Author Topic: Reordering for coordinate list  (Read 1077 times)

0 Members and 1 Guest are viewing this topic.

Offline zhangguangping

  • ATK Guru
  • ****
  • Posts: 182
  • Country: cn
  • Reputation: 2
    • View Profile
Reordering for coordinate list
« on: April 12, 2017, 08:27 »
Dear developer,

I find the reorder function in coordinate list only order one of the components, but the other two components are out of order (in version 2015.1). However, sometimes, the ordering of the other two components maybe also useful. Therefore, a function that can set a primary order component, a second one and the third one is proposed in the feature version.

With best regards,

/Guangping Zhang

Offline Anders Blom

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 4972
  • Country: dk
  • Reputation: 78
    • View Profile
    • QuantumWise
Re: Reordering for coordinate list
« Reply #1 on: April 12, 2017, 08:37 »
I haven't tried myself, but does it work if you first sort by your 3rd criterion, then 2nd, then 1st?
I.e., when you sort the following times, does it keep the sort order of the previous one (as best possible)?

Offline Anders Blom

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 4972
  • Country: dk
  • Reputation: 78
    • View Profile
    • QuantumWise
Re: Reordering for coordinate list
« Reply #2 on: April 12, 2017, 09:23 »
Oh, I just remembered, there is a separate plugin for sorting! Under Coordinate Tools also, see "Sort" :)

Offline zhangguangping

  • ATK Guru
  • ****
  • Posts: 182
  • Country: cn
  • Reputation: 2
    • View Profile
Re: Reordering for coordinate list
« Reply #3 on: April 13, 2017, 04:01 »
Oh, I just remembered, there is a separate plugin for sorting! Under Coordinate Tools also, see "Sort" :)

Dear Anders Blom,

That's what I wanted. When I constructe a molecular junction and then remove a fraction of the electrode at each side to do a extend molecule optimization. After the optimization, I usually restore the removed electrode to recover the molecular junction for a basis to contructed a device using "Device From Bulk". However, at this time, VNL almost can not find the "Left electrode length" and "Right electrode length" due to the desctruction of the periodic condition of electrode in C direction. So, a sort of the coordinates according to the same rule will be helpful in this case. And I have tried the "Sort" function, it works in my case to construct a device from a optimized central region.

Thanks very much for your help.

With best regards,

/Guang-Ping Zhang

Offline Anders Blom

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 4972
  • Country: dk
  • Reputation: 78
    • View Profile
    • QuantumWise
Re: Reordering for coordinate list
« Reply #4 on: April 13, 2017, 04:27 »
Glad it works for you, however it's good practice, and actually more or less required, that you constrain the electrode copy part of the central region when you optimize the device. Even if your approach works, there is a risk that you don't have a precisely crystalline geometry in the electrodes, which introduces additional scattering.

Offline zhangguangping

  • ATK Guru
  • ****
  • Posts: 182
  • Country: cn
  • Reputation: 2
    • View Profile
Re: Reordering for coordinate list
« Reply #5 on: April 13, 2017, 05:18 »
Glad it works for you, however it's good practice, and actually more or less required, that you constrain the electrode copy part of the central region when you optimize the device. Even if your approach works, there is a risk that you don't have a precisely crystalline geometry in the electrodes, which introduces additional scattering.

The procedure I do is as follows.
1) First, I contructed a junction with six layers of, for example Au(111), at each side, and a moleucle sandwiched between in (configuration I).  2) And then, copy configuration I for twice. There, I will have three configuaration I.
3) For the first one, I remove the two outermost Au(111) layers at each side to contructed a "extend molecule" consisting a molecule and four Au(111) layers at each side for the following geometric optimizing (configuration II).
4) For the second one, I remove all the atoms except the two lelftmost layers of Au(111) (configuration III).
5) For the third one, I remove all atoms except the two rightmost layers of Au(111) (configuration IV).
6) During the optimization of configuration II, the two leftmost layers of Au(111) will be fixed and the two rightmost layers of Au(111) will set to be rigid. All other atoms will be fully relaxed, the optimized geometry is called configuration II-opt.
7) After the optimization, I will recover the removed two layers of Au(111) at leftmost using "Drop" function by droping configration III on configuaration II-opt, and get configuration V).
8 ) Figure out the tranlation in xyz directions for the two rightmost layer of Au(111) in configuration II-opt with respect to configuration II.
9) Applied the the tranlation got in step ( 8 ) to configuration IV, and get configuration IV-trans.
10) Recover the two rightmost layers of Au(111) by droping configuration IV-trans on configuration II-opt. Then get configuration VI.

