# Week n+4¶

## Monday 26/7¶

### What I did today:¶

• After my meeting with Katya there’s a few things I need to do:
• Generate MESPs for the efield perturbations
• Potentially redo the relaxed efield perturbations with a more appropriate integration grid for M06-2X
• Redo all the benchmarks in Psi4, since ORCA isn’t behaving nicely
• We’ve decided to not use DCA as an anion, which makes my job slightly easier
• I generated the MESPs and put them into the PyMOL project
• I tweaked the script builder job
• I submitted the better scan jobs which are doing their thing
• I might actually cancel and resubmit these with MESP generation…
• ^this has been done
• I spent a fair amount of time trying to figure out how to extract a clustered trajectory from the dcd files with mda, but I think it’s probably not the best method.
• I’ve set off my MD generator script and it is now building and queueing jobs!
• I’ve suspended these jobs, only because when the IL test jobs finish tomorrow, I want to be able to get them straight back on.

## Tuesday 27/7¶

### What I did today:¶

• All the test runs timed out as expected, so I restarted them, making sure to specify to use P100 gpus so they should finish within 24 hours.
• Submitted all the jobs I generated yesterday
• Worked on finding decent cavities for PCMSolver, which meant troubleshooting compiling geomview, before I learned that nothing else can read .off files… except for https://3dviewer.net
• I settled on using a cavity area of 0.9
• Tweaked the benchmark jobs and brushed up on SCF algorithms to try and speed them up/prevent them from oscillating.
• I settled on using SOSCF, but we shall see how effective that is… (It works for ORCA)
• Mostly spent a lot of time troubleshooting XQuartz again, and also jupyterlab…
• I noticed that the temperatures of the jobs is closer to 200-230K when I’m specifying 343K for the thermostat, trying to follow up on this with Tom or Michael, but it’s late.
• This is because the reported temperature is the average of the temperature of all the particles, which Drudes have their own thermostat that keeps them at ~2K

## Wednesday 28/7¶

### To do:¶

• RDFs of ILs
• Load all geometry into PyMOL
• Update running jobs

### What I did today:¶

• I wrote a script to load all the geometries into PyMOL in different states to make it easier to browse their respective geometries
• I wrote an RDF generation script that uses MDAnalysis, which is much faster than TRAVIS
• I had a collaborator meeting in which Michelle was really happy with the work I’ve done and wants to get it into a paper, but now I’m feeling overwhelmed
• So I just sent an email to Katya about feeling overwhelmed
• The IL pMD jobs have begun to run at speeds of ~50-55ns/day, which should mean they’ll take just over two days to complete which is reasonable
• I managed to get MDAnalysis cluster selection and exporting working! and I managed to speed up the .dcd to .pdb conversion process, and I managed to get the geometry to read properly in pymol
• The RDFs I’m getting form MDAnalysis don’t feel quite right, so I’m trying again with TRAVIS to see if looks any different.
• This didn’t but I realised the error. I was computing all atoms of one molecule against thee other, as opposed ot sat the N-S distance.
• I also used mdtraj as per Tom’s suggestion which made this process so much faster!
• Using both of these things, I’m still not seeing any difference between the two IL test jobs, but I think it might have something to do with the constant acceleration. If I compare that RDF to one of the in progress na1* jobs, I do see a difference!
• I also forgot that I need to do comparative MD between field off and on, so I’ve queued up those jobs as well!

• I might also get them queued on M3 just in case they get picked up there sooner… - done
• I also noticed that my MonARCH jobs weren’t submitted under partner, so I’ve fixed that up

From meeting with Michelle

For a paper! (Special edition of PCCP?)
Based on analysis, we want an e-field pointing this way…

When we do, we get a rate increase of X resulting in kinetics of Y

What will give the best selectivity between the two?… (biggest ΔE)
If we only want R?… (biggest S barrier)
If we only want S?… (biggest R barrier)

Then Michelle can demonstrate how to get this through functionality
Then Joseph can test this experimentally
this can then be explored thorugh a Hammett study to detemrine the effectiveness of the susbtituent

To turn this into a paper then we need to do this for a few derivatives

Then do this in multiple solvents to demonstrate that polar solvents prevent this and that non-polar solvents help this and there’s a midpoint between the two of solubility and effectiveness

Use the Check JOC paper that Michelle just did on the Aldol reaction as a guide

## Thursday/Friday¶

### What I did today:¶

• I did do things… I just haven’t been feeling the best.
• I worked on pymol scripts and analysis of the efield scans mostly

### Benchmark Matrix¶

Job na1r - r na1t - r na1p - r na1r - s na1t - s na1p - s
Non-catalysed benchmark queued queued queued queued queued queued
R benchmark (static) queued queued queued queued queued queued
S benchmark (static) queued queued queued queued queued queued
Catalysed benchmark (relaxed)
R benchmark (relaxed)
S benchmark (relaxed)

### MD Run Matrix¶

IL na1r-r na1t-1-r na1t-2-r na1t-3-r na1p-r na1r-s na1t-1-s na1t-2-s na1t-3-s na1p-s
$$\il{C4mpyr}{OTF} - \vec F =0.1\:V\cdot\AA^{-1}$$​​​​ running queued queued queued queued running queued queued queued running
$$\il{C4mpyr}{MSO4} - \vec F =0.1\:V\cdot\AA^{-1}$$​​​ done running running running done done failed running running done
$$\il{C4mpyr}{TCM} - \vec F =0.1\:V\cdot\AA^{-1}$$​​ done failed done done done failed done done failed done
$$\il{C4mpyr}{OTF} - \vec F =0.0\:V\cdot\AA^{-1}$$​​​ queued queued queued queued queued queued queued queued queued queued
$$\il{C4mpyr}{MSO4} - \vec F =0.0\:V\cdot\AA^{-1}$$​​​ queued queued queued queued queued queued queued queued queued queued
$$\il{C4mpyr}{TCM} - \vec F =0.0\:V\cdot\AA^{-1}$$​​ queued queued queued queued queued queued queued queued queued queued