Restarting from Second to Last Iteration and MAB Questions

[shelby case]

Hello, 

I got the following h5-related error from my latest WESTPA run; please see the attached file. The run completed normally, but upon restarting for more iterations, it gave this error. Is there a way to recover my h5 file or start from the second to last iteration? 

Separately, I have been using the following west.cfg file for my MAB WESTPA run, but it doesn't appear to be sampling the space I hoped it would; please see attachments. I am still new to using this feature. Is there something I'm missing to make it work as intended? 

Thank you, 

Shelby Case 

[Hayden Scheiber]

Hey Shelby,

I typically see that HDF5 error when a process has opened an HDF5 file with write access and has not yet released it. For example, if a weighted ensemble simulation is currently running with that HDF5 file.

Anytime a file is open for write access, the API "locks" the file to prevent write access from another process. This is done to prevent file corruption. Only one compute process can access an HDF5 file at a time with write access. In the future, ensure that all processes that access your HDF5 file close correctly, which would normally release the file lock. 

If you have a crashed or zombie process that didn't properly release your HDF5 file, it may be stuck locked. A trick to get around the lock is to copy your HDF5 file, delete the initial copy, then rename your copied version back to the original name. See this discussion for more info.

As for your MAB issue, please be more specific as to your expectations. I can only guess as to what your expectations are.
Currently you have configured MAB to place 20 equally spaced adaptive bins among walkers with pcoords between -inf and 39 Angstroms. The end points of your MAB scheme will adapt to the minimal and maximal walkers within this range. 

This is defined by your MAB outer bin. Notice that you have

          boundaries:
            - [-inf, 39., 67., inf]
        mappers:
          - type: MABBinMapper
            nbins: [20]
            at: [0]
The "at: [0]" implies you will place your MAB binning scheme in the outer bin that contains this point, i.e. the rectilinear bin with boundaries at -inf and 39 Angstroms.

I assume you intended to sample the space of your progress coordinate greater than 39 Angstroms? 

Lets say you don't want to limit your MAB scheme with an upper boundary of 39 Angstroms. All you would have to do is change your outer bin boundaries to
          boundaries:
            - [-inf, inf]

This would create a single outer bin from -inf to +inf, so your inner MAB binning scheme could vary without constraint.

If, on the other hand, you wanted to sample the outer bin from 39 to 67 Angstroms, you would have to change your mab "at" directive to:

at: [53]

I just chose 53 as the midpoint, but any number greater than 39 and less than 67 would work. You could also create multiple MAB schemes at different outer bins.

Hope this helps,

Hayden

[Hayden Scheiber]

I forgot to mention if your west.h5 file is corrupted, you can use "w_truncate -n X" (this is a westpa CLI tool) to truncate your westpa simulation to iteration X-1 (i.e. this erases data in your west.h5 file for iterations X and beyond). Note this does not delete your per-iteration HDF5 files (if you are using them), i.e. you will need to manually truncate data within $WEST_SIM_ROOT/traj_segs/ and $WEST_SIM_ROOT/seg_logs/.

Cheers,

Hayden

[Leung, Jeremy]

Thanks for answering this so comprehensively, Hayden!

I'd also like to add that if you use `w_truncate` without the `-n` option (or provide 0), it'll automatically truncate only the last iteration. :)

Best,

Jeremy L.

---
Jeremy M. G. Leung, PhD
Postdoctoral Associate, Chemistry (Chong Lab)
University of Pittsburgh | 219 Parkman Avenue, Pittsburgh, PA 15260
jml...@pitt.edu | [He, Him, His]

-- 
You received this message because you are subscribed to the Google Groups "westpa-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to westpa-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/westpa-users/1df54463-5eff-49ac-bdfc-80b972eeb949n%40googlegroups.com.

[shelby case]

Thank you, Hayden and Jeremy! This is very helpful!

Best,

Shelby