VASP自旋轨道耦合计算错误汇总

VASP自旋轨道耦合计算错误汇总静态计算时，报错VERYBADNEWS!Internal内部errorinsubroutine子程序IBZKPT:Reciprocal倒数的latticeandk-latticebelongtodifferentclassoflattices.Oftenresultsarestilluseful...48INCAR参数设置：对策：根据所用集群，修改INCAR中NPAR。将NPAR=4变成NPAR=1，已解决！错误：subspacematrix类错误报错：静态和能带计算中出现警告：WARNING:Sub-Space-Matrixisnothermitian共轭inDAV结构优化出现错误：WARNING:Sub-Space-MatrixisnothermitianinDA/4-4.681828688433112E-002对策：通过将默认AMIX=0.4，修改成AMIX=0.2（或0.3），问题得以解决。以下是类似的错误：WARNING:Sub-Space-Matrixisnothermitianinrmm-3.00000000000000RMM:22-0.167633596124E+02-0.57393E+00-0.44312E-0113260.221E+00BRMIX:veryseriousproblemstheoldandthenewchargedensitydifferoldchargedensity:28.00003new28.060930.111E+00错误：WARNING:Sub-Space-Matrixisnothermitianinrmm-42.5000000000000ERRORFEXCP:suppliedExchange-correletiontableistoosmall,maximalindex:4794错误：结构优化Bi2Te3时，log文件：WARNINGinEDDIAG:subspacematrixisnothermitian1-0.199E+01RMM:2000.179366581305E+01-0.10588E-01-0.14220E+007180.261E-01BRMIX:veryseriousproblemstheoldandthenewchargedensitydifferoldchargedensity:56.00230new124.7039466F=0.17936658E+01E0=0.18295246E+01dE=0.557217E-02curvature:0.00expectdE=0.000E+00dEforcontlinesearch0.000E+00ZBRENT:fatalerrorinbracketingpleasererunwithsmallerEDIFF,orcopyCONTCARtoPOSCARandcontinue但是，将CONTCAR拷贝成POSCAR，接着算静态没有报错，这样算出来的结果有问题吗？对策1：用这个CONTCAR拷贝成POSCAR重新做一次结构优化，看是否达到优化精度！对策2：用这个CONTCAR拷贝成POSCAR，并且修改EDIFF（目前参数EDIFF=lE-6），默认为10-4错误：WARNING:Sub-Space-MatrixisnothermitianinDA/1-7.626640664998020E-003网上参考解决：对策1：减小POTIM:IBRION=0，标准分子动力学模拟。通过POTIM控制步长。POTIM:当IBRION=1,2或3时，是力的一个缩放常数（相当于确定原子每步移动的大小），默认值为0.5。对策2：改IBRION=1,采用准牛顿算法来优化原子的位置。原IBRION=2，采用共轭梯度算法来优化原子的位置对策3：修改ISMEAR对策4：换成CG弛豫（共轭梯度算法）IBRION=2（决定结构优化过程中，原子如何移动或弛豫）IBRION=2离子是否运动，1不运动但做NSW外循环。0动力学模拟，1准牛顿法离子弛豫2CG法离子弛豫，3采用衰减二阶运动方程离子弛豫，INCARrelax中设置IBRION=2，未解决！对策5：用的CG算符，出现的错误是CG算符不能算，在INCAR中加上IALG=Fast（电子优化采用blockedDavidson方法［IALGO=38:IALG=Normal］和RMM-DIIS算法［IALGO=48:IALG=VeryFast］混合）试一试IALG二Fast(两种方法混用)IALG=Very_Fast等价于IALGO=48)IALG二Normal(等价于IALGO=38)INCAR中加上IALG二Fast已解决！(1QL、2QL已解决，3QL以上未解决)VASPFORUM:theerrorisduetoaLAPCKcall(ZHEGV):ZHEGVcomputesalltheeigenvalues本征值，andoptionally随意地，theeigenvectorsofacomplexgeneralizedHermitian-definiteeigenproblem.theremaybeseveralreasonsforthaterror:theRMM-DIISdiagonalisationalgorithmisnotstableforyourspecificsetupofthecalculation.