UT theoretical chemistry code forum

I have use the LANCZOS methods to find the saddles for H atom moving on the SWCNT. Here is my INCAR and POSCAR,

nano h
1.00000000000000
15.0000000000000000 0.0000000000000000 0.0000000000000000
0.0000000000000000 15.0000000000000000 0.0000000000000000
0.0000000000000000 0.0000000000000000 4.2300000000000004
24 1
Direct
0.1345999999999990 0.1340666666666690 0.1011820330969260
0.1696000000000030 0.0454000000000008 0.6319148936170220
0.1829333333333310 0.0489999999999995 0.9926713947990540
0.9518666666666680 0.1843999999999970 0.0990543735224563
0.1244000000000010 0.1241333333333330 0.4614657210401920
0.0456666666666692 0.1695333333333340 0.6309692671394770
0.0487333333333311 0.1828000000000000 0.9919621749409020
0.8212666666666660 0.0474666666666650 0.0938534278959793
0.9550666666666670 0.1698666666666640 0.4593380614657220
0.8763999999999970 0.1238000000000030 0.6271867612293180
0.8655333333333320 0.1336666666666700 0.9877068557919630
0.8661999999999990 0.8635999999999980 0.1002364066193880
0.8322666666666690 0.0446000000000026 0.4555555555555560
0.8312000000000030 0.9541333333333309 0.6278959810874700
0.8169999999999999 0.9510666666666689 0.9884160756501220
0.0495333333333363 0.8163333333333330 0.1016548463356980
0.8762666666666660 0.8748000000000000 0.4605200945626460
0.9549333333333370 0.8300666666666670 0.6319148936170220
0.9512000000000000 0.8169999999999999 0.9926713947990540
0.1827333333333310 0.9510000000000010 0.1009456264775390
0.0457999999999998 0.8301333333333361 0.4621749408983430
0.1242666666666640 0.8758666666666670 0.6323877068557950
0.1342000000000000 0.8661999999999990 0.9931442080378260
0.1694666666666650 0.9545999999999990 0.4617021276595780
0.6892459999999990 0.0948939999999986 0.3927399527186790

SYSTEM=LANCZOS
IBRION=3
POTIM=0.0
NSW=500
EDIFF=1E-6
#SYMPREC=1E-6
EDIFFG=-0.02
PREC=Norm
ISMEAR=0
SIGMA=0.2
LREAL=Auto
IOPT=2

# LANCZOS PARAMETERS
ICHAIN = 3
Snl = 20
Sltol = 1E-2
SdR = 1E-3

The problems are (1) the calculations are not convergence, even I set EDIFFG=-0.02 which is more than your example INCAR; (2) the result CONTCAR is seems not more different from the POSCAR.
I get:
nano h
1.00000000000000
15.0000000000000000 0.0000000000000000 0.0000000000000000
0.0000000000000000 15.0000000000000000 0.0000000000000000
0.0000000000000000 0.0000000000000000 4.2300000000000004
24 1
Direct
0.1139987968661520 0.1134666887588572 0.1307697121135137
0.1546159361252957 0.0414278785876555 0.6304808867981254
0.1545962353224801 0.0414348892428534 0.9618310071029186
0.9593712421034472 0.1537085299349736 0.1298583998184256
0.1140159198719429 0.1134512645197510 0.4611433708353667
0.0422034678566349 0.1548974781584861 0.6307597820029432
0.0422251716174233 0.1549976573264892 0.9607345966820636
0.8454492665526793 0.0409903333476437 0.1282572539210104
0.9593136754357142 0.1535988510834564 0.4610600686357811
0.8879383085329859 0.1121208938235962 0.6297291605815287
0.8879837174886382 0.1120640791487155 0.9604282838598270
0.8880919532424553 0.8864574819658863 0.1299898208702474
0.8439535091796040 0.0414303887317226 0.4616013139125786
0.8466651170631023 0.9581445240078035 0.6296986500613305
0.8467997874923642 0.9581351927712382 0.9605340096581442
0.0423873402755044 0.8446159468588648 0.1309514622204074
0.8880898701881017 0.8866142955782063 0.4611589151796340
0.9594613507462291 0.8444072495756064 0.6308871885191110
0.9594339238330116 0.8443061179620226 0.9609184470745016
0.1542891456139330 0.9587898828342287 0.1304889789057125
0.0423949269550667 0.8446294493071701 0.4612424878814991
0.1134797885004990 0.8869257721121487 0.6305900181068651
0.1134539573129896 0.8869265891279567 0.9619020911293173
0.1543095424643909 0.9587741303932799 0.4621093406118883
0.6857479423888926 0.0961169094371604 0.3812177732493166

Can you help me how to converge to the saddle points,3x

The configuration in your POSCAR does not make much sense to me. It seems to contain 4 cluster of 6 C atoms, each separated by about 10 A. The single H atom is then about 2.5 A from the nearest C atom. I am not really surprised that you are having a hard time converging this structure. If I am mistaken, please correct me.

Ok, I took a closer look at your structure and it makes sense to me know (I thin), it relies on the periodicity of the cell. I assume you have it minimized as well.

Did you start with a reasonable MODECAR file or did you have VASP generate one randomly? It is always much better to start with a guess for the lowest mode rather than have to spend many iterations initially looking for the it. You had set Snl = 20, how many ionic calculations had you made? It is very well conceivable that all 20 would be spent on converging onto the initial lowest modes, if starting from a random guess. That would explain why you hadn't seen any considerable movement yet.

In this case it would be easy to come up with an guess for the lowest mode. Have the H atom be moving towards the surface and all the C atoms stationary.