rules.legal_action should not be copied when building envcpy._helper_action_env)
  • [???] add multi agent
  • [???] make observation read only / immutable for all its properties (and not just for prod_p)
  • [???] better logging
  • [???] shunts in observation too, for real (but what to do when backend is not shunt compliant to prevent the stuff to break)
  • [???] model agent acting at different time frame
  • [???] model delay in observations
  • [???] model delay in action
  • [???] Code and test the "load from disk" method
  • [???] Make the redispatching data independent from the time step (eg instead of "in MW / step" have it in "MW / h") and have grid2op convert it to MW / step
  • [???] add a "plot action" method
  • [???] in MultiEnv, when some converter of the observations are used, have each child process to compute it in parallel and transfer the resulting data.
  • [???] "asynch" multienv
  • [???] properly model interconnecting powerlines

[1.8.0] - 2022-12-12

  • [BREAKING] now requires numpy >= 1.20 to work (otherwise there are issues with newer versions of pandas).
  • [BREAKING] issue #379 requires different behaviour depending on installed gym package.
  • [BREAKING] cooldowns are not consistent between env.step and obs.simulate. If obs.time_before_cooldown_line[l_id] > 0 it will be illegal, at the next call to env.step (and obs.simulate) to modify the status of this powerline l_id. Same for obs.time_before_cooldown_sub[s_id] > 0 if trying to modify topology of substation s_id. This also impacts the maintenances and hazards. This is also linked to github issue #148
  • [FIXED] a bug when using a Runner with an environment that has been copied (see #361)
  • [FIXED] issue #358
  • [FIXED] issue #363
  • [FIXED] issue #364
  • [FIXED] issue #365 and #376 . Now the function(s) gridobj.process_shunt_data and gridobj.process_grid2op_shunt_data are called gridobj.process_shunt_static_data
  • [FIXED] issue #367
  • [FIXED] issue #369
  • [FIXED] issue #374
  • [FIXED] issue #377 by adding a special method backend.update_thermal_limit_from_vect
  • [ADDED] the "packaging" python package is now required to install grid2op. It allows to support different gym versions that changes behavior regarding numpy pseudo random generator.
  • [ADDED] the function act.remove_line_status_from_topo to ignore the line status modification that would be induced by "set_bus" or "change_bus" when some cooldown applies on the powerline.
  • [IMPROVED] clarify documentation of gym compat module (see #372 and #373) as well as the doc for MultifolderWithCache (see #370)

[1.7.2] - 2022-07-05

  • [FIXED] seeding issue #331
  • [FIXED] clarify doc about fixed size matrices / graphs #330
  • [FIXED] improved the behaviour of obs._get_bus_id and obs._aux_fun_get_bus : when some objects were on busbar 2 they had a "wrong" bus id (it was lagged by 1) meaning an empty "bus" was introduced.
  • [FIXED] an issue with obs.state_of(...) when inspecting storage units (see #340)
  • [FIXED] an issue with act0 + act1 when curtailment was applied (see #340)
  • [FIXED] a slight "bug" in the formula to compute the redispatching cost for L2RPN 2022 competition.
  • [IMPROVED] possibility to pass the env variable _GRID2OP_FORCE_TEST to force the flag

    of "test=True" when creating an environment. This is especially useful when testing to prevent downloading of data.

  • [IMPROVED] support of "kwargs" backend arguments in MultiMixEnv see first item of version 1.7.1 below

[1.7.1] - 2022-06-03

  • [BREAKING] The possibility to propagate keyword arguments between the environment and the runner implied adding some arguments in the constructor of PandapowerBackend. So if you made a class that inherit from it, you should add these arguments in the constructor (otherwise you will not be able to use the runner) [This should not impact lot of codes, if any]
  • [FIXED] a documentation issue #281
  • [FIXED] a bug preventing to use the FromChronix2grid chronics class when there is an opponent on the grid.
  • [FIXED] a documentation issue #319 on notebook 11
  • [FIXED] some issues when the backend does not support shunts data (caused during the computation of the size of the observation) Tests are now performed in grid2op/tests/test_educpp_backend.py
  • [FIXED] a bug when downloading an environment when the archive name is not the same as the environment names (attempt to delete a non existing folder). This is the case for l2rpn_wcci_2022 env. For this env, your are forced to use grid2op version >= 1.7.1
  • [FIXED] an issue when converting a "done" action as a graph, see #327
  • [ADDED] score function for the L2RPN WCCI 2022 competition
  • [IMPROVED] adding the compatibility with logger in the reward functions.
  • [IMPROVED] when there is a game over caused by redispatching, the observation is not updated, as it is the case for other type of game over (improved consistency)
  • [IMPROVED] it is now possible to make an environment with a backend that cannot be copied.
  • [IMPROVED] the arguments used to create a backend can be (if used properly) re used (without copy !) when making a Runner from an environment for example.
  • [IMPROVED] description and definition of obs.curtailment_limit_effective are now consistent (issue #321)

[1.7.0] - 2022-04-29

  • [BREAKING] the L2RPNSandBoxScore, RedispReward and EconomicReward now properly computes the cost of the grid (there was an error between the conversion from MWh - cost is given in $ / MWh - and MW). This impacts also ScoreICAPS2021 and ScoreL2RPN2020.
  • [BREAKING] in the "gym_compat" module the curtailment action type has for dimension the number of dispatchable generators (as opposed to all generators before) this was mandatory to fix issue #282
  • [BREAKING] the size of the continuous action space for the redispatching in case of gym compatibility has also been adjusted to be consistent with curtailment. Before it has the size of env.n_gen now np.sum(env.gen_redispatchable).
  • [BREAKING] move the _ObsEnv module to Environment (was before in Observation).
  • [BREAKING] adding the curtailment_limit_effective in the observation converted to gym. This changes the sizes of the gym observation.
  • [FIXED] a bug preventing to use backend.update_from_obs when there are shunts on the grid for PandapowerBackend
  • [FIXED] a bug in the gym action space: see issue #281
  • [FIXED] a bug in the gym box action space: see issue #283
  • [FIXED] a bug when using MultifolderWithCache and Runner (see issue #285)
  • [FIXED] a bug in the env.train_val_split_random where sometimes some wrong chronics name were sampled.
  • [FIXED] the max value of the observation space is now 1.3 * pmax to account for the slack bus (it was 1.01 of pmax before and was not sufficient in some cases)
  • [FIXED] a proper exception is added to the "except" kwargs of the "info" return argument of env.step(...) (previously it was only a string) when redispatching was illegal.
  • [FIXED] a bug in env.train_val_split_random when some non chronics files where present in the "chronics" folder of the environment.
  • [FIXED] an error in the redispatching: in some cases, the environment detected that the redispatching was infeasible when it was not and in some others it did not detect when it while it was infeasible. This was mainly the case when curtailment and storage units were heavily modified.
  • [FIXED] now possible to create an environment with the FromNPY chronixcs even if the "chronics" folder is absent.
  • [FIXED] a bug preventing to converte observation as networkx graph with oldest version of numpy and newest version of scipy.
  • [FIXED] a bug when using max_iter and Runner in case of max_iter being larger than the number of steps in the environment and nb_episode >= 2.
  • [FIXED] a bug in the hashing of environment in case of storage units (the characteristics of the storage units were not taken into account in the hash).
  • [FIXED] a bug in the obs.as_dict() method.
  • [FIXED] a bug in when using the "env.generate_classe()" #310
  • [FIXED] another bug in when using the "env.generate_classe()" on windows #311
  • [ADDED] a function normalize_attr allowing to easily scale some data for the BoxGymObsSpace and BoxGymActSpace
  • [ADDED] support for distributed slack in pandapower (if supported)
  • [ADDED] an attribute self.infos for the BaseEnv that contains the "info" return value of env.step(...)
  • [ADDED] the possibility to shuffle the chronics of a GymEnv (the default behavior is now to shuffle them)
  • [ADDED] two attribtues for the observation: obs.gen_margin_up and obs.gen_margin_down
  • [ADDED] support for hashing chronix2grid related components.
  • [ADDED] possibility to change the type of the opponent space type from the make(...) command
  • [ADDED] a method to "limit the curtailment / storage" action depending on the availability of controllable generators (see act.limit_curtail_storage(...))
  • [ADDED] a class to generate data "on the fly" using chronix2grid (for now really slow and only available for a single environment)
  • [ADDED] a first version (for testing only) for the l2rpn_wcci_2022 environment.
  • [ADDED] a method to compute the "simple" line reconnection actions (adding 2 actions per lines instead of 5) in the action space (see act_space.get_all_unitary_line_set_simple())
  • [IMPROVED] better difference between env_path and grid_path in environments.
  • [IMPROVED] addition of a flag to control whether pandapower can use lightsim2grid (to solve the powerflows) or not
  • [IMPROVED] clean the warnings issued by pandas when used with pandapower
  • [IMPROVED] doc of observation module (some attributes were missing)
  • [IMPROVED] officially drop python 3.6 supports (which could not benefit from all the features)
  • [IMPROVED] add support for setting the maximum number of iteration in the PandaPowerBackend
  • [IMPROVED] when the curtailment / storage is too "strong" at a given step, the environment will now allow every controllable turned-on generators to mitigate it. This should increase the possibility to act on the curtailment and storage units without "breaking" the environment.
  • [IMPROVED] have dedicated type of actions / observation for L2RPN competition environments, defined in the "conf.py" file (to make possible the use of different grid2op version transparently)
  • [IMPROVED] on some cases, the routine used to compute the redispatching would lead to a "redispatch" that would change even if you don't apply any, for no obvious reasons. This has been adressed, though it's not perfect.
  • [IMPROVED] finer resolution when measuring exectution times

