CCL:G: G03 "Error in orthogonal basis" + Bad length for file.



 Sent to CCL by: Guilherme Menegon Arantes [dinamica~!~webcable.com.br]
 Hi there list,
 Any tip about the following error seen in Gaussian 03 (B.04) on a SGI
 Altix running SUSE LINUX Enterprise Server 9 (ia64), kernel
 2.6.5-7.201-rtgfx:
  (Enter /usr/local/g03/l302.exe)
  NPDir=0 NMtPBC=     1 NCelOv=     1 NCel=       1 NClECP=     1 NCelD=
 1
          NCelK=      1 NCelE2=     1 NClLst=     1 CellRange=     0.0.
  One-electron integrals computed using PRISM.
  Error in orthogonal basis= 4.77D+01 with   295 functions, increasing
 threshold to 2.00D-06.
  Error in orthogonal basis= 9.49D+01 with   295 functions, increasing
 threshold to 4.00D-06.
 ...
  Error in orthogonal basis= 9.49D+01 with   295 functions, increasing
 threshold to 5.24D-01.
  Error in orthogonal basis= 9.49D+01 with   288 functions, increasing
 threshold to 1.05D+00.
  Error in orthogonal basis= 2.27D+01 with   145 functions, increasing
 threshold to 2.10D+00.
  Error in orthogonal basis= 2.22D+01 with    30 functions, increasing
 threshold to 4.19D+00.
  Error in orthogonal basis= 2.13D+00 with     1 functions, increasing
 threshold to 8.39D+00.
  Bad length for file.
 FileIO: IOper= 1 IFilNo(1)=  -685 Len=           0 IPos=           0 Q=
 1142739344
  dumping /fiocom/, unit = 1 NFiles =    46 SizExt =    524288 WInBlk =
 2048
                    defal = T LstWrd =     9723904 FType=2 FMxFil=10000
 ...
  Base         20480
  End          65536
  End1         65536
  Wr Pntr      20480
  Rd Pntr      20480
  Length       45056
  Error termination in NtrErr:
  NtrErr Called from FileIO.
 ---------------------------------------------------------
 This job route session:
  %nosave
  %mem=97440512
  %chk=temp/oi.chk
  #p mp2/6-311+G(df,p) maxdisk=1920mw scan counterpoise=2
 ---------------------------------------------------------
 What puzzles me is that this looks like a disk/scratch file error, but
 there is plenty of space and the files never get bigger than 1GB (not at
 this stage of the calculation!).
 This "Error in orthogonal basis" appears to be caused by a linear
 dependence in the basis set (not seen in G98 for the same job).
 If I tunr off this lin. dep. check (using  iop(3/32=2)), the SCF does
 not converge.
 I googled around and also looked in the Gaussian web site, but no help.
 Thanks for your attention,
 Guilherme Menegon Arantes            Sao Paulo, Brasil
 ______________________________________________________