Any Sollution For Fizzled workflow in intermediate state at fireworks?


#1

Hi, my workflow stooped after GGAU_optimize_structure_(2x)…it was fizzled at VASP_db_insertion…what can i do now ?


#2

Are you referring to a workflow that is linked to in your user dashboard after submitting a structure via e.g. our Crystal Toolkit app? If so, let us know the workflow ID. If you are running a workflow yourself, please consult / post to the matproj-develop Google group.


#3

Dear @Suman_Gandi

if your workflow stopped during the GGAU_optimize_structure_(2x) step, it usually means that the structure (.cif or POSCAR) provided was unable to converge to a minimum energy structure. For such cases, I have found that resubmitting the structure after changing the lattice parameter, i.e. increasing the lattice parameter on one direction so that the muffin-tin radius of two ions don’t overlap, or other similar corrections, will most probably solve the problem.

However, if your structure is a very large supercell, I think the runtime exceeds a certain (default) computational wall-time after which your job will be defused by the workflow manager. [I am not completely sure of how this works, but if you check the fireworks workflow outputs, these details can be seen as to why your submission was fizzled].

Similarly, these workflows with WF-IDs 1566099, 1566104, 1566120, 1566137 have fizzled after the optimization step. I have tried re-submission with several such work around, however it fizzled again. I have not found a way to solve for such cases, otherwise changing the lattice parameter and ion coordinates usually work, because you are essentially reducing the energy (arising from strain, ion repulsion-interactions, etc.) of your structure.

Perhaps other super users know how to deal with such cases. ( @dwinston )


#4

@Suman_Gandi, the solutions @George_Yumnam provided are great ideas to solve problems with optimization runs. As for the DB insertion tasks, they might fizzle due to issues on our end. We’ll look into the workflow IDs that you provide(d) as soon as our workload allows. Please get back to us if you don’t see any changes in the fireworks states or hear from us in the next couple of weeks.


#5

I have an elasticity calculation request for mp-540771 (Workflow 242839) which has fizzled. What does that mean and what should expect in terms of my calculation finishing.

I also have a structure prediction that I submitted, Workflow 1652513, which also says fizzled.


#6

@StephenFilippone Your structure suggestion fizzled at the core optimization step (http://fireworks.dash.materialsproject.org/wf/1652513). I would consider the advice of @George_Yumnam above.

As for your elasticity calculation request (http://elastic.dash.materialsproject.org/wf/242839), I will forward this issue to our internal czar for these calculations – it is a calculation requested by multiple people at this point.

Regarding expectations, keep in mind that structure submissions and property requests are taken as input from the MP user community and are mixed in with internal team priorities in light of our allocated computational budgets at supercomputing centers, so there is no guarantee that calculations will be fixed and rerun if fizzled or even be run at all. Nevertheless, we do want to help users like yourself; gentle nudging can go a long way insofar as it indicates you care enough about the request to follow up on it. :slight_smile:


#7

@dwinston Thanks for you help. It appears that the workflow for my elasticity calculation request has been completed. Is this actually the case? where can I find the results of that calculation? And actually I was interested in getting the elasticity calculation so it could be used to determine matching substrates. Do you know what I would have to do to make that happen?


#8

@StephenFilippone that should all be automatic. Workflow completion to display of the resulting elastic properties (and updated substrate data) should take less than a day (we run a daily cron job to rebuild from completed workflows).