[1.6.5] - 2022-01-19

  • [BREAKING] the function "env.reset()" now reset the underlying pseudo random number generators of all the environment subclasses (eg. observation space, action space, etc.) This change has been made to ensure reproducibility between episodes: if env.seed(...) is called once, then regardless of what happens (basically the number of "env.step()" between calls to "env.reset()") the "env.reset()" will be generated with the same prng (drawn from the environment) This effect the opponent and the chronics (when maintenance are generated "on the fly").
  • [BREAKING] the name of the python files for the "Chronics" module are now lowercase (complient with PEP). If you did things like from grid2op.Chronics.ChangeNothing import ChangeNothing you need to change it like from grid2op.Chronics.changeNothing import ChangeNothing or even better, and this is the preferred way to include them: from grid2op.Chronics import ChangeNothing. It should not affect lots of code (more refactoring of the kind are to be expected in following versions).
  • [BREAKING] same as above for the "Observation" module. It should not affect lots of code (more refactoring of the kind are to be expected in following versions).
  • [FIXED] a bug for the EpisodeData that did not save the first observation when "add_detailed_output" was set to True and the data were not saved on disk.
  • [FIXED] an issue when copying the environment with the opponent (see issue #274)
  • [FIXED] a bug leading to the wrong "backend.get_action_to_set()" when there were storage units on the grid.
  • [FIXED] a bug in the "BackendConverter" when there are storage on the grid
  • [FIXED] issue #265
  • [FIXED] issue #261
  • [ADDED] possibility to "env.set_id" by giving only the folder of the chronics and not the whole path.
  • [ADDED] function "env.chronics_handler.available_chronics()" to return the list of available chronics for a given environment
  • [ADDED] possibility, through the Parameters class, to limit the number of possible calls to obs.simulate(...) see param.MAX_SIMULATE_PER_STEP and param.MAX_SIMULATE_PER_EPISODE (see issue #273)
  • [ADDED] a class to generate a "Chronics" readable by grid2op from numpy arrays (see #271)
  • [ADDED] an attribute delta_time in the observation that tells the time (in minutes) between two consecutive steps.
  • [ADDED] a method of the action space to show a list of actions to get back to the original topology (see #275) env.action_space.get_back_to_ref_state(obs)
  • [ADDED] a method of the action to store it in a grid2op independant fashion (using json and dictionaries), see act.as_serializable_dict()
  • [ADDED] possibility to generate a gym DiscreteActSpace from a given list of actions (see #277)
  • [ADDED] a class that output a noisy observation to the agent (see NoisyObservation): the agent sees the real values of the environment with some noise, this could used to model inacurate sensors.
  • [IMPROVED] observation now raises Grid2OpException instead of RuntimeError
  • [IMRPOVED] docs (and notebooks) for the "split_train_val" #269
  • [IMRPOVED] the "env.split_train_val(...)" function to also generate a test dataset see #276

[1.6.4] - 2021-11-08

  • [BREAKING] the name of the python files for the "agent" module are now lowercase (complient with PEP). If you did things like from grid2op.Agent.BaseAgent import BaseAgent you need to change it like from grid2op.Agent.baseAgent import BaseAgent or even better, and this is the preferred way to include them: from grid2op.Agent import BaseAgent. It should not affect lots of code.
  • [FIXED] a bug where the shunt had a voltage when disconnected using pandapower backend
  • [FIXED] a bug preventing to print the action space if some "part" of it had no size (empty action space)
  • [FIXED] a bug preventing to copy an action properly (especially for the alarm)
  • [FIXED] a bug that did not "close" the backend of the observation space when the environment was closed. This might be related to Issue#255
  • [ADDED] serialization of current_iter and max_iter in the observation.
  • [ADDED] the possibility to use the runner only on certain episode id (see runner.run(..., episode_id=[xxx, yyy, ...]))
  • [ADDED] a function that returns if an action has any change to modify the grid see act.can_affect_something()
  • [ADDED] a ttype of agent that performs predefined actions from a given list
  • [ADDED] basic support for logging in environment and runner (more coming soon)
  • [ADDED] possibility to make an environment with an implementation of a reward, instead of relying on a reward class.
  • [ADDED] a possible implementation of a N-1 reward
  • [IMPROVED] right time stamp is now set in the observation after the game over.
  • [IMPROVED] correct current number of steps when the observation is set to a game over state.
  • [IMPROVED] documentation to clearly state that the action_class should not be modified.
  • [IMPROVED] possibility to tell which chronics to use with the result of env.chronics_handler.get_id() (this is also compatible in the runner)
  • [IMPROVED] it is no more possible to call "env.reset()" or "env.step()" after an environment has been closed: a clean error is raised in this case.

[1.6.3] - 2021-08-21

  • [FIXED] a bug that allowed to use wrongly the function backend.get_action_to_set() even when the backend has diverged (which should not be possible)
  • [FIXED] a bug leading to non correct consideration of the status of powerlines right after the activation of some protections (see Issue#245 )
  • [IMPROVED] the PandaPowerBackend is now able to load a grid with a distributed slack bus. When loaded though, the said grid will be converted to one with a single slack bus (the first slack among the distributed)
  • [IMPROVED] massive speed-ups when copying environment or using obs.simulate (sometimes higher than 30x speed up)
  • [IMPROVED] experimental compatibility with different frameworks thanks to the possibility to serialize, as text files the class created "on the fly" (should solve most of the "pickle" error). See env.generate_classes() for an example usage. Every feedback is appreciated.

[1.6.2] (hotfix) - 2021-08-18

  • [FIXED] an issue when using obs.simulate with _AlarmScore (major bug)
  • [FIXED] now properly initialized the "complete_action_class" of the backend (minor bug)

[1.6.2] - 2021-07-27

  • [ADDED] the complete support for pickling grid2op classes. This is a major feature that allows to use grid2op way more easily with multiprocessing and to ensure compatibility with more recent version of some RL package (eg ray / rllib). Note that full compatibility with "multiprocessing" and "pickle" is not completely done yet.

[1.6.1] - 2021-07-27

  • [FIXED] a bug in the "env.get_path_env()" in case env was a multimix (it returned the path of the current mix instead of the path of the multimix environment)
  • [FIXED] a bug in the backend.get_action_to_set() and backend.update_from_obs() in case of disconnected shunt with backend that supported shunts (values for p and q were set even if the shunt was disconnected, which could lead to undefined behaviour)
  • [IMPROVED] now grid2op is able to check if an environment needs to be updated when calling grid2op.update_env() thanks to the use of registered hash values.
  • [IMPROVED] now grid2op will check if an update is available when an environment is being downloaded for the first time.

[1.6.0] (hotfix) - 2021-06-23

  • [FIXED] issue Issue#235 issue when using the "simulate" feature in case of divergence of powerflow.

