Closed mostafa-razavi closed 4 years ago
The Auxiliary Nested PSO method found the following solution. Only U^res was optimized in the inner PSO with a phia=0.7, Nsnapshots=500, and target_Neff=25
Conclusion: 1) The distinction between CH3c and CH3b is not justifiable anymore, because their diameter and energy parameters are very close. (sig_CH3b=3.8, eps_CH3b=120)
Todo: 1) Modify the branched alkane molecules and replace all CH3c's with CH3b
1) Neopentane's liquid density seems to have a wrong trend. 2) The highest Tr IC did not converge, because the Tr was above 0.85 (around 0.9). Note that for NP. the lowest Tr=0.6. 3) The IT points seem to have too much noise. It might be because T^IT is too low and those points are inside the force field's phase envelope.
1) 22DMH is better than NP.
Todo:
1) Make a new NP.itic at Tr^IT=1.2, ignore the lowest IC and run reference simulations for NP at select points: 2) Modify all TranSFF0 branched molecules by replacing CH3c with CH3b and start optimizing a new set of parameters for C site type focusing more on sigma.
Here is the 3.8-105_6.4-5 looks like with wrong bond length B=1.40
Here is the 3.8-105_6.4-5 looks like with corrected bond length B=1.54
Here is the 3.8-120_6.4-5 looks like with corrected bond length B=1.54
Both are way off. A new nested aux pso is running for NP