# Differences with OpenAi Gym¶

## Changing reward and observations¶

Contrarily to OpenAI Gym where learning tasks are predefined, Ecole gives the user the tools to easily extend and customize environments. This is because the objective with Ecole is not only to provide a collection of challenges for machine learning, but really to solve combinatorial optimization problems more efficiently. If different data or tweaking the control task delivers better performances, it is an improvement! This is why Ecole let users change the environment reward and observation using RewardFunction and ObservationFunction.

## Parameter to reset¶

In OpenAI Gym, reset does not take parameters whereas is Ecole reset() takes a problem instance as a mandatory input. This is because when doing machine learning for optimization, there is no practical interest in solving the same problem over and over again. What is important is that the machine learning model is able to generalize to unseen problems. This is typically done by training on mutliple problem instances.

This setting is similar to multi-task reinforcement learning, where each problem instance is a task and one aims to generalize to unseen tasks. An alternative way to implement this is found in MetaWorld, where instead of passing the task as a parameter to reset, an supplementary set_task method is defined in the environment.

## Done on reset¶

In Ecole, reset() returns the same done flag as in step(). This is because nothing prevent an initial state from also being a terminal one. It is not only a theoretical consideration. For instance, in Branching, the initial state would typically be on the root node, prior to making the first branching decision. However, modern solvers have powerful presolvers, and it is not uncommon that the solution to the problem is found without needing to branch on any variable.

## Action set¶

Ecole defines an action set at every transition of the environment, while OpenAI Gym defines an action_space as a static variable of the environment. Ecole environments are more complex. For instance in Branching the set of valid actions changes, not only with every episode, but also with every transition! The action_set is required to make the next call to step(). We chose to add it as a return type to step() and reset() to emphasize this difference.

## Reward offset¶

In reset() a reward_offset is returned. This is not only a difference with OpenAI Gym, but also not part of the MDP formulation. It has not purpose for learning algorithms, rather it is meant for evaluating the complete solving procedure. RewardFunction are often designed so that their cumulative sum match a metric on the terminal state, such as solving time or number of LP iterations. This is because final metrics are often all that matter. However, for learning, a single reward on the terminal state is hard to learn from. It is then divided over all intermediate transitions in the episode.

Rather than providing a different mean of evaluating such metrics, we chose to reuse the environments to compute the cummulative sum, and therfore need the reward_offset for exactly matching the metric.

## No observation on terminal states¶

On terminal states, in OpenAI Gym as in Ecole, no further action can be taken and the environment need to be reset(). Most of the time, a terminal state in Ecole is a solved problem. This means that some complex observations cannot be extracted because they required information that simply do not exist. For instance NodeBipartite extract some information about the LP solution of the current branch-and-bound node. When the problem is solved, for example on a terminal state of the Branching environment, there is not a current node where this information could be extracted. For these reasons, one would find a None instead of an observation on terminal states.

In machine learning, the observation of a terminal state is actually never used. It is not given to a policy to take the next action (because there are not any), and hence never used for learning either. In OpenAI Gym, one can often get an observation on terminal states anyhow because the environment can easily extrapolate it. This can make the code easier to write, but is not required.