[1.6.0] - 2021-06-22

  • [BREAKING] (but transparent for everyone): the disc_lines attribute is now part of the environment, and is also containing integer (representing the "order" on which the lines are disconnected due to protections) rather than just boolean.
  • [BREAKING] now the observation stores the information related to shunts by default. This means old logs computed with the runner might not work with this new version.
  • [BREAKING] the "Runner.py" file has been renamed, following pep convention "runner.py". You should rename your import from grid2op.Runner.Runner import Runner to from grid2op.Runner.runner import Runner (NB we higly recommend importing the Runner like from grid2op.Runner import Runner though !)
  • [FIXED]: the L2RPN_2020 score has been updated to reflect the score used during these competitions (there was an error between DoNothingAgent and RecoPowerlineAgent) [see Issue#228 ]
  • [FIXED]: some bugs in the action_space.get_all_unitary_redispatch and action_space.get_all_unitary_curtail
  • [FIXED]: some bugs in the GreedyAgent and TopologyGreedy
  • [FIXED]: Issue#220 flow_bus_matrix did not took into account disconnected powerlines, leading to impossibility to compute this matrix in some cases.
  • [FIXED]: Issue#223 : now able to plot a grid even if there is nothing controllable in grid2op present in it.
  • [FIXED]: an issue where the parameters would not be completely saved when saved in json format (alarm feature was absent) (related to Issue#224 )
  • [FIXED]: an error caused by the observation non being copied when a game over occurred that caused some issue in some cases (related to Issue#226 )
  • [FIXED]: a bug in the opponent space where the "previous_fail" kwargs was not updated properly and send wrongly to the opponent
  • [FIXED]: a bug in the geometric opponent when it did attack that failed.
  • [FIXED]: Issue#229 typo in the AlarmReward class when reset.
  • [ADDED] support for the "alarm operator" / "attention budget" feature
  • [ADDED] retrieval of the max_step (ie the maximum number of step that can be performed for the current episode) in the observation
  • [ADDED] some handy argument in the action_space.get_all_unitary_redispatch and action_space.get_all_unitary_curtail (see doc)
  • [ADDED] as utils function to compute the score used for the ICAPS 2021 competition (see from grid2op.utils import ScoreICAPS2021 and the associate documentation for more information)
  • [ADDED] a first version of the "l2rpn_icaps_2021" environment (accessible with grid2op.make("l2rpn_icaps_2021", test=True))
  • [IMPROVED] prevent the use of the same instance of a backend in different environments
  • [IMPROVED] Issue#217 : no more errors when trying to load a grid with unsupported elements (eg. 3w trafos or static generators) by PandaPowerBackend
  • [IMPROVED] Issue#215 : warnings are issued when elements present in pandapower grid will not be modified grid2op side.
  • [IMPROVED] Issue#214 : adding the shunt information in the observation documentation.
  • [IMPROVED] documentation to use the env.change_paramters function.

[1.5.2] - 2021-05-10

  • [BREAKING]: allow the opponent to chose the duration of its attack. This breaks the previous "Opponent.attack(...)" signature by adding an object in the return value. All code provided with grid2op are compatible with this new change. (for previously coded opponent, the only thing you have to do to make it compliant with the new interface is, in the opponent.attack(...) function return whatever_you_returned_before, None instead of simply whatever_you_returned_before)
  • [FIXED]: Issue#196 an issue related to the low / high of the observation if using the gym_compat module. Some more protections are enforced now.
  • [FIXED]: Issue#196 an issue related the scaling when negative numbers are used (in these cases low / max would be mixed up)
  • [FIXED]: an issue with the IncreasingFlatReward reward types
  • [FIXED]: a bug due to the conversion of int to float in the range of the BoxActionSpace for the gym_compat module
  • [FIXED]: a bug in the BoxGymActSpace, BoxGymObsSpace, MultiDiscreteActSpace and DiscreteActSpace where the order of the attribute for the conversion was encoded in a set. We enforced a sorted list now. We did not manage to find a bug caused by this issue, but it is definitely possible. This has been fixed now.
  • [FIXED]: a bug where, when an observation was set to a "game over" state, some of its attributes were below the maximum values allowed in the BoxGymObsSpace
  • [ADDED]: a reward EpisodeDurationReward that is always 0 unless at the end of an episode where it returns a float proportional to the number of step made from the beginning of the environment.
  • [ADDED]: in the Observation the possibility to retrieve the current number of steps
  • [ADDED]: easier function to manipulate the max number of iteration we want to perform directly from the environment
  • [ADDED]: function to retrieve the maximum duration of the current episode.
  • [ADDED]: a new kind of opponent that is able to attack at "more random" times with "more random" duration. See the GeometricOpponent.
  • [IMPROVED]: on windows at least, grid2op does not work with gym < 0.17.2 Checks are performed in order to make sure the installed open ai gym package meets this requirement (see issue Issue#185 )
  • [IMPROVED] the seed of openAI gym for composed action space (see issue openai/gym#2166): in waiting for an official fix, grid2op will use the solution proposed there openai/gym#2166 (comment) )

[1.5.1] - 2021-04-15

  • [FIXED]: Issue#194: (post release): change the name of the file platform.py that could be mixed with the python "platform" module to _glop_platform_info.py
  • [FIXED]: Issue #187: improve the computation and the documentation of the RedispReward. This has an impact on the env.reward_range of all environments using this reward, because the old "reward_max" was not correct.
  • [FIXED] Issue #181 : now environment can be created with a layout and a warning is issued in this case.
  • [FIXED] Issue #180 : it is now possible to set the thermal limit with a dictionary
  • [FIXED] a typo that would cause the attack to be discarded in the runner in some cases (cases for now not used)
  • [FIXED] an issue linked to the transformation into gym box space for some environments, this might be linked to Issue #185
  • [ADDED] a feature to retrieve the voltage angle (theta) in the backend (backend.get_theta) and in the observation.
  • [ADDED] support for multimix in the GymEnv (lack of support spotted thanks to Issue #185 )
  • [ADDED] basic documentation of the environment available.
  • [ADDED] Issue #166 : support for simulate in multi environment settings.
  • [IMPROVED] extra layer of security preventing modification of observation_space and action_space of environment
  • [IMPROVED] better handling of dynamically generated classes
  • [IMPROVED] the documentation of the opponent

[1.5.0] - 2021-03-31

  • [BREAKING] backend.check_kirchoff() method now returns also the discrepancy in the voltage magnitude and not only the error in the P and Q injected at each bus.
  • [BREAKING] the class method "to_dict" used to serialize the action_space and observation_space has been renamed cls_to_dict to avoid confusion with the to_dict method of action and observation (that stores, as dictionary the instance of the action / observation). It is now then possible to serialize the action class used and the observation class used as dictionary to (using action.cls_to_dict)
  • [BREAKING] for backend class implementation: need to upgrade your code to take into account the storage units if some are present in the grid even if you don't want to use storage units.
  • [BREAKING] the backend runpf method now returns a flag indicating if the simulation was successful AND (new) the exception in case there are some (it now returns a tuple). This change only affect new Backends.
  • [BREAKING] rename the attribute "parameters" of the "observation_space" to _simulate_parameters to avoid confusion with the parameters attributes of the environment.
  • [BREAKING] change of behaviour of the env.parameters attribute behaviour. It is no more possible to modified it with env.parameters = ... and the env.parameters.PARAM_ATTRIBUTE = xxx will have not effect at all. Use env.change_parameters(new_parameters) for changing the environment parameters and env.change_forecast_parameters(new_param_for_simulate) for changing the parameters used for simulate. (NB in both case you need to perform a "env.reset()" for the new parameters to be used. Any attempt to use an environment without a call to 'env.reset()' will lead to undefined behaviour).
  • [BREAKING] env.obs_space.rewardClass is not private and is called env.obs_space._reward_func. To change this function, you need to call env.change_reward(...)
  • [BREAKING] more consistency in the observation attribute names, they are now gen_p, gen_q and gen_v instead of prod_p, prod_q and prod_v (old names are still accessible for backward compatibility in the observation space) but conversion to json / dict will be affected as well as the converters (eg for gym compatibility)
  • [FIXED] Issue #164: reward is now properly computed at the end of an episode.
  • [FIXED] A bug where after running a Runner, the corresponding EpisodeData's CollectionWrapper where not properly updated, and did not contain any objects.
  • [FIXED] A bug when the opponent should chose an attack with all lines having flow 0, but one being still connected.
  • [FIXED] An error in the obs.flow_bus_matrix when active_flow=False and there were shunts on the powergrid.
  • [FIXED] obs.connectivity_matrix now properly takes into account when two objects are disconnected (before it was as if there were connected together)
  • [FIXED] some surprising behaviour when using obs.simulate just before or just after a planned maintenance operation.
  • [FIXED] a minimal bug in the env.copy method (the wrong simulated backend was used in the observation at right after the copy).
  • [FIXED] a bug in the serialization (as vector) of some action classes, namely: PowerlineSetAction and PowerlineSetAndDispatchAction and PowerlineChangeDispatchAndStorageAction
  • [FIXED] a bug preventing to use the obs.XXX_matrix() function twice
  • [FIXED] issue Issue #172: wrong assertion was made preventing the use of env.train_val_split_random()
  • [FIXED] issue Issue #173: a full nan vector could be converted to action or observation without any issue if it had the proper dimension. This was due to a conversion to integer from float.
  • [FIXED] an issue preventing to load the grid2op.utils submodule when installed not in "develop" mode
  • [FIXED] some issue with the multiprocessing of the runner on windows
  • [ADDED] more complete documentation for the runner.
  • [ADDED] a convenient function to evaluate the impact (especially on topology) of an action on a state (obs + act)
  • [ADDED] a property to retrieve the thermal limits from the observation.
  • [ADDED] documentation of the main elements of the grid and their "modeling" in grid2op.
  • [ADDED] parameters are now checked and refused if not valid (a RuntimeError is raised)
  • [ADDED] support for storage unit in grid2op (analog as a "load" convention positive: power absorbed from the grid, negative: power given to the grid having some energy limit and power limit). A new object if added in the substation.
  • [ADDED] Support for sparse matrices in obs.bus_connectivity_matrix
  • [ADDED] In the observation, it is now possible to retrieve the "active flow graph" (ie graph with edges having active flows, and nodes the active production / consumption) and "reactive flow graph" (see flow_bus_matrix)
  • [ADDED] more consistent behaviour when using the action space across the different type of actions. Now it should understand much more way to interact with it.
  • [ADDED] lots of action properties to manipulate action in a more pythonic way, for example using act.load_set_bus = ... instead of the previously way more verbose act.update({"set_bus": {"loads_id": ...}}) (this applies for load, gen, storage, line_or and line_ex and to set_bus and change_bus and also to storage_p and redispatch so making 12 "properties" in total)
  • [ADDED] an option to retrieve in memory the EpisodeData of each episode computed when using the runner. see runner.run(..., add_detailed_output=True)
  • [ADDED] the option as_csr_matrix in obs.connectivity_matrix function
  • [ADDED] convenient option to get the topology of a substation from an observation (obs.sub_topology(sub_id=...))
  • [ADDED] some basic tests for the environments shipped with grid2op.
  • [ADDED] grid2op now ships with the l2rpn_case14_sandbox environment
  • [ADDED] a function to list environments available for testing / illustration purpose.
  • [ADDED] a function of the observation to convert it to a networkx graph (obs.as_networkx())
  • [ADDED] support for curtailment feature in grid2op (curtailment on the renewable generator units).
  • [ADDED] better backward compatibility when reading data generated with previous grid2op version.
  • [IMPROVED] simplify the interface for the gym converter.
  • [IMPROVED] simplify the interface for the env.train_val_split and env.train_val_split_random
  • [IMPROVED] print of an action now limits the number of decimal for redispatching and storage units

