Failed to converge with new code cpa2021v02.tgz (October 12, 2025) but works good with old ones

Posted on : April 28, 2026 (Tue) 08:20:21

by Jerry Lee

Dear All,

I found the following input can converged with the old version cpa2002v010.tgz (22Feb2022), but failed to converge with the latest version cpa2021v02.tgz (October 12, 2025). What is wrong with the new code?

Thanks
Jerry

c----------------------Co3Sn2S2----------------------------------
c***********************************************
c *** Self-consistent calculation ***
go data/Co3Sn2S2_pot
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
rhb 10.1686 1.0 1.0 60.128 60.128 60.128
c------------------------------------------------------------
c edelt ewdth reltyp sdftyp magtyp record
0.0001 2.4 srals pbeasa mag 2nd
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 8 300 0.025
c------------------------------------------------------------
c ntyp
4
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Co 1 1 0.0 2 27 100
S 1 1 0.0 2 16 100
Sn1 1 1 0.0 2 50 100
Sn2 1 1 0.0 2 50 100
c------------------------------------------------------------
c natm
7
c------------------------------------------------------------
c atmicx atmtyp
1/2a 0 0 Co
0 1/2b 0 Co
0 0 1/2c Co
0.2805a 0.2805b 0.2805c S
0.7195a 0.7195b 0.7195c S
1/2a 1/2b 1/2c Sn1
0 0 0 Sn2
c------------------------------------------------------------

[Re:02] Failed to converge with new code cpa2021v02.tgz (October 12, 2025) but works good with old ones

Posted on : April 28, 2026 (Tue) 23:08:22

by Jerry Lee

Thanks you for your suggestion! But it still cannot converge with the new code cpa2021v02.tgz (October 12, 2025). Anyway I will switch back to the old version cpa2002v010.tgz (22Feb2022)

c----------------------Co3Sn2S2----------------------------------
c***********************************************
c *** Self-consistent calculation ***
go data/Co3Sn2S2_pot
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
rhb 10.1686 1.0 1.0 60.128 60.128 60.128
c------------------------------------------------------------
c edelt ewdth reltyp sdftyp magtyp record
0.0001 2.4 srals pbeasa mag init
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 8 300 0.025
c------------------------------------------------------------
c ntyp
4
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Co 1 1 0.02 2 27 100
S 1 1 0.0 2 16 100
Sn1 1 1 0.0 2 50 100
Sn2 1 1 0.0 2 50 100
c------------------------------------------------------------
c natm
7
c------------------------------------------------------------
c atmicx atmtyp
1/2a 0 0 Co
0 1/2b 0 Co
0 0 1/2c Co
0.2805a 0.2805b 0.2805c S
0.7195a 0.7195b 0.7195c S
1/2a 1/2b 1/2c Sn1
0 0 0 Sn2
c------------------------------------------------------------
c *** Self-consistent calculation ***
go data/Co3Sn2S2_pot
c------------------------------------------------------------
c brvtyp a c/a b/a alpha beta gamma
rhb 10.1686 1.0 1.0 60.128 60.128 60.128
c------------------------------------------------------------
c edelt ewdth reltyp sdftyp magtyp record
0.0001 2.4 srals pbeasa mag 2nd
c------------------------------------------------------------
c outtyp bzqlty maxitr pmix
update 8 300 0.025
c------------------------------------------------------------
c ntyp
4
c------------------------------------------------------------
c type ncmp rmt field mxl anclr conc
Co 1 1 0.0 2 27 100
S 1 1 0.0 2 16 100
Sn1 1 1 0.0 2 50 100
Sn2 1 1 0.0 2 50 100
c------------------------------------------------------------
c natm
7
c------------------------------------------------------------
c atmicx atmtyp
1/2a 0 0 Co
0 1/2b 0 Co
0 0 1/2c Co
0.2805a 0.2805b 0.2805c S
0.7195a 0.7195b 0.7195c S
1/2a 1/2b 1/2c Sn1
0 0 0 Sn2
c------------------------------------------------------------

[Re:01] Failed to converge with new code cpa2021v02.tgz (October 12, 2025) but works good with old ones

Posted on : April 28, 2026 (Tue) 10:09:12

by root

Try the following input as the first step
(you need a blank before Co, otherwise it is understood as a comment line)

.....
c type ncmp rmt field mxl anclr conc
Co 1 1 0.02 2 27 100
S 1 1 0.0 2 16 100
Sn1 1 1 0.0 2 50 100
Sn2 1 1 0.0 2 50 100
.....

Namely, give a magnetic field on the Co site. Run the SCF cycle until some stable solution is attained (no necessary of convergence). As the second step, switch off the field and continue normal scf calculations. You will get a convergence.