221212 Meeting ideas

[ehhartmann]

Topology and Parameter Patch Refactor

Detailed Balance update

• " It states that at equilibrium, each elementary process is in equilibrium with its reverse process."
• Roux paper used symmetric momentum reversal for MD/MC simulations to ensure detailed balance
• momentum reversal was applied to short neMD simulations which were used to calculate ΔE of a reaction
  -> we already have ΔE for HAT and it's reversible, homolysis is seen as irreversible at the moment (which is close to the truth?!)
  -> unclear how this would apply to rejection free kMC
  -> reaction MD is only used to get product state, no meaning associated to it
• do we want to stay in detailed balance? Can we regain detailed balance?
  -> view time before reaction as in equilibrium, reaction MD + first part of next pulling MD as transformation to the next equilibrium --> in detailed balance

Using KIMMDY

• which features should be part of KIMMDY v2?
  • better (espaloma) parameters? partial charges?
  • documentation?
• what do we want to use KIMMDY v2 for?
• do we immediately want to continue developing KIMMDY after v2 release?
• which features do we want to implement next?

Open questions from last meeting

• How to get prefactor? QM/MM? pure QM?
• MC w/ rejection:
  • cutoff in time domain

• How to mix reactions?
  • option 1: Only do one kind at a time
  • option 2: Competing rates need to be evaluated at the same time (as we currently do)
  • Competing rates can have diff prefactors
• maybe recalculate prefactor for homolysis in QMMM together with HAT QMMM

[KRiedmiller]

Protocol:

• Topology changes
  • goal: general solution for changing topology
  • in memory representation
  • Testing in place
  • Todo: regression tests with new KIMMDY versions
  • Bonds can be broken and rebuild from atom index
  • Atom types get changed using patches -> parameters get changed by factors or absolute values
  • Todo: charges must add up after applying patch!
  • Todo: Check /# in FF
  • How to store radical? Change atom name
    • gmx might complain
    • usefull for analysis
• Detailed Balance update
  • Momentum reversal:
    • Each process in eq w/ iths reversed process
    • To ensure this, vel reversed
  • Not important for practical application
  • MC step starts from eq, ends in eq, should be in detailed balance all the time, exluding the reaction itself
  • Maybe just try
• Using KIMMDY/Release
  • Partial charges should be good; they are!
  • Espaloma parameters as new verions; already implemented!
  • Publishing:
    • Methods
    • one use case / example like hexala for statistics and details
    • one big one to show off
    • maybe use ala-ala-dopa-ala-ala
      • would be specifically for collagen
  • Documentation
    • example
    • auto generated technical documentation
  • Features:
    • Espaloma -> next paper
    • Collagen -> next paper
    • MC w/ rejection + testing -> first release!
• Open Questions
  • Prefactor:
    • QM/MM
    • Ab initio MD: g09 issue: frozen part, harmonic restrains needed
    • calculating for homolysis and HAT together makes sense
  • Topology changer as standalone tool?
    • for MDAnalysis
    • for gromacs
    • would be alot of work
    • for now keep it inside KIMMDY