[1.4.0] - 2020-12-10

  • [CHANGED] The parameters FORECAST_DC is now deprecated. Please use change_forecast_parameters(new_param) with new_param.ENV_DC=... instead.
  • [FIXED] and test the method backend.get_action_to_set
  • [FIXED] an error for the voltage of the shunt in the PandapowerBackend
  • [FIXED] PowerLineSet and PowerSetAndDispatch action were not properly converted to vector.
  • [ADDED] a method to set the state of a backend given a complete observation.
  • [ADDED] a utils module to store the data of some environment and be able to compute the scores (as in the neurips l2rpn competitions). This module might move at a different place in the future
  • [ADDED] a function to "split" an environment into train / validation using os.symlink
  • [ADDED] the implementation of + operator for action (based on previously available +=)
  • [ADDED] A more detailed documentation on the representation of the topology and how to create a backend
  • [ADDED] A easier way to set up the topology in backend (eg. get_loads_bus)
  • [ADDED] A easier way to set up the backend, with automatic computation of some attributes (eg. *_to_sub_pos, sub_info, dim_topo) if needed.
  • [ADDED] A function to change the parameters used by the environment (or obs_env) "on the fly" (has only impact AFTER env.reset is called) (see change_parameters and change_forecast_parameters)
  • [IMPROVED] PandaPowerBackend now should take less time to when reset.
  • [IMPROVED] some speed up in the grid2op computation

[1.3.1] - 2020-11-04

  • [FIXED] the environment "educ_case14_redisp"
  • [FIXED] notebooks are now working perfectly

[1.3.0] - 2020-11-02

  • [BREAKING] GymConverter has been moved to grid2op.gym_compat module instead of grid2op.Converter
  • [FIXED] wrong computation of voltage magnitude at extremity of powerlines when the powerlines were disconnected.
  • [FIXED] Issue #151: modification of observation attributes 3 could lead to crash
  • [FIXED] Issue #153: negative generator could happen in some cases
  • [FIXED] an error that lead to wrong normalization of some generator (due to slack bus) when using the gymconverter.
  • [FIXED] a bug that prevented runner to read back previously stored data (and now a test to check backward compatibility down to version 1.0.0)
  • [FIXED] small issue that could lead to non reproducibility when shuffling chronics
  • [FIXED] a bug in obs.bus_connectivity_matrix() when powerlines were disconnected
  • [ADDED] a class to deactivate the maintenance and hazards in the chronics from file GridStateFromFileWithForecastsWithoutMaintenance
  • [ADDED] a keyword argument in the matplotlib plot information on the grid (plot_helper.plot_info(..., coloring=...))
  • [ADDED] a function to change the color palette of powerlines (plot_helper.assign_line_palette)
  • [ADDED] a function to change the color palette of generators (plot_helper.assign_gen_palette)
  • [ADDED] Support the attack of the opponent in the EpisodeData class
  • [ADDED] Now the observations are set to a "game over" state when a game over occurred see BaseObservation.set_game_over
  • [ADDED] a method to plot the redispatching state of the grid PlotMatplot.plot_current_dispatch
  • [ADDED] the documentation of Episode module that was not displayed.
  • [IMPROVED] silence the warning issue when calling MultiEnv.get_seeds
  • [IMPROVED] the tolerance of the redispatching algorithm is now more consistent between the precision of the solver used and the time when it's
  • [IMPROVED] make faster and more robust the optimization routine used during redispatching
  • [IMPROVED] error message when the state fails because of infeasible redispatching

[1.2.3] - 2020-09-25

  • [ADDED] l2rpn-baselines package dependency in the "binder" environment.
  • [FIXED] binder integration that was broken momentarily
  • [FIXED] an issue in the sampling of redispatching action (ramp up and ramp down were inverted)
  • [FIXED] an issue causing errors when using action_space.change_bus and action_space.set_bus
  • [FIXED] an issue in the sampling: redispatching and "change_bus" where always performed at the same time
  • [FIXED] Issue #144: typo that could lead to not display some error messages in some cases.
  • [FIXED] Issue #146: awkward behaviour that lead to not calling the reward function when the episode was over.
  • [FIXED] Issue #147: un consistency between step and simulate when cooldowns where applied (rule checking was not using the right method).
  • [FIXED] An error preventing the loading of an Ambiguous Action (in case an agent took such action, the EpisodeData would not load it properly).
  • [IMPROVED] overall documentation of BaseEnv and Environment
  • [IMPROVED] rationalize the public and private part of the API for Environment and BaseEnv. Some members have been moved to private attribute (their modification would largely alterate the behaviour of grid2op).
  • [IMPROVED] internal functions are tagged as "Internal, do not use" in the documentation.
  • [IMPROVED] Improved documentation for the Environment and MultiMixEnvironment.

[1.2.2] - 2020-08-19

[1.2.1] - 2020-08-18

  • [ADDED] a function that allows to modify some parameters of the environment (see grid2op.update_env)
  • [ADDED] a class to convert between two backends
  • [FIXED] out dated documentation in some classes
  • [FIXED] Issue #140: illegal action were not properly computed in some cases, especially in case of divergence of the powerflow. Also now the "why" the action is illegal is displayed (instead of a generic "this action is illegal").
  • [FIXED] LightSim Issue #10: copy of whole environments without needing pickle module.
  • [UPDATED] a missing class documentation Chronics.Multifolder in that case.

