CP2K计算PETG键能的cutoff的选择问题

alexanderwdz

准备来计算PETG（体系只含有C,H,O三种元素，含有苯环）的键能，选择的基组是 DZVP-GTH 势能是BLYP。CP2K的网站上确实有一个关于如何找到cutoff的例子，但是我没怎么看懂。 请问针对我这样的系统如果我任意给定一个相对较大的cutoff的值（比如300）会对结果有很大的影响吗？

alexanderwdz

附带程序如下
&GLOBAL
PROJECT petg_all
RUN_TYPE ENERGY_FORCE
PRINT_LEVEL HIGH
&END GLOBAL

&FORCE_EVAL
METHOD Quickstep
&SUBSYS
  &KIND C
   ELEMENT C
   BASIS_SET DZVP-GTH
   POTENTIAL GTH-BLYP-q4
  &END KIND
  &KIND H
   ELEMENT H
   BASIS_SET DZVP-GTH
   POTENTIAL GTH-BLYP-q1
  &END KIND
  &KIND O
   ELEMENT O
   BASIS_SET DZVP-GTH
   POTENTIAL GTH-BLYP-q6
  &END KIND
  &CELL
   A 60.000000000 0.000000000  0.000000000
   B 0.000000000  60.000000000 0.000000000
   C 0.000000000  0.000000000  60.000000000
  &END CELL
  &COORD
    O 30.6745964572 7.9710778033 50.5802286318
    C 29.7859753384 8.7342544861 51.4153131626
    H 30.3568036895 9.4537959405 51.9873785517
    H 29.1504327531 9.3642946210 50.7740729052
    C 29.0327221527 7.8163957799 52.3667999976
    H 29.8371661914 7.2827075860 52.9344386955
    C 28.1412477072 6.8288804500 51.6084102374
    H 27.5900483862 6.2244612305 52.3135121059
    H 28.7070053832 6.2313841068 50.8934658597
    C 26.9960822021 7.5879263452 50.8861907051
    H 26.3925238586 6.8681371956 50.3025718123
    H 27.4637904226 8.3347494431 50.1681081840
    C 26.0767443608 8.4915739751 51.7612702226
    H 25.6607387515 9.1987212959 51.0173884981
    C 26.9348321158 9.3060458424 52.7551094704
    H 26.2447953568 9.6267885434 53.5474121585
    H 27.3041929926 10.2574339667 52.3494552252
C 28.1257678964 8.5463565710 53.4375893673
    H 27.6225287146 7.7687639642 54.0564828463
    H 28.6598536982 9.1716792531 54.1138832547
    C 24.8691559658 7.7446011744 52.4287511765
    H 24.2591446767 8.5072837074 53.0065773552
    H 25.1390622306 6.8365868331 53.0913233304
    O 23.9820118273 7.2890634421 51.4466015249
    C 23.3499569937 6.0316085849 51.4246139509
    O 23.3685586989 5.3202856644 52.3926354436
    C 22.8294629155 5.6677897334 50.0576358578
    C 22.7200929416 4.3089489732 49.5838273072
    H 22.8336413471 3.5018287441 50.3871517834
    C 22.3781784830 4.0805261503 48.2446399774
    H 22.2029548756 3.0016022055 47.9292884546
    C 22.0293367219 5.1783014560 47.3849123296
    C 22.1870796631 6.4400934871 47.9143783726
    H 21.8881363204 7.3288886228 47.4082238881
    C 22.5058530044 6.7194439412 49.2032053352
    H 22.4113262940 7.7480748986 49.5675328049
    C 21.7076817796 4.9482381916 45.9708577310
    O 21.2531914267 5.7403773314 45.1884127495
    O 35.7735954725 2.5381080549 42.8968866824
    C 35.0894909448 3.1438476246 43.9544303848
    H 35.1792407994 2.5991027176 44.9427518911
    H 34.0300221539 3.1039027710 43.6523032738
    C 35.4034405533 4.6572819756 44.1709486118
    H 36.4833840161 4.8548037495 44.3223632192
    H 35.3248876950 5.1444763006 43.2211567928
    O 34.6216974746 5.3459737069 45.1251632063
    C 34.9419256431 6.0779415805 46.1924022099
    O 36.0157925302 6.5095409646 46.4775813929
    C 33.7264234668 6.3449934041 47.0010719719
    C 32.4648974727 6.5772213738 46.4083848190
    H 32.3678640450 6.4115641212 45.3590168361
    C 31.3825493110 7.0044768257 47.1355386868
    H 30.4041323548 7.1239575956 46.6293888918
    C 31.5408015535 7.2193152068 48.5752743267
    C 32.7820758129 7.0018518857 49.1750938371
    H 32.9634471134 7.3118096178 50.2011242085
    C 33.8096890384 6.4731900393 48.4221066063
H 34.7989558988 6.5814559026 48.9022272067
    C 30.3442058734 7.7464972687 49.2628994813
    O 29.2872934248 7.9519132919 48.7703926924
    H 36.5926404442 2.1613428441 43.1471426208
    O 21.9574948310 3.6878167199 45.6012242768
    H 21.8847277979 3.5955324985 44.6812824919
  &END COORD
&END SUBSYS

