search

mostafa-razavi / ITIC-paper

0 stars 0 forks source link

How fast is ITIC compared to other methods (GCMC, GEMC)? #2

Closed mostafa-razavi closed 5 years ago

mostafa-razavi commented 6 years ago

Dr Elliott suggested to perform a benchmark comparison between GCMC (Tr=0.55-0.85) and ITIC (Tr=0.45-0.85)

ramess101 commented 6 years ago

@mostafa-razavi

I am still very cautious of trying to perform a rigorous comparison. I think there are a lot of parameters and factors we would need to consider to make this a fair comparison. For example, it is not straight forward to compare the methods since the performance of GCMC depends strongly on the insertion acceptance which depends on the compound, the algorithms, and the temperature. Furthermore, GCMC can use much smaller systems, whereas we are not sure what size of system can be used in ITIC. I am in favor of making a superficial comparison and focusing more on the validation by comparing with GCMC.

mostafa-razavi commented 6 years ago

@ramess101 Yes. I agree. I asked Mohammad if they can give me some numbers about their GCMC simulations. If yes, it might be enough for the superficial comparison.

ramess101 commented 6 years ago

@mostafa-razavi

This is what I said in our JCTC publication:

One advantage of ITIC is that all the simulations are performed in the NVT ensemble. In fact, although ITIC requires roughly three simulations for each saturation temperature, the total simulation time is typically comparable to the traditional VLE methods, i.e. GEMC or GCMC. This is primarily because NVT systems converge quickly, as they do not require expensive particle insertion/deletion or volume fluctuation moves.