What is the most accurate molecular dynamic simulation method

Question

First of all, I want to say that I am extremely new to MD simulations, so I apologize for any of these questions if they appear to be something that should be well known.

I know that the classical type of picture that MD packages such as Gromacs give us are not really useful for simulating chemical reactions and non local electron phenomenon, so I am interested in pinning down a very precise and accurate simulation method before beginning a lot of work in the MD area. I also know that many people really like car-parrinello molecular dynamics (CPMD) in order to circumvent the problems classical simulations (like those that can be run with Gromacs) has with only dealing with localized electrons.
However, CPMD was developed in 1985 So, I was wondering if there are any more recent advancements in doing highly accurate quantum mechanical modeling that will give better results than just CPMD.

Also, I am interested in exactly what types of properties these simulations will give you. I am most interested in adsorption enthalpies between two materials and optical properties, such as exciton energies and optical oscillator strengths etc. But it would be nice to know if there are any properties MD can't give

Answer 1

The area you are asking about is known formally as ab initio molecular dynamics. Marx and Hütter a few years ago wrote a comprehensive monograph on recent updates in the method, and can be recommended for providing a useful summary of the (almost) current state-of-the-art in the field.

However, one thing that should definitely be considered before doing any serious amount of work in the MD community is understanding what the relevant time and length scales you are working with, and how well those translate into the particular branch of MD you want to run. For instance, you aren't going to be able to simulate 10,000 or 100,000 atoms for 100 nanoseconds using CPMD or any other branch of ab initio molecular dynamics. Similarly, if you want to observe processes occurring at the electronic scale, as you mentioned, classical MD is a poor choice.