Polyconf dihedral_solver() does not exit correctly when first monomer cannot be solved

[recombinatrix]

dihedral_solver() is meant to try each multiplicity of a dihedral to generate a valid conformation

If dihedral_solver() resolves the first dihedral successfully, then later fails to resolve the polymer conformation, it exits with a failure message

If dihedral_solver() never resolves the very first dihedral, it continues to rotate that dihedral forever, and never exits with the failure message

[recombinatrix]

examples

intended behaviour (failure):

failing gracefully because a polymer conformation could not be solved in this instance
https://github.com/OMaraLab/polyconstruct/blob/main/polyconf_examples/02c_build_PMMA_atactic.py

intended behaviour (success)

complete solve, sometimes backtracking
https://github.com/OMaraLab/polyconstruct/blob/main/polyconf_examples/02d_build_PMMA_atactic_with_linear_extend.py

bug (fails to fail)

cannot resolve first dihedral, stays rotating it forever instead of failing gracefully

#!/usr/bin/env python

from polyconf.Monomer import Monomer
from polyconf.Polymer import Polymer

print ('creating an isotactic PMMA polymer') 

dummies="CMA CN CP CQ"

PMMA_isotactic=Polymer(Monomer('MMAD_bonds.pdb')) # initialise

adds=49 # we will lay 49 additional monomers

for i in range (0,49):
        PMMA_isotactic.extend( 
            Monomer('MMAD_bonds.pdb'), 
            n=PMMA_isotactic.maxresid(), 
            nn=PMMA_isotactic.newresid(),
            names=dict(P='CMA',Q='CA',R='CN',S='C'), 
            joins=[('C','CA')], 
            ) 

alkanes=PMMA_isotactic.gen_pairlist(J='C',K='CA',first_resid=1,same_res=False,last_resid=49,mult=3)
PMMA_isotactic.dihedral_solver(alkanes,dummies=dummies,cutoff=2.0) # extreme cutoff to force failure on step one

[lunamorrow]

Fixed in #73

However, the dihedral_solver() function would benefit from an overhaul of its method in the near future. It currently has a pair of nested while loops which are not amenable to the maintenance and future development of PolyConf. I can tackle this with your approval if you are happy for me to @recombinatrix?

[recombinatrix]

I'm surprised that worked; I'd tried these approaches before and they either did not work, or broke tqdm.

Regardless, PR approved, pls merge.

[lunamorrow]

I've got the magic touch. Merging now :)

[recombinatrix]

As far as solver, I'm happy to discuss that as part of the next release, but not at this stage.

Solver is brittle, but with the exception of the bug you just fixed it, implements the search through dihedral space correctly. I don't want to refactor any functional algorithms while we're under review, but I'm happy to discuss for future versions.

[lunamorrow]

Yes agreed. I've made a new issue for it which we can come back to :)