&DFT
  BASIS_SET_FILE_NAME  GTH_BASIS_SETS
  POTENTIAL_FILE_NAME  GTH_POTENTIALS
  &QS
   EPS_DEFAULT 1.0E-10
  &END QS
  &MGRID
   NGRIDS 4
   CUTOFF 300
   REL_CUTOFF 60
  &END MGRID
  &XC
   &XC_FUNCTIONAL BLYP
   &END XC_FUNCTIONAL
  &END XC
  &SCF
   SCF_GUESS ATOMIC
   EPS_SCF 1.0E-7
   MAX_SCF 300
   &DIAGONALIZATION
    ALGORITHM STANDARD
   &END DIAGONALIZATION
   &MIXING
    METHOD BROYDEN_MIXING
    ALPHA 0.4
    NBROYDEN 8
   &END MIXING
  &END SCF
&END DFT
&PRINT
   &FORCES ON
   &END FORCES
&END PRINT
&END FORCE_EVAL

啦啦黑还黑

对于cp2k来说，300 Ry的cutoff是很粗糙的精度。google group上有关于CUTOFF值的建议，对于包含Na, N, O, F, Ni, Ga等元素的计算需要设置到1000 Ry确保计算精度。不过你可以自己做个收敛性测试，只要你关心的性质对于CUTOFF收敛了就可以。

alexanderwdz

啦啦黑还黑 发表于 2019-2-22 20:02
对于cp2k来说，300 Ry的cutoff是很粗糙的精度。google group上有关于CUTOFF值的建议，对于包含Na, N, O, F, ...

那请问rel_cutoff的值相应应该怎么取呢

啦啦黑还黑

一般取50，精确计算用60 Ry

a386514315

啦啦黑还黑 发表于 2019-2-22 20:02
对于cp2k来说，300 Ry的cutoff是很粗糙的精度。google group上有关于CUTOFF值的建议，对于包含Na, N, O, F, ...

您好，请问这个图有原链接吗？谢谢

qli0218

请问可以对一个体系中的不同原子分别设置该原子的cutoff值吗？谢谢

yxdd98

qli0218 发表于 2024-5-7 20:41
请问可以对一个体系中的不同原子分别设置该原子的cutoff值吗？谢谢

应该不可以，直接取cutoff需求值最大的那类原子的cutoff作为体系的cutoff就成

不想飞的猫头鹰

a386514315 发表于 2023-12-13 23:12
您好，请问这个图有原链接吗？谢谢

https://groups.google.com/g/cp2k/c/DVCV0epl7Wo
之前看sob老师说过，cp2k振动分析是用有限差分的方式实现的，CUTOFF不够大容易造成能量不可忽视的虚假波动，即能量相对于原子位移缺乏足够理想的内在的平移不变性。然后给出了网上论坛给出的资源，测试了PBE结合GTH赝势达到1E-5 Hartree以下的平移不变性需要的CUTOFF。这个和能量、受力达到收敛需要的CUTOFF是两回事。个人感觉算键能其实跟能量差不多，应该不用图中这么大。