能量最小化后NVT 升温初始步 LINCS 崩溃（madrid 2019的H3O+体系，CUDA error #700）

kogareru1z

我在对纯H3O+体系进行nvt升温测试时模拟崩溃。我首先进行了能量最小化，log显示如下：
NOTE 1 [file init.top, line 14]:
  System has non-zero total charge: 8.500000
  Total charge should normally be an integer. See
  https://manual.gromacs.org/current/user-guide/floating-point.html
  for discussion on how close it should be to an integer.
WARNING 1 [file init.top, line 14]:
  You are using Ewald electrostatics in a system with net charge. This can
  lead to severe artifacts, such as ions moving into regions with low
  dielectric, due to the uniform background charge. We suggest to
  neutralize your system with counter ions, possibly in combination with a physiological salt concentration.
NOTE 2 [file em.mdp]:
  The optimal PME mesh load for parallel simulations is below 0.5
  and for highly parallel simulations between 0.25 and 0.33,
  for higher performance, increase the cut-off and the PME grid spacing.
但最后还是成功收敛了：

但之后进行nvt升温模拟，程序报错崩溃

我认为是[constraints]导致的问题，当我注释掉H-H的约束条件后，模拟可以进行，但HOH的角度无法控制，H3O+不是刚体而且结构明显不合理

结构文件是采用的是J. Chem. Phys. 162, 171101 (2025)的补充材料，应该不会出错才对啊...

kogareru1z

再次测试发现，将constraint_algorithm = LINCS 改为 SHAKE，nvt模拟不再报错，查看最后生成的结构文件，H3O+也保持刚体；这是否是lincs的算法bug？查看官方文档发现lincs算法似乎不能与角度约束一起使用？这里约束三对H-H距离是否就变相等于约束了HOH角度从而导致模拟崩溃？