[1.2.0] - 2020-08-03

  • [ADDED] ActionSpace.sample method is now implemented
  • [ADDED] DeltaRedispatchRandomAgent: that takes redispatching actions of a configurable [-delta;+delta] in MW on random generators.
  • [FIXED] Issue #129: game over count for env_actions
  • [FIXED] Issue #127: Removed no longer existing attribute docstring indisponibility
  • [FIXED] Issue #133: Missing positional argument space_prng in Action.SerializableActionSpace
  • [FIXED] Issue #131: Forecast values are accessible without needing to call obs.simulate beforehand.
  • [FIXED] Issue #134: Backend iadd actions with lines extremities disconnections (set -1)
  • [FIXED] issue Issue #125
  • [FIXED] issue Issue #126 Loading runner logs no longer checks environment actions ambiguity
  • [IMPROVED] issue Issue #16 improving openai gym integration.
  • [IMPROVED] Issue #134 lead us to review and rationalize the behavior of grid2op concerning the powerline status. Now it behave more rationally and has now the following behavior: if a powerline origin / extremity bus is "set" to -1 at one end and not modified at the other, it will disconnect this powerline, if a powerline origin / extremity bus is "set" to 1 or 2 at one end and not modified at the other, it will reconnect the powerline. If a powerline bus is "set" to -1 at one end and set to 1 or 2 at its other end the action is ambiguous.
  • [IMPROVED] way to count what is affect by an action (affect the cooldown of substation and powerline and the legality of some action). And action disconnect a powerline (using the "set_bus") will be considered to affect only this powerline (and not on its substations) if and only if the powerline was connected (otherwise it affects also on the substation). An action that connects a powerline (using the "set_bus") will affect only this powerline (and not its substations) if and only if this powerline was disconnected ( otherwise it affects the substations but not the powerline). Changing the bus of an extremity of a powerline if this powerline is connected has no impact on its status and therefor it considers it only affects the corresponding substation.
  • [IMPROVED] added documentation and usage example for CombineReward and CombineScaledReward

[1.1.1] - 2020-07-07

  • [FIXED] the EpisodeData now properly propagates the end of the episode
  • [FIXED] MultiFolder.split_and_save function did not use properly the "seed"
  • [FIXED] issue Issue 122
  • [FIXED] Loading of multimix environment when they are already present in the data cache.
  • [UPDATED] notebook 3 to reflect the change made a long time ago for the ambiguous action (when a powerline is reconnected)

[1.1.0] - 2020-07-03

  • [FIXED] forgot to print the name of the missing environment when error in creating it.
  • [FIXED] an issue in MultiFolder.sample_next_chronics that did not returns the right index
  • [FIXED] an issue that prevented the EpisodeData class to load back properly the action of the environment. This might have side effect if you used the obs.from_vect or act.from_vect in non conventional ways.
  • [ADDED] some documentation and example for the MultiProcessEnv
  • [IMPROVED] check that the sub environments are suitable grid2op.Environment.Environment in multiprocess env.
  • [FIXED] Minor documentation generation warnings and typos (Parameters, Backend, OpponentSpace, ActionSpace)

[1.0.0] - 2020-06-24

  • [BREAKING] MultiEnv has been renamed SingleEnvMultiProcess
  • [BREAKING] MultiEnv has been abstracted to BaseMultiProcessEnv and the backwards compatible interface is now SingleProcessMultiEnv
  • [BREAKING] the seeds parameters of the Runner.run function has been renamed env_seeds and an agent_seeds parameters is now available for fully reproducible experiments.
  • [FIXED] a weird effect on env.reset that did not reset the state of the previous observation held by the environment. This could have caused some issue in some corner cases.
  • [FIXED] BaseAction.__iadd__ fixed a bug with change actions += operator reported in Issue #116
  • [FIXED] obs.simulate post-initialized reward behaves like the environment
  • [FIXED] LinesReconnectedReward fixes reward inverted range
  • [FIXED] the get_all_unitary_topologies_change now counts only once the "do nothing" action.
  • [FIXED] obs.simulate could sometime returns "None" when the simulated action lead to a game over. This is no longer a problem.
  • [FIXED] grid2op.make will now raise an error if an invalid argument has been passed to it.
  • [FIXED] some arguments were not passed correctly to env.get_kwargs() or env.get_params_for_runner()
  • [ADDED] Issue #110 Adding an agent that is able to reconnect disconnected powerlines that can be reconnected, see grid2op.Agent.RecoPowerlineAgent
  • [ADDED] a clearer explanation between illegal and ambiguous action.
  • [ADDED] MultiEnvMultiProcess as a new multi-process class to run different environments in multiples prallel processes.
  • [ADDED] more control on the environment when using the grid2op.make function.
  • [ADDED] creation of the MultiMixEnv that allows to have, through a unified interface the possibility to interact alternatively with one environment or the other. This is especially useful when considering an agent that should interact in multiple environments.
  • [ADDED] possibility to use simulate on the current observation.
  • [ADDED] the overload of "__getattr__" for environment running in parallel
  • [ADDED] capability to change the powerlines on which the opponent attack at the environment initialization
  • [UPDATED] Backend.PandaPowerBackend.apply_action vectorized backend apply action method for speed.
  • [UPDATED] Issue #111 Converter is better documented to be more broadly usable.
  • [UPDATED] MultiEnv has been updated for new use case: Providing different environments configurations on the same grid and an arbitrary number of processes for each of these.
  • [UPDATED] Behaviour of "change_bus" and "set_bus": it is no more possible to affect the bus of a powerline disconnected.
  • [UPDATED] More control about the looping strategy of the ChronicsHandler that has been refactored, and can now be more easily cached (no need to do an expensive reading of the data at each call to env.reset)

[0.9.4] - 2020-06-12

  • [FIXED] Issue #114 the issue concerning the bug for the maintenance.

[0.9.3] - 2020-05-29

  • [FIXED] Issue #69 MultEnvironment is now working with windows based OS.
  • [ADDED] Issue #108 Seed is now part of the public agent API. The notebook has been updated accordingly.
  • [ADDED] Some function to disable the obs.simulate if wanted. This can lead to around 10~15% performance speed up in case obs.simulate is not used. See env.deactivate_forecast and env.reactivate_forecast (related to Issued #98)
  • [UPDATED] the first introductory notebook.
  • [UPDATED] possibility to reconnect / disconnect powerline giving its name when using reconnect_powerline and disconnect_powerline methods of the action space.
  • [UPDATED] Issue #105 problem solved for notebook 4. based OS.
  • [UPDATED] overall speed enhancement mostly in the VoltageControler, with the adding of the previous capability, some updates in the BackendAction Issued #98
  • [UPDATED] Added PlotMatplot constructor arguments to control display of names and IDs of the grid elements (gen, load, lines). As suggested in Issue #106

[0.9.2] - 2020-05-26

  • [FIXED] GridObject loading from file does initialize single values (bool, int, float) correctly instead of creating a np.array of size one.
  • [FIXED] IdToAct loading actions from file .npy
  • [FIXED] a problem on the grid name import on some version of pandas
  • [ADDED] a function that returns the types of the action see action.get_types()
  • [ADDED] a class to "cache" the data in memory instead of reading it over an over again from disk (see grid2op.chronics.MultifolderWithCache (related to Issued #98) )
  • [ADDED] improve the documentation of the observation class.
  • [UPDATED] Reward LinesReconnectedReward to take into account maintenances downtimes
  • [UPDATED] Adds an option to disable plotting load and generators names when using PlotMatplot

[0.9.1] - 2020-05-20

  • [FIXED] a bug preventing to save gif with episode replay when there has been a game over before starting time step
  • [FIXED] the issue of the random seed used in the environment for the runner.

[0.9.0] - 2020-05-19

  • [BREAKING] Issue #83: attributes name of the Parameters class are now more consistent with the rest of the package. Use NB_TIMESTEP_OVERFLOW_ALLOWED instead of NB_TIMESTEP_POWERFLOW_ALLOWED, NB_TIMESTEP_COOLDOWN_LINE instead of NB_TIMESTEP_LINE_STATUS_REMODIF and NB_TIMESTEP_COOLDOWN_SUB instead of NB_TIMESTEP_TOPOLOGY_REMODIF
  • [BREAKING] Issue #87: algorithm of the environment that solves the redispatching to make sure the environment meet the phyiscal constraints is now cast into an optimization routine that uses scipy.minimize to be solved. This has a few consequences: more dispatch actions are tolerated, computation time can be increased in some cases, when the optimization problem cannot be solved, a game over is thrown, scipy is now a direct dependency of grid2op, code base of grid2op is simpler.
  • [BREAKING] any attempt to use an un intialized environment (eg after a game over but before calling env.reset will now raise a Grid2OpException)
  • [FIXED] Issue #84: it is now possible to load multiple environments in the same python script and perform random action on each.
  • [FIXED] Issue #86: the proper symmetries are used to generate all the actions that can "change" the buses (SerializationActionSpace.get_all_unitary_topologies_change).
  • [FIXED] Issue #88: two flags are now used to tell the environment whether or not to activate the possibility to dispatch a turned on generator (forbid_dispatch_off) and whether or not to ignore the gen_min_uptimes and gen_min_downtime propertiers (ignore_min_up_down_times) that are initialized from the Parameters of the grid now.
  • [FIXED] Issue #89: pandapower backend should not be compatible with changing the bus of the generator representing the slack bus.
  • [FIXED] Greedy agents now uses the proper data types dt_float for the simulated reward (previously it was platform dependant)
  • [ADDED] A way to limit EpisodeReplay to a specific part of the episode. Two arguments have been added, namely: start_step and end_step that default to the full episode duration.
  • [ADDED] more flexibilities in IdToAct converter not to generate every action for both set and change for example. This class can also serialize and de serialize the list of all actions with the save method (to serialize) and the init_converter method (to read back the data).
  • [ADDED] a feature to have multiple difficulty levels per dataset.
  • [ADDED] a converter to transform prediction in connectivity of element into valid grid2op action. See Converter.ConnectivitiyConverter for more information.
  • [ADDED] a better control for the seeding strategy in Environment and MultiEnvironment to improve the reproducibility of the experiments.
  • [ADDED] a chronics class that is able to generate maintenance data "on the fly" instead of reading the from a file. This class is particularly handy to train agents with different kind of maintenance schedule.