-->useALGO=Normal(blockedDavidson)orALGO=Fast(5stepsblockedDavidson,RMM-DIIS)用ALGO=NormalIALGO=48或者ALGO=Fast2)maybeyourinputgeometrywasnotreasonable(erroroccursattheveryfirstionicstep,pleasehavealookforthegeometrydataofyourruninOUTCAR)orthelastionicrelaxationstepleadtoanunreasonablegeometry(comparetheinputandoutputgeometriesofthelastionicrelaxationstepsinXDATCAR).Inthatcase(2b)itcanbehelpfulto-->switchtoadifferentrelaxationalgorithm(IBRION-tag)-->reducethestepsizeofthefirststepbysettingPOTIMsmallerthanthedefaultvalue改变IBRION，减少步长POTIMTheinstallationoftheLAPACKonyourmachinewasnotdoneproperly:usetheLAPACKwhichisdeliveredwiththecode(vasp.4.lib/lapack_double.o)IftheerrorpersistalthoughyouswitchedtotheDavidsonalgorithm:onsomearchitectures(especiallySGI)someLAPACKroutinesarenotworkingproperly.However,itispossibletoavoidtheusageoftheZHEGVsubroutinebycommentingtheline#defineUSEZHEEVXindavidson.F,subrot.F,andwavprenoio.FandrecompilingVASP.关于Mixing方法的调试：针对这类错误：DAV:13-0.242323773333E+030.98155E+02-0.87140E+01488320.949E+01BRMIX:veryseriousproblemstheoldandthenewchargedensitydifferoldchargedensity:252.00012new252.299790.809E+01WARNING:Sub-Space-MatrixisnothermitianinDAV90.133520549894753解决办法只需调整AMIX,BMIX的值，把他们设置小一些。Mixing方法：IMIX=typeofmixing混合、混频，AMIX=linearmixingparameter，AMIN=minimalmixingparameter,BMIX=cutoffwavevectorforKerkermixingscheme,AMIX_MAG=linearmixingparameterformagnetization,BMIX_MAG=cutoffwavevectorforKerkermixingschemeformag,WC=weightfactorforeachstepinBroydenmixingscheme,INIMIX=typeofinitialforeachstepinBroydenmixingscheme,MIXPRE=typeofpreconditioninginBroydenmixingscheme,MAXMIX=maximumnumberstepsstoredinBroydenmixer.一般采用其默认值，除非在电子迭代难以收敛的情况，才手动设置AMIX和BMIX等参数值。】对策：grepAMIXOUTCARAMIX=0.40;BMIX=1.00AMIX_MAG=1.60;BMIX_MAG=1.00initialmixingisaKerkertypemixingwithAMIX=0.4000andBMIX=1.0000设置初始值AMIX=0.0100；BMIX=0.0001AMIX=0.1000；BMIX=0.0010AMIX=0.20;BMIX=0.01AMIX=0.2、BMIX=0.001AMIX=0.3、BMIX=0.1AMIX=0.4收敛值AMIX=0.01;BMIX=0.00AMIX=0.10;BMIX=0.00AMIX=0.20;BMIX=0.01AMIX=0.2、BMIX=0.001AMIX=0.3、BMIX=0.1AMIX=0.40;BMIX=1.00结果计算无误计算无误计算无误计算无误计算无误静态log:WARNINGinEDDRMM:callto**AMIX=0.02AMIX=0.1AMIX=0.02;BMIX=1.00AMIX=0.10;BMIX=1.00计算无误静态log:WARNINGinEDDRMM:calltoZHEGVfailed,returncode=63**,能带一样**AMIX=0.3AMIX=0.30;BMIX=1.00静态log:WARNINGinEDDRMM:calltoZHEGVfailed,returncode=63**,能带一样**BMIX=0.0001AMIX=0.40;BMIX=0.00以上参数设置，得到的能带图都一样，如下图:综上：设置AMIX=0.2（或0.3）,BMIX默认（省事,算出正确的能带。计算无误K"M等于1.0），可以保证计算过程无误。还需进一步调整其他参数,ZHEGVfailed,returncode=63**,能带一样警告：算1QL弛豫、静态、能带时，都有这个提示：ADVICETOTHISUSERRUNNING'VASP/VAMP'(HEARYOURMASTER'SVOICE...):Youhavea(moreorless)'smallsupercell'andforsmallercellsitisrecommendedtousethereciprocal-spaceprojectionscheme!Therealspaceoptimizationisnotefficientforsmallcellsanditisalsolessaccurate...ThereforesetLREAL=.FALSE.intheINCARfile对策：对于较小的晶胞(原子数小于20)，设置LREAL=.FALSE.