bpampel / bdld Goto Github PK

It might be good to reconsider the documentation or implementation of the "stride" and "write_stride" parameters of the trajectory action, as these behave a bit differently than for the other actions:

• The stride just specifies what (every Nth point) is written to the files, while always the full trajectory is stored in memory
• The write_stride specifies how often these selected points are written to the file (to not open the files and append a single line every time)

The reason for storing the full trajectories in memory was the usage of the Trajectory for the Histogram, where the full trajectories were typically desired.

Two solutions:

• Clarify the documentation
• Change the code, such that changing the stride actually influences the number/stride of particle positition stored in memory. This would mean that a stride !=1 results also in a different behavior of the Histogram

The latter is probably the better option, but requires a lot more work and might break historic input files.

At least document that you also need to do

git submodule init
git submodule update

to get the helpers module.

Alternative: restructure code, such that this is not a submodule.

When a stride N is specified for any actions, this will not process the action the first N times.
While this is fine for e.g. FES or DeltaF, this might not be the desired behavior for the trajectory action, if one wants to monitor the initial position.

Consider also writing the initial value to file for the trajectory action.

The "time" field is not given in the header.