[0.8.2] - 2020-05-13

  • [FIXED] Issue #75: PlotGrid displays double powerlines correctly.
  • [FIXED] Action += operator (aka. __iadd__) doesn't create warnings when manipulating identical arrays containing NaN values.
  • [FIXED] Issue #70: for powerline disconnected, now the voltage is properly set to 0.0
  • [UPDATED] Issue #40: now it is possible to retrieve the forecast of the injections without running an expensive "simulate" thanks to the obs.get_forecasted_inj method.
  • [UPDATED] Issue #78: parameters can be put as json in the folder of the environment.
  • [UPDATED] minor fix for env.make
  • [UPDATED] Challenge tensorflow dependency to tensorflow==2.2.0
  • [UPDATED] make documentation to reflect API changes of 0.8.0

[0.8.1] - 2020-05-05

  • [FIXED] Issue #65: now the length of the Episode Data is properly computed
  • [FIXED] Issue #66: runner is now compatible with multiprocessing again
  • [FIXED] Issue #67: L2RPNSandBoxReward is now properly computed
  • [FIXED] Serialization / de serialization of Parameters as json is now fixed

[0.8.0] - 2020-05-04

  • [BREAKING] All previously deprecated features have been removed
  • [BREAKING] grid2op.Runner is now located into a submodule folder
  • [BREAKING] merge of env.time_before_line_reconnectable into env.times_before_line_status_actionable which referred to the same idea: impossibility to reconnect a powerilne. Side effect observation have a different size now ( merging of obs.time_before_line_reconnectable into obs.time_before_cooldown_line). Size is now reduce of the number of powerlines of the grid.
  • [BREAKING] merge of act.vars_action into env.attr_list_vect which implemented the same concepts.
  • [BREAKING] the runner now save numpy compressed array to lower disk usage. Previous saved runner are not compatible.
  • [FIXED] grid2op.PlotGrid rounding error when casting from np.float32 to python.float
  • [FIXED] grid2op.BaseEnv.fast_forward_chronics Calls the correct methods and is now working properly
  • [FIXED] __iadd__ is now properly implemented for the action with proper care given to action types.
  • [UPDATED] MultiEnv now exchange only numpy arrays and not class objects.
  • [UPDATED] Notebooks are updated to reflect API improvements changes
  • [UPDATED] grid2op.make can now handle the download & caching of datasets
  • [UPDATED] Test/Sample datasets provide datetime related files .info
  • [UPDATED] Test/Sample datasets grid_layout.json
  • [UPDATED] grid2op.PlotGrid Color schemes and optional infos displaying
  • [UPDATED] grid2op.Episode.EpisodeReplay Improved gif output performance
  • [UPDATED] Action and Observation are now created without having to call init_grid(gridobject) which lead to small speed up and memory saving.

[0.7.1] - 2020-04-22

  • [FIXED] a bug in the chronics making it not start at the appropriate time step
  • [FIXED] a bug in "OneChangeThenNothing" agent that prevent it to be restarted properly.
  • [FIXED] a bug with the generated docker file that does not update to the last version of the package.
  • [FIXED] numpy, by default does not use the same datatype depending on the platform. We ensure that floating value are always np.float32 and integers are always np.int32
  • [ADDED] a method to extract only some part of a chronic.
  • [ADDED] a method to "fast forward" the chronics
  • [ADDED] class `grid2op.Reward.CombinedScaledReward`: A reward combiner with linear interpolation to stay within a given range.
  • [ADDED] `grid2op.Reward.BaseReward.set_range`: All rewards have a default setter for their reward_min and reward_max attributes.
  • [ADDED] `grid2op.PlotGrid`: Revamped plotting capabilities while keeping the interface we know from grid2op.Plot
  • [ADDED] grid2op.replay binary: This binary is installed with grid2op and allows to replay a runner log with visualization and gif export
  • [ADDED] a LicensesInformation file that put a link for all dependencies of the project.
  • [ADDED] make multiple dockers, one for testing, one for distribution with all extra, and one "light"
  • [UPDATED] test data and datasets are no longer included in the package distribution
  • [UPDATED] a new function make_new that will make obsolete the "grid2op.download" script in future versions
  • [UPDATED] the python "requests" package is now a dependency

[0.7.0] - 2020-04-15

  • [BREAKING] class grid2op.Environment.BasicEnv has been renamed BaseEnv for consistency. As this class should not be used outside of this code base, no backward compatibility has been enforced.
  • [BREAKING] class grid2op.Environment.ObsEnv has been renamed _ObsEnv to insist on its "privateness". As this class should not be used outside of this code base, no backward compatibility has been enforced.
  • [BREAKING] the "baselines" directory has been moved in another python package that will be released soon.
  • [DEPRECATION] grid2op.Action.TopoAndRedispAction is now grid2op.Action.TopologyAndDispatchAction.
  • [FIXED] Performances caveats regarding `grid2op.Backend.PandaPowerBackend.get_topo_vect`: Reduced the method running time and reduced number of direct calls to it.
  • [FIXED] Command line install scripts: Can now use grid2op.main and grid2op.download after installing the package
  • [FIXED] a bug that prevented to perform redispatching action if the sum of the action was neglectible (eg 1e-14) instead of an exact 0.
  • [FIXED] Manifest.ini and dockerfile to be complient with standard installation of a python package.
  • [ADDED] a notebook to better explain the plotting capabilities of grid2op (work in progrress)
  • [ADDED] grid2op.Backend.reset as a way for backends to implement a faster way to reload the grid. Implemented in grid2op.Backend.PandaPowerBackend
  • [ADDED] grid2op.Action.PowerlineChangeAndDispatchAction A subset of actions to limit the agents scope to 'switch line' and 'dispatch' operations only
  • [ADDED] grid2op.Action.PowerlineChangeAction A subset of actions to limit the agents scope to 'switch line' operations only
  • [ADDED] grid2op.Action.PowerlineSetAndDispatchAction A subset of actions to limit the agents scope to 'set line' and 'dispatch' operations only
  • [ADDED] grid2op.Action.PowerlineSetAction A subset of actions to limit the agents scope to 'set line' operations only
  • [ADDED] grid2op.Action.TopologySetAction A subset of actions to limit the agents scope to 'set' operations only
  • [ADDED] grid2op.Action.TopologySetAndDispatchAction A subset of actions to limit the agents scope to 'set' and 'redisp' operations only
  • [ADDED] grid2op.Action.TopologyChangeAction A subset of actions to limit the agents scope to 'change' operations only
  • [ADDED] grid2op.Action.TopologyChangeAndDispatchAction A subset of actions to limit the agents scope to 'change' and 'redisp' operations only
  • [ADDED] grid2op.Action.DispatchAction A subset of actions to limit the agents scope to 'redisp' operations only
  • [ADDED] a new method to plot other values that the default one for plotplotly.
  • [ADDED] a better plotting utilities that is now consistent with PlotPlotly, PlotMatplotlib and PlotPyGame
  • [ADDED] a class to replay a logger using PlotPyGame class (grid2op.Plot.EpisodeReplay)
  • [ADDED] a method to parse back the observations with lower memory footprint and faster, when the observations are serialized into a numpy array by the runner, and only some attributes are necessary.
  • [ADDED] fast implementation of "replay" using PlotPygame and EpisodeData
  • [UPDATED] overall documentation: more simple theme, easier organization of each section.