At last, I need to use "Device from bulk" to construct a real two probe system based on configuration VI. But, usually it fails due to the destruction of the peridic condition in C direction from a rearrangement of the atoms (I do not know at which step it occurs, maybe during the "Drop" step). In order to make VNL get the periodic condition in C direction, a reorder of the atoms in configuration VI will be needed. But, the "reorder" in "Coordinate List" will only make a sort for one componet. However, "Sort" function will make it.

I think the procedure above will be simple and quick one for constructing a two probe system in VNL, and is there a risk to have a precisely crystalline geometry in the electrodes, that introduces additional scattering?

With best regards,

/Guangping Zhang

Offline zhangguangping

  • ATK Guru
  • ****
  • Posts: 182
  • Country: cn
  • Reputation: 2
    • View Profile
Re: Reordering for coordinate list
« Reply #6 on: April 15, 2017, 12:01 »
Glad it works for you, however it's good practice, and actually more or less required, that you constrain the electrode copy part of the central region when you optimize the device. Even if your approach works, there is a risk that you don't have a precisely crystalline geometry in the electrodes, which introduces additional scattering.

The procedure I do is as follows.
1) First, I contructed a junction with six layers of, for example Au(111), at each side, and a moleucle sandwiched between in (configuration I).  2) And then, copy configuration I for twice. There, I will have three configuaration I.
3) For the first one, I remove the two outermost Au(111) layers at each side to contructed a "extend molecule" consisting a molecule and four Au(111) layers at each side for the following geometric optimizing (configuration II).
4) For the second one, I remove all the atoms except the two lelftmost layers of Au(111) (configuration III).
5) For the third one, I remove all atoms except the two rightmost layers of Au(111) (configuration IV).
6) During the optimization of configuration II, the two leftmost layers of Au(111) will be fixed and the two rightmost layers of Au(111) will set to be rigid. All other atoms will be fully relaxed, the optimized geometry is called configuration II-opt.
7) After the optimization, I will recover the removed two layers of Au(111) at leftmost using "Drop" function by droping configration III on configuaration II-opt, and get configuration V).
8 ) Figure out the tranlation in xyz directions for the two rightmost layer of Au(111) in configuration II-opt with respect to configuration II.
9) Applied the the tranlation got in step ( 8 ) to configuration IV, and get configuration IV-trans.
10) Recover the two rightmost layers of Au(111) by droping configuration IV-trans on configuration II-opt. Then get configuration VI.

At last, I need to use "Device from bulk" to construct a real two probe system based on configuration VI. But, usually it fails due to the destruction of the peridic condition in C direction from a rearrangement of the atoms (I do not know at which step it occurs, maybe during the "Drop" step). In order to make VNL get the periodic condition in C direction, a reorder of the atoms in configuration VI will be needed. But, the "reorder" in "Coordinate List" will only make a sort for one componet. However, "Sort" function will make it.

I think the procedure above will be simple and quick one for constructing a two probe system in VNL, and is there a risk to have a precisely crystalline geometry in the electrodes, that introduces additional scattering?

With best regards,

/Guangping Zhang

I currently find that strictly following steps 1-10 can give a good basis for the two-probe system construction and it does not need a sort action.

With best regards,

/Guangping Zhang

Offline Anders Blom

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 4972
  • Country: dk
  • Reputation: 78
    • View Profile
    • QuantumWise
Re: Reordering for coordinate list
« Reply #7 on: April 15, 2017, 22:41 »
Your method works, but I think I see an easier approach (although it requires a little bit longer computation...). Couldn't you just use Conf 1 and make the left 5 layers fixed, and the right-most 5 layers rigid?

Offline zhangguangping

  • ATK Guru
  • ****
  • Posts: 182
  • Country: cn
  • Reputation: 2
    • View Profile
Re: Reordering for coordinate list
« Reply #8 on: April 16, 2017, 05:10 »
Your method works, but I think I see an easier approach (although it requires a little bit longer computation...). Couldn't you just use Conf 1 and make the left 5 layers fixed, and the right-most 5 layers rigid?

Dear Anders Blom,

Yes, your method is much easier, but it require a bit more computations, what is I want to reduced.

With best regards,

/Guangping Zhang