deepmodeling
Memory Leak
Describe the bug
我在计算UN的时候，用小轨道，算300步scf不会爆炸，但是内存会上升。用中等轨道就爆炸了，调节OMP可以跑完300步，但是最后输出会报错。（这个内存泄漏应该跟安装无关，我在cpu，gpu，以及一位网友的gpu上都出现了这个） 还有这个体系一般都是用VASP算，文章的声子谱可以接近实验值。我想用abacus算，但是在U加到一定的数值就完全不能收敛，磁性材料很难收敛，希望abacus努力，vasp太慢了。这个加强收敛的办法我都试过了，包括uramp和omc=1或者2，那些帮助收敛的参数我都测试过，都用的是小+U下面收敛总时间最短的（qq群里已经没有建议了，说的建议我都试过了）。中等轨道的速度大于慢三倍，但是晶格常数更接近实验，可惜还是没收敛（222超包没收敛，原胞收敛了）
Expected behavior
不要内存爆炸，且让这个体系能更好收敛，狠狠diff vasp
To Reproduce
No response
Environment
No response
Additional Context
No response
Task list for Issue attackers (only for developers)
- [ ] Verify the issue is not a duplicate.
- [ ] Describe the bug.
- [ ] Steps to reproduce.
- [ ] Expected behavior.
- [ ] Error message.
- [ ] Environment details.
- [ ] Additional context.
- [ ] Assign a priority level (low, medium, high, urgent).
- [ ] Assign the issue to a team member.
- [ ] Label the issue with relevant tags.
- [ ] Identify possible related issues.
- [ ] Create a unit test or automated test to reproduce the bug (if applicable).
- [ ] Fix the bug.
- [ ] Test the fix.
- [ ] Update documentation (if necessary).
- [ ] Close the issue and inform the reporter (if applicable).
Thanks for your advice! I have changed the title to English for convenience. We will try to find the problem recently.
We've done some tests, this OOM problem seeems to be related to mixing_restart in DFT+U method.
Our testing revealed that the OOM occurs after 'SCF restart after this step!'. Attached are the original file and memory usage data from the calculation.