，计算结果比较精确。而对于较大的晶胞，设置LREAL=Auto，这样计算速度比较快。本体系含原子5个，INCAR中LREAL=Auto。设置所有INCAR中的LREAL=.FALSE.，重新算一遍。对于1QL2QL3QL原子数分别为5、10、15，LREAL=.False.对于4QL5QL6QL原子数分别为20、25、30,LREAL=Auto自旋轨道耦合计算时，静态和能带计算中出现的错误：ERROR:noncollinearcalculationsrequirethatVASPiscompiledwithouttheflag-DNGXhalfand-DNGZhalf分析：VASP手册中关于自旋轨道耦合计算的描述（翻译版）：非线性计算和自旋轨道耦合：旋量是由GeorgKresse在VASP代码中引入的。这个代码是由DavidHobbs编写，用于处理非线性磁结构。自旋轨道耦合计算是由OlivierLebacqandGeorgKresse共同实现的。只有VASP4.5以上的版本才支持旋量的计算。在INCAR中设置LNONCOLLINEAR=.TRUE•允许执行完全非线性磁结构的计算。VASP有能力读入之前非磁或非线性计算得到的WAVECAR和CHGCAR文件，然而它不可能扭转局域在指定原子处的磁场。因此在实际操作中，我们推荐分两步执行非线性计算：第一步，计算计算非磁性基态，产生WAVECAR和CHGCAR文件。第二步，读入WAVECAR和CHGCAR文件，通过设置MAGMOM参数，提供初始的磁矩。对于非线性设置，在MAGMOM这一行，每个离子必须设置三个值。这三项分别对应每个离子在x,y,z方向的初始局域磁矩值。MAGMOM=100010这一行，给第一个原子赋予的初始磁矩值沿x方向，第二个原子的初始磁矩值沿y方向。注意：只有在ICHARG=2（即不读入之前CHGCAR的情况）或者CHGCAR文件中只包含电荷但是不包括磁密度数据的情况（即之前那一步进行了非磁的计算）下，才需要通过MAGMOM设定初始磁矩值。LSORBIT-tagSupportedasofVASP.4.5.【设置LSORBIT=.TRUE.表示计算自旋轨道耦合，并附带自动设置了LNONCOLLINEAR=.TRUE.】LSORBIT=.TRUE.只能用于PAW赝势，不能用于超软赝势。如果不考虑自选轨道耦合，则能量不依赖磁矩的方向，也就是说，旋转所有的磁矩以同一个角度，让它们拥有相等的能量。不考虑自选轨道耦合的时候，不需要定义自旋量子化坐标。开启自旋轨道耦合设置以下参数：LSORBIT=.TRUE.SAXIS=s_xs_ys_z（自旋量子化轴，默认值SAXIS=（0+,0,1））GGA_COMPAT=.FALSE.!应用球面截断能到梯度场其中SAXIS默认=（0+,0,1）（0+表示沿x轴方向一个无穷小的正数）。当需要计算亚meV能量尺度的微小能量差异（一般指磁各向异性计算的情况）时，需要设置GGA_COMPAT这个参数。现在所有关于坐标轴（Sx,Sy,Sz）的磁矩都给出来了，我们采用VASP中给出关于这个坐标轴所有磁矩和自旋状量子读写惯例。这包括INCAR文件中的MAGMOM行，OUTCAR和PROCAR文件中的总和局域磁矩，WAVECAR文件中的类自旋轨道，CHGCAR文件中的磁密度。笛卡尔坐标系中的磁分量由以下等式得到：m=cos(P)cos(a)加曲$—sin(a)maxis+sin(P)*cos(a)maxisTOC\o"1-5"\h\zxxyzm=cos(P)sin(a)m+cos(a)maxis+sin(P)sin(a)m«xisyxyzm=—sin(P)maxis+cos(P)maxiszxz其中，maxis是外部可见的磁矩值，此处的a是SAXIS矢量(sx,sy,sz)和笛卡尔坐标x轴的夹角，卩是SAXIS矢量和笛nIs2+s2|卡尔坐标z轴的夹角,以下等式得到逆变化:P=atan—szmaxis=cos(P)cos(a)m+cos(P)sin(a)m+sin(P)mTOC\o"1-5"\h\zxxyzmaxis=—sin(a)m+cos(a)myxymaxis=sin(P)cos(a)m+sin(P)sin(a)m+cos(P)m,xyz不难看出，默认值（sx,sy,sz）=（0+,0,1），两个角度都是0,即卩=0和a=0。在这种情况下，内部转换简单地等于外部地转换：mx=呼,my=彎，mz=maxis，第二种重要的情况，是mr=0和營s=0，在这种情况下:m=sin(P)*cos(a)maxis=maxiss/、s2+s2+s2xzzz'xyzm=ym=cos(P)maxis=maxiss/、、.s2+s2+s2zxzzxyz因此现在磁矩是平行于SAXIS矢量。这样有两种方式去旋转自旋到任意方向，即通过改变初始的磁矩MAGMOM或改变SAXIS。为了给计算赋予平行于一个选定的矢量（x,y,z）的初始磁矩，可以通过设定（假定是单原子原胞）：MAGMOM=xyzSAXIS=001!局域磁矩xyz!量子轴平行于z轴或者MAGMOM=00total_magnetic_moment!局域磁矩平行于SAXISSAXIS=xyz两种设置都必须在相同能量的标准/辐射（原则、!量子轴平行于矢量（x,y,z）根源）场，但是要实现第二种方法，通常更加精确。第二种方法，也允许读入之前存在的WAVECAR文件（由线性计算还是非线性计算产生的都可以），然后继续用一个不同的自旋方向计算。当读入一个非线性WAVECAR文件，自旋假定平行于SAXIS（因此VASP将仅仅输出一个z轴方向的磁矩）。推荐计算磁各项异性的步骤如下:先做线性计算，得到一个WAVECAR和CHGCAR文件。