[0.6.1] - 2020-04-??

  • [FIXED] Issue #54: Setting the bus for disconnected lines no longer counts as a substation operation.
  • [FIXED] if no redispatch actions are taken, then the game can no more invalid a provided action due to error in the redispatching. This behavior was caused by increase / decrease of the system losses that was higher (in absolute value) than the ramp of the generators connected to the slack bus. This has been fixed by removing the losses of the powergrid in the computation of the redispatching algorithm. side effect for the generator connected to the slack bus, the ramp min / up as well as pmin / pmax might not be respected in the results data provided in the observation for example.
  • [FIXED] a bug in the computation of cascading failure that lead (sometimes) to diverging powerflow when in the fact the powerflow did not diverge.
  • [FIXED] a bug in the OneChangeThenNothing agent.
  • [FIXED] a bug that lead to impossibility to load a powerline after a cascading failure in some cases. Now fixed by resetting the appropriate vectors when calling "env.reset".
  • [FIXED] function env.attach_render that uses old names for the grid layout
  • [ADDED] Remember last line buses: Reconnecting a line without providing buses will reconnect it to the buses it was previously connected to (origin and extremity).
  • [ADDED] Change lines status (aka. switch_line_status) unitary actions for subclasses of AgentWithConverter.
  • [ADDED] Dispatching unitary actions for subclasses of AgentWithConverter.
  • [ADDED] CombinedReward. A reward combiner to compute a weighted sum of other rewards.
  • [ADDED] CloseToOverflowReward. A reward that penalize agents when lines have almost reached max capacity.
  • [ADDED] DistanceReward. A reward based on how far way from the original topology the current grid is.
  • [ADDED] BridgeReward. A reward based on graph connectivity, see implementation in grid2op.Reward.BridgeReward for details

[0.6.0] - 2020-04-03

  • [BREAKING] grid2op.GameRules module renamed to grid2op.RulesChecker
  • [BREAKING] grid2op.Converters module renamed grid2op.Converter
  • [BREAKING] grid2op.ChronicsHandler renamed to grid2op.Chronics
  • [BREAKING] grid2op.PandaPowerBackend is moved to grid2op.Backend.PandaPowerBackend
  • [BREAKING] RulesChecker.Allwayslegal is now Rules.Alwayslegal
  • [BREAKING] Plotting utils are now located in their own module grid2op.Plot
  • [DEPRECATION] HelperAction is now called ActionSpace to better suit open ai gym name. Use of HelperAction will be deprecated in future versions.
  • [DEPRECATION] ObservationHelper is now called ObservationSpace to better suit open ai gym name. Use of ObservationHelper will be deprecated in future versions.
  • [DEPRECATION] Action class has been split into BaseAction that serve as an abstract base class for all action class, and CompleteAction (that inherit from BaseAction) for the class allowing to perform every modification implemented in grid2op.
  • [DEPRECATION] Observation class has renamed BaseObservation that serve as an abstract base class for all observation classes. Name Observation will be deprecated in future versions.
  • [DEPRECATION] Agent class has renamed BaseAgent that serve as an abstract base class for all agent classes. Name Agent will be deprecated in future versions.
  • [DEPRECATION] Reward class has renamed BaseReward that serve as an abstract base class for all reward classes. Name Reward will be deprecated in future versions.
  • [DEPRECATION] LegalAction class has renamed BaseRules that serve as an abstract base class for all type of rules classes. Name LegalAction will be deprecated in future versions.
  • [DEPRECATION] typo fixed in PreventReconection class (now properly named PreventReconnection)
  • [ADDED] different kind of "Opponent" can now be implemented if needed (missing deep testing, different type of class, and good documentation)
  • [ADDED] implement other "rewards" to look at. It is now possible to have an environment that will compute more rewards that are given to the agent through the "information" return argument of env.step. See the documentation of Environment.other_rewards.
  • [ADDED] Alternative method to load datasets based on new dataset format: MakeEnv.make2
  • [ADDED] Layout of the powergrid is part of the GridObject and is serialized along with the action_space and observation_space. Plotting utilities no longer require specific layout (custom layout can still be provided)
  • [ADDED] A new kind of actions that can change the value (and buses) to which shunt are connected. This support will be helpfull for the VoltageControler class.
  • [FIXED] Loading L2RPN_2019 dataset
  • [FIXED] a bug that prevents the voltage controler to be changed when using grid2op.make.
  • [FIXED] time_before_cooldown_line vector were output twice in observation space (see issue 47 part 1)
  • [FIXED] the number of active bus on a substation was not computed properly, which lead to some unexpected behavior regarding the powerlines switches (depending on current stats of powerline, changing the buses of some powerline has different effect) (see issue 47 part 2)
  • [FIXED] wrong voltages were reported for PandapowerBackend that causes some isolated load to be not detected (see issue 51 )
  • [FIXED] improve the install script to not crash when numba can be installed, but cannot be loaded. (see issue 50 )
  • [UPDATED] import documentation of Space especially in case someone wants to build other type of Backend

[0.5.8] - 2020-03-20

  • [ADDED] runner now is able to show a progress bar
  • [ADDED] add a "max_iter" in the runner.
  • [ADDED] a repository in this github for the baseline (work in progress)
  • [ADDED] include grid2Viz in a notebook (the notebook "StudyYourAgent")
  • [ADDED] when a file is not present in the chronics, the chronics_handler behaves as if nothing changes. If no files at all are provided, it raises an error.
  • [ADDED] possibility to change the controler for the generator voltage setpoints (See VoltageControler for more information). It can be customized as of now.
  • [ADDED] lots of new tests for majority of classes (ChronicsHandler, BaseAction, Observations etc.)
  • [FIXED] voltages are now set to 0 when the powerline are disconnected, instead of being set to Nan in pandapower backend.
  • [FIXED] ReadPypowNetData does not crash when argument "chunk_size" is provided now.
  • [FIXED] some typos in the Readme
  • [FIXED] some redispatching declared illegal but are in fact legal (due to a wrong assessment) (see issue 44)
  • [FIXED] reconnecting a powerline now does not count the mandatory actions on both its ends (previously you could not reconnect a powerline with the L2RPN 2019 rules because it required acting on 2 substations) as "substation action"
  • [UPDATED] add a blank environment for easier use.
  • [UPDATED] now raise an error if the substations layout does not match the number of substations on the powergrid.
  • [UPDATED] better handling of system without numba issue 42)
  • [UPDATED] better display of the error message if all dispatchable generators are set issue 39
  • [UPDATED] change the link to the doc in the notebook to point to readthedoc and not to local documentation.
  • [UPDATED] Simulate action behavior result is the same as stepping given perfect forecasts at t+1

[0.5.7] - 2020-03-03

  • [ADDED] a new environment with consistant voltages based on the case14 grid of pandapower (case14_relistic)
  • [ADDED] a function to get the name on the element of the graphical representation.
  • [ADDED] a new class to (PlotMatPlotlib) to display the grid layout and the position of the element, as well as their name and ID
  • [ADDED] possibility to read by chunk the data (memory efficiency and huge speed up at the beginning of training) (issue 21)
  • [ADDED] improved method to limit the episode length in chronics handler.
  • [ADDED] a method to project some data on the layout of the grid (GetLayout.plot_info)
  • [FIXED] a bug in the simulated reward (it was not initialized properly)
  • [FIXED] add the "prod_charac.csv" for the test environment case14_test, case14_redisp, case14_realistic and case5_example
  • [FIXED] fix the display bug in the notebook of the l2rpn starting kit with the layout of the 2 buses
  • [UPDATED] now attaching the layout metadata directly into the environment
  • [UPDATED] obs.simulate now has the same code as env.step this include the same signature and the possibility to simulate redispatching actions as well.
  • [UPDATED] Notebook 6 to train agent more efficiently (example: prediction of actions in batch)
  • [UPDATED] PlotGraph to derive from GridObjects allowing to be inialized at creation and not when first observation is loaded (usable without observation)
  • [UPDATED] new default environment (case14_realistic)
  • [UPDATED] data for the new created environment.
  • [UPDATED] implement redispatching action in obs.simulate
  • [UPDATED] refactoring Environment and ObsEnv to inherit from the same base class.

[0.5.6] - 2020-02-25

  • [ADDED] Notebook 6 to explain multi environment
  • [ADDED] more type of agents in the notebook 3
  • [FIXED] Environment now properly built in MultiEnvironment
  • [FIXED] Notebook 3 to now work with both neural network
  • [FIXED] remove the "print" that displayed the path of the data used in MultiEnvironment
  • [UPDATED] the action space for "IdToAct" now reduces the number of possible actions to only actions that don't directly cause a game over.

