Intra add

[nyee]

This pull request fixes the duplicate problem between Intra_R_Add_Exocyclic and Intra_R_Add_Endocyclic and adds some groups for rings in the Intra_R_Add_exocyclic family.

[mention-bot]

@nyee, thanks for your PR! By analyzing the history of the files in this pull request, we identified @connie, @jwallen and @mliu49 to be potential reviewers.

[zjburas]

The latest commit forbids Intra_R_Add_Endo from ever adding to the double bond of an aromatic ring, such as in the following reactions:

Intra_R_Add_Exo already picks up these reactions in the other resonance form, so I think it is redundant for Endo to find it as well.

The new forbidden group does not, however, prevent reactions like the following from happening, which I think was a concern for @nyee ::

[mliu49]

I want to make sure I understand. So in general, we want to allow reactions like the latter, where *2 and *3 are the only labeled atoms on the ring, to happen in Intra_R_Add_Endocyclic (because they should behave similarly to linear reactants, and there's usually no risk of duplication with Exocyclic). However, for aromatics, the same reaction would be duplicated in Exocyclic due to resonance, which is what the latest commit fixes.

[zjburas]

That is my understanding, but @nyee will have to confirm.

[nyee]

Nothing in this patch addresses the problem of aromatics which is a specific case of this issue .

The last commit is actually forbidding endo reactions where the radical is on the benzylic position. This is because the atom labelling changes, such that *4 is on the ring in this case. In the other cases, where the radical is further from the ring, the atom labelling with always have *5 on the ring.

[zjburas]

@nyee , @mliu49 is referring to the latest commit above: 8508636. See the explanation above.

[nyee]

Oh sorry, I didn't see the commit. Yes this seems correct to me now.

[mliu49]

I have some observations and questions after testing a few reactants:

1. 4d1723d forbid radicals two carbons away from the ring from reacting with the meta or para positions. However, radicals one carbon away from the ring can still react (eg. benzyl radical can still add to meta and para sites). Do we want to forbid those as well?
2. The reaction degeneracies seem to be wrong. I'm testing with KineticsFamily.generateReactions(), so no resonance is being considered.
   • Observations:
     • For benzyl radical (ignoring how feasible these reactions are for now)
       • ortho: degeneracy = 2
       • meta: degeneracy = 4
       • para: degeneracy = 2
     • For propylbenzene and pentylbenzene primary radicals
       • ortho: degeneracy = 2
       • meta: degeneracy = 4
       • para: degeneracy = 1
     • For pentylbenzene primary radical
       • ortho: degeneracy = 1
       • meta: degeneracy = 2
       • para: degeneracy = 1
     • For 3-(methyl/ethyl/propyl/etc.)-cyclohexene primary radicals
       • ortho: degeneracy = 2
       • meta: degeneracy = 2
   • Hypotheses:
     • For aromatics, degeneracies are decreasing with chain length because we don't have sufficiently long groups
     • For all species with the double bond on a ring, we are matching both the linear backbone groups and the cyclic backbone groups. Therefore, we get twice as many atom mappings, and twice the degeneracy. Put another way, we implicitly assume that the groups in logic OR nodes are mutually exclusive, at least for degeneracy calculations. In this case, that's not true because the cyclic groups are a subset of the linear groups.
3. For the new cyclic groups you added in Intra_R_Add_Exocyclic, why do beta groups have long and short, but alpha groups do not? I think this is the reason the meta degeneracies for aromatics are twice the ortho degeneracies, which I think should be equal in theory.

[zjburas]

@mliu49 , Regarding question 1, yes, we should certainly make forbidden groups preventing benzylic carbon from adding to the meta or para position. I didn't forbid them in the initial commit because I was only thinking about my specific system.

It is also probably reasonable to forbid radicals 3 or even 4 carbons away from the ring from adding to the para position (and maybe meta too). I suppose it would be best to examine the potential TS's for these additions to decide what exactly should be forbidden. I can work on this.

[zjburas]

The latest commit forbids carbon radicals in either the alpha or gamma position (in addition to the eta already banned) relative to a phenyl side group F

[zjburas]

The latest commit forbids carbon radicals in either the alpha or gamma position (in addition to the beta already banned) relative to a phenyl side group from adding to the meta or para positions. Examining the possible TS's for these reactions, they all seem unlikely to happen. Even forbidding such reactions out to the delta carbon radical might be justified.

The forbidden groups could also probably be made for general. For example, the linear side chain need not consist entirely of carbons, nor do all of the atoms need to be close-shell. Making these forbidden groups more general might help systems with heteroatom containing cyclics (like oxygenates). I don't know what the usual philosophy is regarding how general to make forbidden groups, so I will await your advice.

The same commit also bans the phenyl group from self-ring-closing to form a 3-5 or 4-4 membered bicyclic.

[mliu49]

Will be replaced by new pull request soon.