加入以下参数:LSORBIT=.TRUE.ICHARG=11!非自洽计算，读入CHGCARLMAXMIX=4!对于d电子元素设置LMAXMIX=4,f电子元素设置LMAXMIX=6!在线性计算中，需要设置LMAXMIXSAXIS=xyz!磁场的方向NBANDS=2*线性计算能带数GGA_COMPAT=.FALSE.!在梯度场中应用球面截断能VASP读入WA/ECAR和CHGCAR文件，将自旋量子轴对齐SAXIS矢量，这意味着现在磁场平行于SAXIS矢量,执行非线性计算。通过比较不同方向的能量，可以确定磁各向异性。请记住，原则上，在VASP中一个完全地自洽计算（ICHARG=1）也是有可能的，但是这种情况将会允许自旋波函数从它们的初始值旋转到平行于SAXIS矢量，直到获得正确的基态，也就是，直到磁矩平行于易磁化轴。实际操作中，这种旋转非常缓慢，直到自旋获得少量能量重新定位。因此，如果收敛标准太精确，完全地自洽计算可以得到一个比较合理的结果（我们实验过的几种自洽计算都没有问题。）要非常小心对称性。我们建议选择计算自旋轨道耦合时，完全关掉对称性（ISYM=0）。通常会从一个自旋方向到另一个自旋方向k点的设置会发生改变，进而恶化转换的结果（如果k点改变WA/ECAR将不会被正确地重新读取）。GGA_COMPAT通常需要，应该被设置，因为磁各向异性能量通常需要精确到亚meV数量级。当计算自旋轨道耦合，特别是磁各向异性时通常需要非常小心：能量差异非常小，k点的收敛冗长而且缓慢，需要耗费大量的计算时间。此外，这一特征-尽管长期存在于VASP中-在最新的版本中依然存在，你可以尝试频繁地升级发现这一点。不敢保证，你的结果是有用的！此处根据README文件做了一个小小的总结:20.11.2003:提出的GGA程序轻微的破坏了非正交体系晶胞的对称型。球面截断能应用于梯度及互逆空间中的所有中间结果。GGA引起的轻微的改变（通常每个原子0.1meV）,却对磁各项异性很重要。05.12.2003:继续…现在VASP.4.6默认旧的行为GGA_COMPAT=.TRUE.，新的行为将可以通过在INACR中设置GGA_COMPAT=.FALSE得至V。12.08.2003:主要的错误出现在symmetry.F和paw.F：非线性计算的对称性例程没有正确的执行。如果你阅读了以上内容，就会意识到在VASP.4.6和VASP.5.2版本中进行非线性计算推荐设置GGA_COMPAT=.FALSE.，这样可以提升GGA计算的数值精度。VASP:Non-collinearcalculationsandspinorbitcoupling:Spinors旋量wereincludedbyGeorgKresseintheVASPcode.Thecoderequiredforthetreatment处理ofnon-collinearmagneticstructureswaswrittenbyDavidHobbs,andspin-orbitcouplingwasimplemented实施、执行byOlivierLebacqandGeorgKresse.SpinorsareonlysupportedasofVASP.4.5.Subsections:分段、子章节、下一级栏目LNONCOLLINEAR-tagSupported支持asofVASP.4.5.ySettingLNONCOLLINEAR=.TRUE.intheINCARfileallowstoperformfullynon-collinearmagneticstructurecalculations.VASPiscapable有能力的ofreadingWAVECARandCHGCARfilesfromprevious之前的non-magnetic非磁orcollinear线性calculations,itishowevernotpossibletorotate旋转、转动themagneticfieldlocallyonselectedatoms.Hence因此,inpractice在实践中，werecommend推荐toperformnon-collinearcalculationsintwosteps:First,calculatethenonmagneticgroundstate基态andgenerateaWAVECARandCHGCARfile.Second,readtheWAVECARandCHGCARfile,andsupply提供initialmagneticmomentsbymeansoftheMAGMOMtag(compareSec.6.13).Foranon-collinearsetup,threevaluesmustbesuppliedforeachionintheMAGMOMline.Thethreeentriescorrespondtotheinitiallocalmagneticmomentforeachioninx,yandzdirectionrespectively.ThelineMAGMOM=100010Initialises赋初值themagneticmomentonthefirstatominthex-direction,andonthesecondatomintheydirection.Mind,thattheMAGMOMlinesuppliesinitialmagneticmomentsonlyifICHARG=2,oriftheCHGCARfilecontainsonlychargebutnomagnetisationdensity.LSORBIT=.TRUE.Switches接通、开启onspin-orbitcouplingandautomaticallysetsLNONCOLLINEAR=.TRUE..Thisoption选项、选择worksonlyforPAWpotentialsandisnotsupportedforultrasoftpseudopotentials.Ifspin-orbitcouplingisnotincluded,theenergydoesnotdependon依赖thedirectionofthemagneticmoment,i.e.