各位老师如何评价Amber22？

ChaosChiao

Amber22的主要更新如下：

• Major extensions for alchemical free energy calculations using soft-core potentials, and some advanced lambda scheduling options. New ability to use SHAKE with TI.
• REST2-like enhanced sampling.
• Methods for studying protein-protein interactions using Gaussian accelerated MD (PPI-GaMD).
• Updates to self-guided Langevin dynamics, using both momentum and force guiding factors; new connections to the canonical ensemble.
• Kernel Modified MD allows on-the-fly updates to forcefields based on training configurations.
• Support in pmemd and mdgx for more types of "extra points".
  

AmberTools22的主要更新如下：

• New and updated forcefields:
  
  • Lipid21 major update for lipids
  • Polarizable water model, OPC3-pol
  • New parameters for some fluorinated amino acids
    
• Quick package for Hartree-Fock and DFT electronic stucture calculations, with GPU support. Quick is integrated into sander for QM/MM simulations, and AmberTools22 contains significant performance improvements, a new geometry optimizer, and support for spin-unrestricted calculations.
• The 3D-RISM code now supports periodic boundaries, and has improved performance and scaling for open boundaries
• Detailed changelog for cpptraj. Changes include GIST improvements, new cluster analysis, and a "prepareforleap" script with support for carbohydrates.
• The PyRESP program supports RESP fitting for both additive and polariable force fields.
• New constant pressure protocols for the Amber pGM polarizable force field.
  

相对于Amber20，又无必要升级？这个Quick目前如何？看着似乎在QM/MM计算速度上有较大提升。
另外我在手册中看到有一章是写机器学习的，但十分简略，还不清楚amber未来是否会进一步支持机器学习力场的应用。

rpestana94

I think if you don't need one of the new characteristic of amber you can work fine with amber20

HZW

Polarizable water model, OPC3-pol这个感觉有待测试，不管速度还是准确性（相对于目前普遍使用的非极化水模型）