A problem I had when using the CUSTOM function to define a CV

[Hang LYU]

Dear developers and users of Plumed,

   I am trying to repeat a metadynamics simulation of a nucleation process of Ge2Sb2Te5 reported in a paper(Adv. Funct. Mater. 2015, 25, 6407–6413). I follow their supporting infomation to define the CV. In their calculation, they used something similar to the local steinhardt order parameter to monitor the crystallinity of the system.

   

     The final CV is define in the following way:

    

     As shown by the fomula above, they randomly selected an atom i and calculated the switch function between i and j. The final CV is an average sum with the switch function fij as the weight. In order to implement this CV. I wrote the following lines in plumed.dat.

UNITS LENGTH=A TIME=fs

i: GROUP ATOMS=52 

j: GROUP ATOMS=1-72 

### To calculate the local steinhardt parameter ###

q6: Q6 SPECIES=1-72 SWITCH={RATIONAL R_0=3.86 D_0=0 NN=24 MM=48} 

lq6: LOCAL_Q6 SPECIES=q6 SWITCH={RATIONAL R_0=5.52 D_0=0 NN=24 MM=48}

### To calculate the local steinhardt parameter ###

### To get the fij in the final CV ###

con_mat: CONTACT_MATRIX GROUPA=j GROUPB=i SWITCH={RATIONAL R_0=7.82 D_0=0 NN=24 MM=48}

real_con_mat: COORDINATIONNUMBER WEIGHT=con_mat.w

### To get the fij in the final CV ###

### To get the final CV ###

m: CUSTOM ARG1=lq6 ARG2=real_con_mat FUNC=x*y PERIODIC=NO

CV: COMBINE ARG=m PERIODIC=NO

### To get the final CV ###

   However, when I test this input file using plumed driver, the following error was reported. Seems like that I cannot use the CUSTOM function to multiply lq6 and fij. Is there other command in plumed can be applied to obtaine the final CV.

Best wishes,

Hang LYU

[Gareth Tribello]

Hello

I am slightly confused by what you are doing here.  First of all are you using the master branch of PLUMED or hack-the-tree?

I don’t understand these lines:

con_mat: CONTACT_MATRIX GROUPA=j GROUPB=i SWITCH={RATIONAL R_0=7.82 D_0=0 NN=24 MM=48}

real_con_mat: COORDINATIONNUMBER WEIGHT=con_mat.w

You take the number of atoms from group i (which contains one atom) that are within the coordination sphere of group j (multiple atoms).   This quantity is a vector that can be 1 or zero.  Is this what you want?

If you send a frame and an example input maybe I can take a look later.  I need more detail though.

Good luck

Gareth

[Hang LYU]

Hello,

   Sorry for making the confusion. 

   From the LOCAL_Q6 command, i get a vector lq6 = [lq6.1,  lq6.2,  lq6.3,  lq6.4,  lq6.5,  lq6.6] (assume that there are 6 atoms in the system). 

   

   Following the formula in the paper, I need a vector fij consisting of 1 or 0 depending whether the distance between atom i and atom j is smaller than 7.82 Å. For example, this vector fij  is [1, 0, 0, 1, 1, 0]. Then I can use a dot-product to get the CV. CV = 1 × lq6.1 + 0 × lq6.2 + 0 × lq6.3 + 1 × lq6.4  + 1 × lq6.5  + 0 × lq6.6.

   The frame and the input file are also attach below.

Best wishes,

Hang LYU

[Gareth Tribello]

Hi 

Sorry for taking a while to get back to you on this one.  I have assumed that you are using the hack-the-tree branch.

Given this I would modify your input to the following:

UNITS LENGTH=A TIME=fs

  

a: GROUP ATOMS=52

o: GROUP ATOMS=1-72

q6: Q6 SPECIES=1-72 SWITCH={RATIONAL R_0=3.86 D_0=0 NN=24 MM=48}

lq6: LOCAL_Q6 SPECIES=q6 SWITCH={RATIONAL R_0=5.52 D_0=0 NN=24 MM=48}

dist: DISTANCES ORIGIN=a ATOMS=o

real_con_mat: LESS_THAN ARG1=dist SWITCH={RATIONAL R_0=7.82 D_0=0 NN=24 MM=48}

m: CUSTOM ARG1=lq6_av ARG2=real_con_mat FUNC=x*y PERIODIC=NO

CV1: COMBINE ARG=m PERIODIC=NO

Look at the log and notice that lq6_av is the normalised version of the local_q6 (i.e. divided by the coordination numbered thus between 0 and 1).  If you want to work with the unormalised version of the local_q6 you can use lq6 in the line labelled m.

I hope this helps

Good luck

Gareth

On 25 Mar 2021, at 17:21, Hang LYU <terryh...@gmail.com> wrote:

<plumed.inp>

[Hang LYU]

Dear Gareth,

   Sorry for my late reply. I have tried the code, and it works now.

   Thanks so much for your help.

Best wishes and thanks,
Hang LYU