How to increase the maximum number of atoms ?

Posted on : November 16, 2016 (Wed) 22:04:56

by Andrei Baranovskiy

Dear KKR Administrator,

I've got an unexpected problem when was trying to increase the maximum number of atoms in the unit cell. I need it to be 13 atoms, but after changing natmmax parameter to 13 I've got an error during compilation of the program (see below). Is there any way to fix this problem ? May be it is possible to decrease some parameters without affecting drastically on calculation quality ?

Thank you in advance,
Andrei.

source/specx.o: In function `MAIN__':
specx.f:(.text+0x3d1): relocation truncated to fit: R_X86_64_PC32 against `.bss'
specx.f:(.text+0x3ff): relocation truncated to fit: R_X86_64_PC32 against `.bss'
specx.f:(.text+0x712): relocation truncated to fit: R_X86_64_32 against `.bss'
specx.f:(.text+0x778): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0x799): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0x7a5): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0x7cd): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0x7e5): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0x919): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0x935): relocation truncated to fit: R_X86_64_32S against `.bss'
specx.f:(.text+0xa41): additional relocation overflows omitted from the output
collect2: ld returned 1 exit status
make: *** [specx] Error 1

[Re:06] How to increase the maximum number of atoms ?

Posted on : November 24, 2016 (Thu) 00:07:26

by Andrei Baranovskiy

Dear KKR Administrator,

Thank you very much for your answer.
Now it works.

Best regards,
Andrei.

[Re:05] How to increase the maximum number of atoms ?

Posted on : November 20, 2016 (Sun) 13:33:37

by Administrator

It seems that you are taking a half of the normal lattice constant and twice as large primitive unit cell vectors as normal ones. It may work, but the atomic positions in your input seem completely different from normal YbB12 that has the UB12 structure. Also, you have to increase the number of atoms from 7 to 13 in the input. Also bzqlty=7 is too large. I would take bzqlty=1 for starting and then after some convergence switch to bxqlty=2. [mix=0.02 is till fine for this system.
Another a bit subtle point is that, if non-magnetic calculation is intended (as is shown in the input), spin-orbit coupling is not very much meaningful. The reason in in this KKR-code only the spin diagonal parts are taken into account and thus not making much sense for non-magnetic cases. Also using the open core configuration might be safe, i.e. taking only up to d states for valence.
Standard input file for your system will be seen in the bottom.
Concerning the parameters for specx.f, I will suggest the following:

parameter
& (natmmx=13, ncmpmx=2, msizmx=124, mxlmx=4, nk1x=64, nk3x=0,
& msex=201, ngmx=15, nrpmx=650, ngpmx=650, npmx=350, msr=400)

If you would like to plot the dispersion curves, "nk1x=0, nk3x=300" is recommended.

The standard input for this system is the following:

c------------------------------------------------------------
go data/ybb12
c brvtyp a c/a b/a alpha beta gamma
fcc 14.122, , , , , ,
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 2.0 sra mjw nmag 2nd
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 1 200 0.02
c------------------------------------------------------------
c ntyp
2
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
B 1 1.0 0.0 2
5 100
Yb 1 1.8 0.0 3
70 100
c------------------------------------------------------------
c natm
13
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
1/2 0.1707 0.1707 B
1/2 -0.1707 0.1707 B
1/2 0.1707 -0.1707 B
1/2 -0.1707 -0.1707 B
0.1707 1/2 0.1707 B
-0.1707 1/2 0.1707 B
0.1707 1/2 -0.1707 B
-0.1707 1/2 -0.1707 B
0.1707 0.1707 1/2 B
-0.1707 0.1707 1/2 B
0.1707 -0.1707 1/2 B
-0.1707 -0.1707 1/2 B
0 0 0 Yb

[Re:04] How to increase the maximum number of atoms ?

Posted on : November 17, 2016 (Thu) 23:23:00

by Andrei Baranovskiy

Sure, I'll be very appreciated if you check it.
Best regards,
Andrei.

c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
aux
0.0 1.0 1.0
1.0 0.0 1.0
1.0 1.0 0.0
7.061
c------------------------------------------------------------
c edelt ewidth reltyp sdftyp magtyp record
0.001 2.0 srals mjw nmag init
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 7 200 0.005
c------------------------------------------------------------
c ntyp
2
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
B 1 1.0 0.0 2
5 100
Yb 1 1.0 0.0 3
70 100
c------------------------------------------------------------
c natm
7
c------------------------------------------------------------
c atmicx(in the unit of a) atmtyp
-0.15860 0.50000 0.50000 B
1.15860 0.50000 0.50000 B
0.50000 1.15860 -0158600 B
0.50000 -0.15860 1.15860 B
0.50000 0.50000 -0.15860 B
1.15860 -0.15860 0.50000 B
-0.15860 1.15860 0.50000 B
0.50000 0.50000 1.15860 B
0.50000 -0.15860 0.50000 B
1.15860 0.50000 -0.15860 B
0.50000 1.15860 0.50000 B
-0.15860 0.50000 1.15860 B
0.00000 0.00000 0.00000 Yb
Type :quit<Enter> to exit Vim

[Re:03] How to increase the maximum number of atoms ?

Posted on : November 17, 2016 (Thu) 11:20:44

by Administrator

If you do not mind, let me show your input. Then I can show you suitable parameter sets.

KKR administrator

[Re:02] How to increase the maximum number of atoms ?

Posted on : November 16, 2016 (Wed) 23:26:13

by Andrei Baranovskiy

Dear KKR Administrator,
Thank you very much for your answer.
The list of parameter are presented below.

parameter
& (natmmx=13, ncmpmx=6, msizmx=96, mxlmx=4, nk1x=1550, nk3x=300,
& msex=201, ngmx=15, nrpmx=650, ngpmx=650, npmx=350, msr=400)
c
c--- the following part is used for all cases.
parameter
& (lengx=msex, ndmx=natmmx*(natmmx-1)+1, mxlfac=5-mxlmx,
& lastmx=2900d0/3d0**mxlfac, ntypmx=ncmpmx, nkmx=nk1x+nk3x)

[Re:01] How to increase the maximum number of atoms ?

Posted on : November 16, 2016 (Wed) 22:29:33

by Administrator

This happens when the memory of your system cannot afford the memory size. In such case you have to reduce the size of memory that are not used. Please let me see the parameter statement in source/specx.f .

KKR administrator