Bug: Memory leak in RT-TDDFT + DeePKS when using md_nstep

[xuan112358]

Describe the bug

I found a memory leak issue when performing RT-TDDFT calculations with the DeePKS functional in ABACUS. The memory usage increases continuously and eventually leads to a crash when md_nstep is greater than 1.
However, this issue does not occur in the following cases:

• Using DeePKS with estep_per_md
• Using PBE functional with md_nstep.

Log of memory is shown below with settings of RT-TDDFT+DeePKS+md_nstep

--- 监控开始: Tue May 26 18:34:34 CST 2026 ---
监控目标进程名: abacus_max_para
时间 | 运行小时 | 总内存(GB) | 进程数 | CPU总占用%
2026-05-26 18:34:34 | 0h | 进程 abacus_max_para 未找到
2026-05-26 18:39:34 | .08h | 11.08 GB | 10 ranks | 4565%
2026-05-26 18:44:34 | .16h | 17.59 GB | 10 ranks | 4586%
2026-05-26 18:49:34 | .25h | 23.77 GB | 10 ranks | 4600%
2026-05-26 18:54:34 | .33h | 29.69 GB | 10 ranks | 4607%
2026-05-26 18:59:34 | .41h | 35.52 GB | 10 ranks | 4606%
2026-05-26 19:04:35 | .50h | 41.88 GB | 10 ranks | 4604%
2026-05-26 19:09:35 | .58h | 47.80 GB | 10 ranks | 4606%
2026-05-26 19:14:35 | .66h | 53.61 GB | 10 ranks | 4606%
2026-05-26 19:19:35 | .75h | 59.53 GB | 10 ranks | 4608%
2026-05-26 19:24:35 | .83h | 65.44 GB | 10 ranks | 4609%
2026-05-26 19:29:35 | .91h | 71.36 GB | 10 ranks | 4608%
2026-05-26 19:34:35 | 1.00h | 76.95 GB | 10 ranks | 4606%
2026-05-26 19:39:36 | 1.08h | 82.66 GB | 10 ranks | 4600%
2026-05-26 19:44:36 | 1.16h | 88.05 GB | 10 ranks | 4590%
2026-05-26 19:49:36 | 1.25h | 93.72 GB | 10 ranks | 4596%
2026-05-26 19:54:36 | 1.33h | 99.09 GB | 10 ranks | 4597%
2026-05-26 19:59:36 | 1.41h | 104.67 GB | 10 ranks | 4598%
2026-05-26 20:04:37 | 1.50h | 110.55 GB | 10 ranks | 4602%
2026-05-26 20:09:37 | 1.58h | 117.09 GB | 10 ranks | 4603%
2026-05-26 20:14:37 | 1.66h | 122.14 GB | 10 ranks | 4605%
2026-05-26 20:19:37 | 1.75h | 127.72 GB | 10 ranks | 4606%
2026-05-26 20:24:37 | 1.83h | 134.26 GB | 10 ranks | 4606%
2026-05-26 20:29:37 | 1.91h | 139.20 GB | 10 ranks | 4607%
2026-05-26 20:34:38 | 2.00h | 144.54 GB | 10 ranks | 4609%
2026-05-26 20:39:38 | 2.08h | 150.66 GB | 10 ranks | 4609%
2026-05-26 20:44:38 | 2.16h | 156.39 GB | 10 ranks | 4609%
2026-05-26 20:49:38 | 2.25h | 162.29 GB | 10 ranks | 4609%
2026-05-26 20:54:38 | 2.33h | 168.74 GB | 10 ranks | 4608%
2026-05-26 20:59:38 | 2.41h | 174.64 GB | 10 ranks | 4608%
2026-05-26 21:04:39 | 2.50h | 180.45 GB | 10 ranks | 4608%
2026-05-26 21:09:39 | 2.58h | 186.14 GB | 10 ranks | 4608%
2026-05-26 21:14:39 | 2.66h | 192.05 GB | 10 ranks | 4608%
2026-05-26 21:19:39 | 2.75h | 197.85 GB | 10 ranks | 4608%
2026-05-26 21:24:39 | 2.83h | 204.07 GB | 10 ranks | 4609%
2026-05-26 21:29:39 | 2.91h | 209.65 GB | 10 ranks | 4610%
2026-05-26 21:34:39 | 3.00h | 215.79 GB | 10 ranks | 4608%
2026-05-26 21:39:40 | 3.08h | 221.81 GB | 10 ranks | 4606%
2026-05-26 21:44:40 | 3.16h | 228.30 GB | 10 ranks | 4606%

Expected behavior

No response

To Reproduce

Example files:

github_issue.zip

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).