| 查看: 1508 | 回复: 2 | ||
[求助]
VASP计算出来的BLOCH 波函数的相位问题。 已有1人参与
|
|
以下是我发给VASP-support 的邮件 Dear Sir or Madam, Recently, I tried to output the wavefunction Un (k, r) ( n is the energy band, k is the crystal momentum) calculated by VASP. As a complex wave function, Un(k, r) should be continuous in respect to k and r, physically. While, I found that when r is fixed, Un(k, r) is not continuous in respect to k. I mean, although the absolute value of Un(k, r) is continuous in respect to k, the phase of Un(k, r) is not (as shown in the figure below). I checked the self-consistency scheme of VASP, I saw that the wave function is initialized using a random number generator, and the convergence criteria is related to energy. I wonder it is the random number generator that introduces random phases into the wavefunction Un(k, r) at last. And, this random phase won’t influence the convergence standard. Can you tell me whether what I think is right? Anyway, I think the phases of the wavefunction are very important physically. So, I want to get some advices from vasp-support about how to deal with the random phases. 大概意思就是BLOCH 波函数Un (k, r) 是个复数，但是VASP计算出来的波函数的相位对晶格动量K 不是连续的，这在物理上是不对的。 但是VASP给我回复的邮件好敷衍，就两句话。 please note that WAVECAR only contains the wavefunctions COEFFICIENTS, not the wavefuction itself. the random number generator initiates some starting values of these coefficients。 it has nothing to do with random phase. 他回复的东西完全没用嘛。有没有人比较精通VASP计算原理的，来说说这个random的phase到底来自于计算的哪些步骤。我觉得可能是来自于初始赋值以及迭代过程中矩阵对角化。 如何才能解决这个随机相位的问题呢？ |
回复此楼» 猜你喜欢
科研孤儿太难了
已经有6人回复
-
时间戳变了，能看出什么问题？
已经有23人回复
-
只有每年这种时候来逛逛小木虫
已经有22人回复
-
时间戳又变了
已经有14人回复
-
今天系统多次维护，明天很可能放榜！
已经有10人回复
-
放榜前的不淡定
已经有12人回复
-
言之凿凿，明天官网会通知放榜？
已经有16人回复
-
重要消息，中午系统在维护
已经有12人回复
-
哪位老哥知道今年的国自然具体哪一天放榜？
已经有20人回复
-
今天基金会出结果吗？20260819
已经有15人回复
2楼2017-03-04 07:28:27
ra2ghgzh
木虫 (正式写手)
学痴
- 物理EPI: 1
- 应助: 113 (高中生)
- 金币: 2131.9
- 散金: 5
- 红花: 33
- 帖子: 774
- 在线: 445.5小时
- 虫号: 3739328
- 注册: 2015-03-15
- 性别: GG
- 专业: 原子和分子物理
3楼2017-03-04 16:39:18
40