[0.5.5] - 2020-02-14

  • [ADDED] a easier way to set the thermal limits directly from the environment (env.set_thermal_limit)
  • [ADDED] a new environment with redispatching capabilities (case14_redisp) including data
  • [ADDED] a new convenient script to download the dataset, run python3 -m grid2op.download --name "case14_redisp" from the command line.
  • [ADDED] new rewards to better take into account redispatching (EconomicReward and RedispReward)
  • [ADDED] a method to check if an action is ambiguous (act.is_ambiguous())
  • [ADDED] a method to set more efficiently the id of the chronics used in the environment (env.set_id)
  • [ADDED] env.step now propagate the error in "info" output (but not yet in obs.simulate)
  • [ADDED] notebooks for redispatching (see getting_started/5_RedispacthingAgent.ipynb)
  • [ADDED] now able to initialize a runner from a valid environment (see env.get_params_for_runner)
  • [FIXED] reconnecting too soon a powerline is now forbidden in l2rpn2019 (added the proper legal action)
  • [UPDATED] more information in the error when plotly and seaborn are not installed and trying to load the graph of the grid.
  • [UPDATED] setting an object to a busbar higher (or equal) than 2 now leads to an ambiguous action.
  • [UPDATED] gitignore to really download the "prod_charac.csv" file
  • [UPDATED] private member in action space and observation space (_template_act and _empty_obs) to make it clear it's not part of the public API.
  • [UPDATED] change default environment to case14_redisp
  • [UPDATED] notebook 2 now explicitely says the proposed action is ambiguous in a python cell code (and not just in the comments) see issue (issue 27)

[0.5.4] - 2020-02-06

  • [ADDED] better handling of serialization of scenarios.

[0.5.3] - 2020-02-05

  • [ADDED] parrallel processing of the environment: evaluation in parrallel of the same agent in different environments.
  • [ADDED] a way to shuffle the order in which different chronics are read from the hard drive (see MultiFolder.shuffle)
  • [FIXED] utility script to push docker file
  • [FIXED] some tests were not passed on the main file, because of a file ignore by git.
  • [FIXED] improve stability of pandapower backend.
  • [UPDATED] avoid copying the grid to build observation

[0.5.2] - 2020-01-27

  • [ADDED] Adding a utility to retrieve the starting kit L2RPN 2019 competition.
  • [ADDED] Layout of the powergrid graph of the substations for both the 5bus_example and the CASE_14_L2RPN2019.
  • [FIXED] Runner skipped half the episode in some cases (sequential, even number of scenarios). Now fixed.
  • [FIXED] Some typos on the notebook "getting_started4-StudyYourAgent.ipynb".
  • [FIXED] Error in the conversion of observation to dictionnary. Twice the same keys were used ('time_next_maintenance') for both time_next_maintenance and duration_next_maintenance.
  • [UPDATED] The first chronics that is processed by a runner is not the "first" one on the hardrive (if sorted in alphabetical order)
  • [UPDATED] Better layout of substation layout (in case of multiple nodes) in PlotGraph

[0.5.1] - 2020-01-24

  • [ADDED] extra tag 'all' to install all optional dependencies.
  • [FIXED] issue in the documentation of BaseObservation, voltages are given in kV and not V.
  • [FIXED] a bug in the runner that prevented the right chronics to be read, and output wrong names
  • [FIXED] a bug preventing import if plotting packages where not installed, that causes the documentation to crash.

[0.5.0] - 2020-01-23

  • [BREAKING] BaseAction/Backend has been modified with the implementation of redispatching. If you used a custom backend, you'll have to implement the "redispatching" part.
  • [BREAKING] with the introduction of redispatching, old action space and observation space, stored as json for example, will not be usable: action size and observation size have been modified.
  • [ADDED] A converter class that allows to pre-process observation, and post-process action when given to an BaseAgent. This allows for more flexibility in the action_space and observation_space.
  • [ADDED] Adding another example notebook getting_started/Example_5bus.ipynb
  • [ADDED] Adding another renderer for the live environment.
  • [ADDED] Redispatching possibility for the environment
  • [ADDED] More complete documentation of the representation of the powergrid (see documentation of Space)
  • [FIXED] A bug in the conversion from pair unit to kv in pandapower backend. Adding some tests for that too.
  • [UPDATED] More complete documentation of the BaseAction class (with some examples)
  • [UPDATED] More unit test for observations
  • [UPDATED] Remove the TODO's already coded
  • [UPDATED] GridStateFromFile can now read the starting date and the time interval of the chronics.
  • [UPDATED] Documentation of BaseObservation: adding the units (issue 22)
  • [UPDATED] Notebook getting_started/4_StudyYourAgent.ipynb to use the converter now (much shorter and clearer)

[0.4.3] - 2020-01-20

  • [FIXED] Bug in L2RPN2019 settings, that had not been modified after the changes of version 0.4.2.

[0.4.2] - 2020-01-08

  • [BREAKING] previous saved BaseAction Spaces and BaseObservation Spaces (as dictionnary) are no more compatible
  • [BREAKING] renaming of attributes describing the powergrid across classes for better consistency:
Class Name Old Attribute Name New Attribute Name
Backend

n_lines

n_line
Backend

n_generators

n_gen
Backend

n_loads

n_load
Backend

n_substations

n_sub
Backend

subs_elements

sub_info
Backend

name_loads

name_load
Backend

name_prods

name_gen
Backend

name_lines

name_line
Backend

name_subs

name_sub
Backend

lines_or_to_subid

line_or_to_subid
Backend

lines_ex_to_subid

line_ex_to_subid
Backend

lines_or_to_sub_pos

line_or_to_sub_pos
Backend

lines_ex_to_sub_pos

line_ex_to_sub_pos
Backend

lines_or_pos_topo_vect

line_or_pos_topo_vect
Backend

lines_ex_pos_topo_vect

lines_ex_pos_topo_vect
BaseAction / BaseObservation _lines_or_to_subid line_or_to_subid
BaseAction / BaseObservation _lines_ex_to_subid line_ex_to_subid
BaseAction / BaseObservation _lines_or_to_sub_pos line_or_to_sub_pos
BaseAction / BaseObservation _lines_ex_to_sub_pos line_ex_to_sub_pos
BaseAction / BaseObservation _lines_or_pos_topo_vect line_or_pos_topo_vect
BaseAction / BaseObservation _lines_ex_pos_topo_vect lines_ex_pos_topo_vect
GridValue n_lines n_line
  • [FIXED] Runner cannot save properly action and observation (sizes are not computed properly) now fixed and unit test added
  • [FIXED] Plot utility has a bug in extracting grid information. now fixed
  • [FIXED] gym compatibility issue for environment
  • [FIXED] checking key-word arguments in "make" function: if an invalid argument is provided, it now raises an error.
  • [UPDATED] multiple random generator streams for observations
  • [UPDATED] Refactoring of the BaseAction and BaseObservation Space. They now both inherit from "Space"
  • [UPDATED] the getting_started notebooks to reflect these changes

[0.4.1] - 2019-12-17

  • [FIXED] Bug#14 : Nan in the observation space after switching one powerline [PandaPowerBackend]
  • [UPDATED] plot now improved for buses in substations

[0.4.0] - 2019-12-04

  • [ADDED] Basic tools for plotting with the PlotPlotly module
  • [ADDED] support of maintenance operation as well as hazards in the BaseObservation (and appropriated tests)
  • [ADDED] support for maintenance operation in the Environment (read from the chronics)
  • [ADDED] example of chronics with hazards and maintenance
  • [UPDATED] handling of the AmbiguousAction and IllegalAction exceptions (and appropriated tests)
  • [UPDATED] various documentation, in particular the class BaseObservation
  • [UPDATED] information retrievable BaseObservation.state_of

[0.3.6] - 2019-12-01

  • [ADDED] functionality to restrict action based on previous actions (impacts Environment, RulesChecker and Parameters)
  • [ADDED] tests for the notebooks in getting_started
  • [UPDATED] readme to properly show the docker capability
  • [UPDATED] Readme with docker

[0.3.5] - 2019-11-28

  • [ADDED] serialization of the environment modifications
  • [ADDED] the changelog file
  • [ADDED] serialization of hazards and maintenance in actions (if any)
  • [FIXED] error messages in grid2op.GridValue.check_validity
  • [UPDATED] notebook getting_started/4_StudyYourAgent.ipynb to reflect these changes