Andrea Rizzi

Implement automatic setup of replicate calculations

1

I'm thinking of adding a `replicate` option in the YAML interface that will instruct the experiment builder to run multiple runs. It would look like something like this ```yaml experiments:...

Once the RMSD is released with the next version, we should reactivate that figure in the notebook.

There are a couple of further optimizations that can be done here. 1. The temperature of the state is saved separately from the system in the dictionary `{system, temperature, pressure}`...

We should add an option (`False` by default) to continue runs/setups that raised an exception.

Prevent users to specify a switch_experiment/phase_interval smaller than checkpoint_interval

1

When this happens the simulation never make progress because on the next cycle, it will resume from iteration 0 since no checkpoint was written before.

A reminder for me to add an article about good design practices.

[WIP] Convert restraint forces to inherit from CVCustomForce

2

These features were implemented originally in #365 and merged in #363, but we had to move them to a separate PR because we haven't resolved the performance issues that come...

Parallelize storage reading/writing

19

I'm planning to make some changes in the `Reporter` to split the two monolithic `netcdf` files into more manageable chunks. This is what I'm thinking now: - Solute trajectory: One...

In #437 we introduced the `surface_tension` attribute to `ThermodynamicState`. In `ThermodynamicState.__setstate__`, we should maintain backwards compatibility and use `serialization.get('surface_tension', None)` rather than `serialization['surface_tension']`.

For now, the serialization system is based on calling directly `__getstate__` and `__setstate__`. We should add `to_dict` and `from_dict` methods. Note that many objects have `__get/setstate__` methods that accept arguments...