Segmentational fault when ntry > num_sites

[GoogleCodeExporter]

icc + Intel MKL
FC_FLAGS  = -m64 -warn all -O3 -unroll
CXX_FLAGS = -m64 -Wall -O3 -unroll 

with the attached input files
----
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source      

ggeom              0000000000677249  Unknown               Unknown  Unknown
ggeom              0000000000675B1E  Unknown               Unknown  Unknown
ggeom              00000000006257A2  Unknown               Unknown  Unknown
ggeom              00000000005DFCC3  Unknown               Unknown  Unknown
ggeom              00000000005E3C9B  Unknown               Unknown  Unknown
libpthread.so.0    00007F72C30B2340  Unknown               Unknown  Unknown
ggeom              000000000045C113  Unknown               Unknown  Unknown
ggeom              0000000000403F03  Unknown               Unknown  Unknown
ggeom              0000000000403B26  Unknown               Unknown  Unknown
libc.so.6          00007F72C29F7EC5  Unknown               Unknown  Unknown
ggeom              0000000000403A19  Unknown               Unknown  Unknown

----
[1] same input file but with mu = 1 or mu = 2 work fine
[2] Disabling #BONDS, #SYMM, #PHASE does not remove the bug
[3] Does not depend on the matrix multiplication method -DDQMC_CKB or 
-DDQMC_ASQRD both give same output.

[4] When I use debugging flags I get
----
forrtl: severe (408): fort: (3): Subscript #1 of the array SITELIST has value 0 
which is less than the lower bound of 1

Image              PC                Routine            Line        Source      

ggeom              000000000054F308  dqmc_hubbard_mp_d        1339  
dqmc_hubbard.F90
ggeom              0000000000404933  MAIN__                     74  ggeom.F90
ggeom              0000000000403D76  Unknown               Unknown  Unknown
libc.so.6          00007FFE5F4D7EC5  Unknown               Unknown  Unknown
ggeom              0000000000403C69  Unknown               Unknown  Unknown
----

Original issue reported on code.google.com by iglovi...@gmail.com on 27 Aug 2014 at 9:28

Attachments:

• L80.0_mu0_1409173962.88.in
• chain_t1_tp1_u4_nx8_ny48.geom

[GoogleCodeExporter]

In your input file, ntry (number of sites for global move)  is set to be 30, 
which exceeds the total number of sites (n=8).

I am not sure why global moves are required here, as U=4 is not terribly large.

Original comment by cxc639 on 27 Aug 2014 at 9:46

[GoogleCodeExporter]

Setting ntry = [0,1,2,3,4,5,6,7,8] works fine on my machine.

Original comment by cxc639 on 27 Aug 2014 at 9:46

[GoogleCodeExporter]

I see. Thank you.

I am setting priority to low, but not closing the bug, because I believe code 
should check input parameters for consistency and

ntry < num_sites should be one of these checks

Original comment by iglovi...@gmail.com on 27 Aug 2014 at 9:58

• Changed title: Segmentational fault when ntry > num_sites
• Added labels: Priority-Low
• Removed labels: Priority-High

[GoogleCodeExporter]

I agree we should check the inputs for ntry > nsites, have the code
issue a warning and stop the job.  That's preferable to
a mysterious 'seg fault' especially for someone not in our group.

Likewise, it might be good to have QUEST report when it finds duplicate
pairs of neighbors:  in a ladder geometry the pbc duplicates the
rung hopping.  I remember a long time ago asking Simone whether
QUEST would do the hopping twice so t=2 or just once in such situations.
If I recall even Simopne was not sure so I just checked at U=0.
I have even forgotten what I concluded!  Anyway, it would be nice
if QUEST wrote something to output to alert the user.

Richard Scalettar
Professor, Physics Department
One Shields Ave.
University of California, Davis 95616
fax   530-752-4717
email scalettar@physics.ucdavis.edu
http://scalettar.physics.ucdavis.edu/

Original comment by scalet...@physics.ucdavis.edu on 27 Aug 2014 at 10:24

[GoogleCodeExporter]

I changed the type of issue from "defect" for "enhancement" as this is not 
really a bug of the code.

Original comment by cxc639 on 28 Aug 2014 at 6:04

• Added labels: Type-Enhancement
• Removed labels: Type-Defect

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Input parameter check for ntry has been implemented in ggeom.F90.

Original comment by cxc639 on 28 Aug 2014 at 6:07

• Changed state: Accepted

[GoogleCodeExporter]

Thank you. I verify this.

I am closing bug report.

Original comment by iglovi...@gmail.com on 28 Aug 2014 at 7:43

• Changed state: Fixed