也就是说rotating旋转allmagneticmoments磁矩bythesameangleresultsexactlyinthesameenergy.Hence因此thereisnoneedtodefinethespinquantizationaxis自旋量子化坐标轴,aslongas只要spin-orbitcouplingisnotincluded.Spin-orbitcoupling,however,couples一对、一双thespintothecrystalstructure.Spinorbitcouplingisswitchedon开启byselectingLSORBIT=.TRUE.SAXIS=s_xs_ys_z(quantisationaxisforspin自旋量子化轴)GGA_COMPAT=.FALSE.!applyspherical球面cutoff截断能ongradientfield梯度场wherethedefaultforSAXIS=(0+,0,l)(thenotation符号0+implies意味着aninfinitesimal无穷小smallpositivenumberinxdirection).TheflagGGA_COMPAT(seeSec.6.42)isoptional选项andshouldbesetwhensmallenergydifferencesinthesub副、下标meVregime体制、状态needtobecalculated(oftenthecaseformagneticanisotropycalculations磁各向异性计算).Allmagneticmomentsarenowgivenwithrespectto关于theaxis坐标轴(Sx,Sy,Sz),wherewehaveadopted应用theconvention“惯例0thatallmagneticmomentsandspinor-likequantitieswrittenorreadbyVASParegivenwithrespecttothisaxis.ThisincludestheMAGMOMlineintheINCARfile,thetotalandlocalmagnetizationsintheOUTCARandPROCARfile,thespinor自旋量-likeorbitalsintheWAVECARfile,andthemagnetizationdensityintheCHGCARfile.Withrespecttothecartesian笛卡尔latticevectorsthecomponents组件、部分ofthemagnetizationare(internally内部地、内在的)givenbym=cos(卩)cos(a)maxis-sin(a)maxis+sin(卩)*cos(a)maxisTOC\o"1-5"\h\zxxyzm=cos(卩)sin(a)m+cos(a)maxis+sin(卩)sin(a)maxisyxyzm=—sin(卩)mEs+cos(卩)mEszxzWheremaxisistheexternally夕卜部地visible可得到的，现有的，可见的magneticmomentHere,aistheanglebetweenthe/X入SAXISvector(s%,s『,sz)andthecartesianvectorx,and卩istheanglebetweenthevectorSAXISandthecartesianvectorz:snIs2+s2|a=atanT,p=atan~厂ssxzTheinverse倒转、翻转transformation转化、转换isgivenbymaxis=cos(P)cos(a)m+cos(P)sin(a)m+sin(P)myzxxmaxis=—sin(a)m+cos(a)mxyy-maxis=sin(P)cos(a)m+sin(P)sin(a)m+cos(P)m”xyzItiseasytoseethatforthedefault(sx,s『，sz)=(0+,0,l),bothanglesrepresentationissimplyequivalenttotheexternalrepresentation:arezero,i.e.0a0da=0.Inthiscase,theinternalm=maxis,m=maxis,m=maxisxxyyzzThesecondimportantcase,ismaxjs=°andmaxis=0.Inthiscasem=sin(卩)*cos(a)肌曲=maxiss/、：s2+s2+s2zzzXyzm=ym=cos(卩｝maxis=maxiss/s2+s2+s2zxzzxyzHence因此、今后nowthemagneticmomentisparalleltothevectorSAXIS.Thustherearetwowaystorotatethespinsinanarbitrary任意的direction,eitherbychangingtheinitialmagneticmomentsMAGMOMorbychangingSAXIS.Toinitialisecalculationswiththemagneticmomentparalleltoachosenvector(x,y,z),itistherefore因此possibletoeitherspecify扌旨定