“解决一个又一个报错，计算路上的坑将越来越少”
之前关于wannier90的更新：
《一文搞定VASP+wannier90构建紧束缚模型》
《VASP+wannier90 构建紧束缚模型（补充）》
《QE+wannier90构建紧束缚模型》
《超越wannier90——构建紧束缚模型》
《wanniertools中表面能带投影问题与解决》
《确定wannier90能量窗口脚本》
《wanniertools中电子/声子的NumOccupied参数》
《[视频] SCDM: A unified approach to Wannier localisation》
开始之前先说一下wannier90能量窗口脚本一节中，engwin.py脚本运行报错的原因在于下面备注的地方。
def get_energies(xml_name):# VASP EIGENVAL formatif xml_name == "EIGENVAL":with open("EIGENVAL", "r") as eigenval:raw_content = eigenval.readlines()content = raw_content[7:]entries = []for i, line in enumerate(content):if len(line.split()) == 4:entries.append(i)eng_full = []for i in entries:eng_ki = []for line in content[i+1:]:#if len(line.split()) == 2: # 之前的文件if len(line.split()) == 3: # 这里之前是2，修改为3即可。eng_ki.append(float(line.split()[1]))else:breakeng_full.append(eng_ki)
wannier90常见报错汇总
1. wannier90.wout中出现:
Reading overlaps fromwannier90.mmn : File generated byVASP: unknown system
Error readingwannier90.mmn: 2 1 12 0 0 0 0
Exiting.......
Neighbour not found
原因：一般是因为wannier90.win是从其他地方拷贝的，不是vasp直接生成的，那么你的wannier90.win里面的k点网格可能不匹配（主要是k点的顺序，不是k点的数目）。
什么情况容易报这个错误？
同样的参数，不加soc运行正常，加soc以后报错，通常是因为在不加soc的时候没有设置ISYM=-1，k点简并比较高，生成的wannier90.win中的k-mesh与ISYM=-1时的k-mesh不同导致。
解决办法：删除wannier90.win里面的k-mesh部分，重新运行VASP程序。
wannier90.win中的k-mesh部分：
mp_grid = 12 12 1begin kpoints0.000000000000 0.000000000000 0.0000000000000.083333333333 0.000000000000 0.0000000000000.166666666667 0.000000000000 0.0000000000000.250000000000 0.000000000000 0.0000000000000.333333333333 0.000000000000 0.0000000000000.416666666667 0.000000000000 0.0000000000000.500000000000 0.000000000000 0.000000000000…end kpoints
2. wannier90.werr文件中出现：
Wannier90: Execution started on 2Jun2021 at09:02:24
Exiting.......
param_get_projection: too many projectionsdefined
原因：num_wann 与设置投影数不匹配。
解决办法：不加soc：num_wann 等于每种原子投影轨道数之和
soc：num_wann 等于两倍的每种原子投影轨道数之和
3. Wannier90.werr文件中出现：
Wannier90: Execution started on 2Jun2021 at09:06:32
Found a mismatch in wannier90.eig
Wanted band : 1 found band : 96
Wanted kpoint: 2 found kpoint: 1
A common cause of this error is using thewrong
number of bands. Check your input files.
If your pseudopotentials have shallow corestates remember
to account for these electrons.
Exiting.......
param_read: mismatch in wannier90.eig
原因：num_bands与vasp中NBANDS不匹配
解决办法：grep NBANDS OUTCAR 查看 NBANDS数，并且设置num_bands = NBANDS
另外，注意exclude_bands这个参数，num_bands = NBANDS – exclude_bands
4. Wannier90.werr中出现：
Wannier90: Execution started on 2Jun2021 at09:10:00
Exiting.......
Error: restart requested but wannier90.chkfile not found
原因：你加了restart = plot 这个命令，但却没有wannier90.chk这个文件。Restart是需要读取wannier90.chk的。
解决办法：注释掉restart = plot，然后重新运行wannier90.x wannier90 &得到wannier90.chk.
5. wannier90.wout中出现：
Time to calculate interpolated bandstructure 0.604 (sec)
Exiting.......
wrong irdist_ws
Error: examine the output/error file fordetails
原因：目前不太清楚原因，可能是2.1版本的一个bug，因为1.2版本没问题。在write_hr = true 写出wannier90_hr.dat的时候会出现，不会影响wannier90_hr.dat的写出。
下面是找到的一个疑似官方回答：
On 23 Feb 2018, at12:21, Mohammed Ghadiyali <m786g at live.co.uk> wrote:
> Dear all,
>
> I 'mtrying to generate wannier centres for single, bi and tri layer graphene, touse willson's loop method to calculate Z2 invariant via wainner tools.
>
> Hence I 'mincluding spin orbital coupling, which prohibits plotting of the wanniercentres so the way I can check the accuracy is by comparing the band-structuresgenerated by wannier90 and Quantum ESPRESSO.
>
> Forgraphene I am able to reproduce the band structure with certain deavations, butfor bi and tri layer graphene the calcualtions are failing, gernally with theerror: wrong irdist_ws
We have madesome improvements to this area of the code since the last release (there isalso a new parameter (ws_distance_tol). So you might want to try taking thedevelopment version from github to see if that fixes your issue.
https://github.com/wannier-developers/wannier90
Note that the development version also plotsspinor wave functions (if you also patch pwscf with the version of pw2wannier90.f90supplied).
Jonathan
本文暂时没有评论，来添加一个吧(●'◡'●)