(assuming假定、假设asingleatominthecell)MAGMOM=xyz!localmagneticmomentinx,y,zSAXIS=001!quantisationaxisparalleltozorMAGMOM=00total_magnetic_moment!localmagneticmomentparalleltoSAXISSAXIS=xyz!quantisationaxisparalleltovector(x,y,z)Bothsetupsshouldinprincipleyieldexactlythesameenergy,butforimplementation实现reasonsthesecondmethodisusuallymoreprecise精确.ThesecondmethodalsoallowstoreadapreexistingWAVECARfile(fromacollinearornoncollinearrun),andtocontinuethecalculationwithadifferentspinorientation.WhenanoncollinearWAVECARfileisread,thespinisassumed假定tobeparalleltoSAXIS(hence因此VASPwillinitiallyreportamagneticmomentinthez-directiononly).Therecommended被推荐的procedure过程、步骤forthecalculationofmagneticanisotropiesistherefore因而、表示结果(pleasecheckthesectiononLMAXMIX6.63):StartwithacollinearcalculationandcalculateaWAVECARandCHGCARfile.AddthetagsLSORBIT=.TRUE.ICHARG=11!nonselfconsistentrun,readCHGCARLMAXMIX=4!fordelementsincreaseLMAXMIXto4,f:LMAXMIX=6!youneedtosetLMAXMIXalreadyinthecollinearcalculationSAXIS=xyz!directionofthemagneticfieldNBANDS=2*numberofbandsofcollinearrunGGA_COMPAT=.FALSE.!applysphericalcutoffongradientfieldVASPreadsintheWAVECARandCHGCARfiles,aligns排列thespinquantizationaxisparalleltoSAXIS,whichimplies意味着thatthemagneticfieldisnowparalleltoSAXIS,andperformsanonselfconsistentcalculation.Bycomparingtheenergiesfordifferentorientationsthemagneticanisotropycanbedetermined确定.Pleasemind,thatacompletelyselfconsistentcalculation(ICHARG=1)isinprinciple大体上、原则上alsopossiblewithVASP,butthiswouldallowthethespinorwavefunctionstorotatefromtheirinitialorientationparalleltoSAXISuntilthecorrectgroundstateisobtained,i.e.untilthemagneticmomentisparalleltotheeasyaxis(?=theeasymagneticaxis).Inpracticethisrotationwillbeslow,sincereorientation再定位ofthespingains获得littleenergy.Thereforeiftheconvergence收敛criterion标准isnottootight,sensible明智的resultsmightbeobtainedevenforfullyselfconsistentcalculations(inthefewcaseswehavetried可靠地，试验过的selfconsistentcyworkedwithoutproblems).Beverycarefulwithsymmetry.Werecommend建议toswitchoff关掉symmetry(ISYM=0)altogether完全地，whenspinorbitcouplingisselected.Oftenthek-pointsetchangesfromonetotheotherspinorientation,worsening恶化thetransferabilityoftheresults(alsotheWAVECARfilecannotberereadproperly正确地ifthenumberofk-pointschanges).TheflagGGA_COMPATisusuallyrequiredandshouldbeset,sincemagneticanisotropyenergiesareofteninthesubmeVregime(seeSec.6.42).Generallybeextremely非常careful,whenusingspinorbitcouplingand,specifically特别地,magneticanisotropies:energydifferencesaretiny微小的，k-pointconvergence收敛istedious冗长乏味andslow,andthecomputertimemightbehuge.Additionally此夕卜，thisfeature这一特征--althoughlongimplemented应用inVASP--isstillinalatebetast,aageyoumightdeducefrom推断，从…得出结论thefrequent频繁的updates升级、更新.Nopromise允诺,thatyourresultswillbeuseful!Hereisasmallsummary总结fromtheREADMEfile:20.11.2003:Thepresent提出GGAroutine程序breaksthesymmetryslightly轻微地fornonorthorhombic正交晶系cells.Aspherical球面的cutoffisnowimposedon应用于thegradientsandallintermediate中间的resultsinreciprocal互逆space.ThischangestheGGAresultsslightly(usuallyby0.1meVperatom),butisimportantformagneticanisotropies.05.12.2003:continue...NowVASP.4.6defaultstotheoldbehaviorGGA_COMPAT=.TRUE.,thenewbehaviorcanbeobtainedbysettingGGA_COMPAT=.FALSE.intheINCARfile.12.08.2003:MAJOR主要的BUG故障FIX固定insymmetry.Fandpaw.F:fornon-collinearcalculationsthesymmetryroutines惯例didnotworkproperly正确地Ifyouhavereadthepreviouslines,youwillrealizethatitisrecommended推荐tosetGGA_COMPAT=.FALSE.fornoncollinearcalculationsinVASP.4.6andVASP.5.2,sincethisimprovesthenumericalprecisionofGGAcalculations.degree:BSc：BachelorofScience理科学士MDMS：master硕士PhD：DoctorofPhilosophy博士学位AAAssociatedegreeofArts大专文科学位AASAssociatedegreeofArtsandScience大专文理科学位ASAssociatedegreeofScience大专理科学位BABachelorofArts文科本科学位BSBachelorofScience理科本科学位MAMasterofArts文科硕士MBAMasterofBusinessAdministration商学硕士MSMasterofScience理科硕士Ph.D.DoctorofPhiolosphy哲学(通才)博士JDDoctorofJournalism新闻博士MDDoctorofMedicine医学博士DVMDoctorofVeterinry兽医博士CalltoZHEGVfailedErrorEDDDAV:CalltoZHEGVfailed.Returncode=1318Theearliersolutionsuggestedbyadmin(DOS操作系统中，超级管理员。行政、管理)(suppressing制止的theline#defineUSE_ZHEEVXindavidson.F,subrot.F,andwavpre_noio.FandrecompilingVASP)doesnotwork,i.e.thesameerrormessages,andthesameindication迹象、表示ofZHEGVfailure,stillappear出现.ImayaddnowthattheproblemappearsbothwiththelapackwhichcomeswithVASPandwithasystem-nativelapacklibrary.Thewarningsgivensuggestthattheproblemactuallyappearsatanearlierstage阶段,inwhichamatrixisgeneratedwithinadequate不适当的valueswhichmakeitnonhermitian,andconsequentlyZHEGVfailsevenifworkingcorrectly;thesolutionthuswouldnotbetoavoidusingZHEGV,buttoavoidanincorrectgenerationofthesaidmatrix.Cansomeonegiveanideatoreallysolvetheproblem?答：Pleasetryifitworksbyadding"LSCALAPACK=.FALSE."inyourINCAR.对策：grepLSCALAPACKOUTCAR空设置：LSCALAPACK=.FALSE问：No;adding"LSCALAPACK=.FALSE."inINCARmakesnodfference,theproblemcontinuesthesame.问：Iwassuccessfultofixthisproblem解决此问题byusingIALGO=48insteadofIALGO=Default。unfortunately,whenisetIALGO=48,thenewwarningis：WARNINGinEDDRMM:calltoZHEGVfailed,returncode=6314howtosolvethisproblem?whatdoes"ZBRENT"mean?对策：grepIALGOOUTCARIALGO=68algorithm(INCARALGO=Fast)设置：IALGO=48pleasetryoneofthefollowing:chooseadifferentalgorithmforionicoptimization(IBRION=1)采用准牛顿算法来优化原子位置setADDGRID=.True.inINCAR(onlyforvaspreleases发布管理、释放、豁免445andnewer)对策：grepADDGRIDOUTCAR空grepIBRIONOUTCARIBRION=2ionicrelax:0-MD1-quasi-New2-CG设置ADDGRID=.True.InINCAR设置IBRION=1inINCAR错误：internalERRORRSPHER:runningoutofbuffer001310nonlr.F:OutofbufferRSPHER得到的CONTCAR是空的！结构优化出现错误：Internal内部的、内在的ERRORRSPHER:runningoutofbuffer缓冲001310nonlr.F:OutofbufferRSPHER解决：将-NPAR=-1-修改成4(或者-2)-，问-题得以解决。分两步-Gsc-f非磁线性计算，bands读取-CHGCAR、WAVECAR-做非线性自旋轨道耦合计算)，能带计算出错：ERROR:whilereadingWAVECAR,-planewavecoefficients系数•changed5728628837Solution:Youhavetocontinuewiththeconverged收敛CHGCAR,becausemostprobably,youwillincrease增^加/change改变thek-meshtogetadenser密集的、浓厚的k-gridtocalculatetheDOSaccurately.Then,WAVECARwillnotbereadcorrectlybecausethewavefunction-coefficients波函数-系数arestored存储k-pointwise明智的concerning涉及theREADerrorofCHGCAR:pleasecheckwhethertheFFTmesheshavechanged.pleasemakesurethattheCHGCARreallyisintheworkingdirectory目录atruntime运行时间thefftmeshesofCHGCARarecompatible兼容的Themainpointsisinthissentence"planewavecoefficientschanged",IthinktheISMEARyouusedinscfandnoscfprocessisdifferent,therefore,theplanewavecoefficientschangedinthesetwoprocessisnotidentical完全相同的.YoucanfindthevaluesofNGXF,NGYFandNGZFintheCHGCARorOUTCARofthescf,andthenaddthesethreeparametersintheINCARofthenonscf.OK,theproblemisresolved.在静态计算的CHGCAR或者OUTCAR中找到NGXF,NGYF和NGZF，将这些参数加到非静态计算的INCAR中:grepNGXFOUTCARdimensionx,y,zNGXF=64NGYF=64NGZF=840supportgridNGXF=64NGYF=64NGZF=840NGXF,Y,Zisequivalenttoacutoffof25.43,25.43,25.05a.u.对策：在能带计算INCAR中加入NGXF=64——NG¥F=-64-NGZF=-840修改之后，bands中出现错误：ERROR:-non-collinear-calculations-require-that-VASPis-compiled-without-theflag-DNGXhalfand--DNGZhalf解决：待解决！网上经验：noncollinearcalculationsrequirethatVASPiscompiledwithouttheflag-DNGXhalfand-DNGZhalf.一、请加入SOC1)INCAR中加入LNONCOLLINEAR=.True.LSORBIT=.True.LORBMOM=.True.ISYM=-1(?不对，ISYM取值0,l,2,3)【SAXIS=自旋轴方向；MAGMOM=每个原子的初始磁矩值】2）不要忘记toincludeSOC,please1)addthefollowinglinestoINCARLNONCOLLINEAR二.True.SimplecubeMCenterofanedgeRCornerpointXCenterofafaceFace-centeredcubicLSORBIT二.True.SAXIS=#pleasegivethespinquantizationaxishere,like001forthez-axis)MAGMOM=#pleasegiveatripletofnumbersforeachatomhere,andpleasehavealookatthemanual(chapternon-collinearcalculationsandspin-orbittag)onhowthedirectionofthemagneticmomentshastobedefinedwithrespecttothes