Recent changes to bugs

mac os x compilation

Anonymous — Mon, 08 Apr 2013 21:43:48 -0000

Under Mac OS X 10.6.8 with icc (ICC) 11.1 20091130 with XCode 3.2.2, psi4 builds incorrectly with message "dyld: Library not loaded: libmkl_intel_lp64.dylib" due to incorrect linking.

segfault in CCSD geometry optimzation with analytical grad

Michael Banck — Tue, 13 Sep 2011 22:43:53 -0000

The following input file crashes on psicode-3.4.0 (see also http://bugs.debian.org/641470\):

% 6-31G** H2O Test optimization calculation
psi: (
label = "6-31G** SCF H2O"
jobtype = opt
wfn = ccsd
reference = rhf
dertype = first
basis = "6-31G**"
num_threads = 4
memory = (2048 MB)
zmat = (
o
h 1 roh
h 1 roh 2 ahoh
)
zvars = (
roh 0.96031231
ahoh 104.09437511
)
)

Output (on x86-64):

The PSI3 Execution Driver
PSI3 will perform a RHF CCSD optimisation via analytic gradients.
The following programs will be executed:
input
REPEAT 40
cints
cscf
transqt2
ccsort
ccenergy
cchbar
cclambda
ccdensity
oeprop
transqt --backtr
cints --deriv1
optking --opt_step
END
psiclean

input
REPEAT 40
CYCLE 1
cints
cscf
transqt2
ccsort
ccenergy
cchbar
cclambda
ccdensity
oeprop
transqt --backtr
cints --deriv1
optking --opt_step

Command optking --opt_step was terminated with signal 11

When I try to run the above input on my ThinkPad (x86), I get:

The PSI3 Execution Driver

PSI3 will perform a RHF CCSD optimization via analytic gradients.

The following programs will be executed:

input
REPEAT 40
cints
cscf
transqt2
ccsort
ccenergy
cchbar
cclambda
ccdensity
oeprop
transqt --backtr
cints --deriv1
optking --opt_step
END
psiclean

input
REPEAT (40)
CYCLE 1
cints
cscf
transqt2
ccsort
ccenergy
cchbar
cclambda
ccdensity
oeprop
transqt --backtr
Segmentation fault
cints --deriv1
Segmentation fault
optking --opt_step
Segmentation fault
CYCLE 2

and segfaults thereafter. Not sure if this is a Debian-related problem or possibly reproducable by you?

psi3 3.4.0 raises "Aborted" when chkpt_init() called

Daniel M. Healion — Thu, 11 Aug 2011 23:42:44 -0000

Hi,

I'm trying to link against psi3 to read in some integrals, and molecular properties from the checkpoint file. In the process, I'm getting an error message "Aborted" when the program calls chkpt_init(PSIO_OPEN_OLD). Attached is a stripped down example that reproduces the problem, along with a psi3 input file. Both psi3 and test.cc are compiled using g++ 4.3.6. test.cc is compiled and linked with the commands:

/home/dhealion/amd/bin/g++ -O3 -Wno-write-strings -g -Wall -I. -I/home/dhealion/amd/include -I/home/dhealion/amd/include -I/home/dhealion/amd/include/c++/4.3.6 -I/home/dhealion/amd/include/c++/4.3.6/x86_64-unknown-linux-gnu/32/ -c test.cc
/home/dhealion/amd/bin/g++ -O3 -Wno-write-strings -g -Wall -I. -I/home/dhealion/amd/include -I/home/dhealion/amd/include -I/home/dhealion/amd/include/c++/4.3.6 -I/home/dhealion/amd/include/c++/4.3.6/x86_64-unknown-linux-gnu/32/ -L/home/dhealion/amd/share/sage-4.6.2-amd/local/lib/ -L/home/dhealion/amd/lib64 -o test test.o -L. -L/home/dhealion/amd/lib -lPSI_qt -lPSI_chkpt -lPSI_iwl -lPSI_psio -lPSI_ciomr -lPSI_ipv1 -L/usr/lib64 -lblas -L/usr/lib64 -llapack -lblas -L/home/dhealion/amd/lib64 -lnewmat -L/home/dhealion/amd/lib64 -L/home/dhealion/amd/lib -L/usr/lib64 -L/usr/lib -L/usr/local/lib64 -L/usr/local/lib -lgfortran -lgcc_s -lpthread -lg2c -lm

and run using "psi3 -m; ./test"

If I run the code with the debugger, gdb says :

#0 0x0000003fe3430265 in raise () from /lib64/libc.so.6
#1 0x0000003fe3431d10 in abort () from /lib64/libc.so.6
#2 0x0000000000405e90 in psi::PSIO::get_numvols(unsigned int) ()
#3 0x00000000006117a8 in std::string::_Rep::_S_empty_rep_storage ()
#4 0x0000003fe3403a88 in ?? () from /lib64/libc.so.6
#5 0x0000000000000000 in ?? ()

I can't figure out if there's something I'm doing wrong. Thank you very much,

-Daniel Healion
dhealion<at>uci<dot>edu

Electrostatic Potential

Herma Cuppen — Tue, 26 Jul 2011 12:56:37 -0000

I think I found a bug in the calculation of the electrostatic potential on a 2D grid (oeprop, grid =1 )

In grid_oeprop.cc there are six redundant iy terms in the computation of the nuclear contribution to the electrostatic potential over a grid (around line 190)

x = grid_origin[0] + grid_step_x[0]*ix - grid_step_y[0];
y = grid_origin[1] + grid_step_x[1]*ix - grid_step_y[1];
z = grid_origin[2] + grid_step_x[2]*ix - grid_step_y[2];
for(iy=0;iy<=niy;iy++) {
x += grid_step_y[0];
y += grid_step_y[1];
z += grid_step_y[2];

cstdio includes, yyin?

Anonymous — Fri, 06 May 2011 17:18:57 -0000

I found myself downloading the 3.4.0 package on an ubuntu distribution and found a few bugs. The first ones are a few missing includes for cstdio where you guys use fprintf and fflush. These are easy, but the following is stymying me:

g++ -o ../../../bin/psi3 gprgid.o psi3.o parse_var.o runcmd.o ../../../lib/libPSI_ciomr.a ../../../lib/libPSI_ipv1.a ../../../lib/libPSI_psio.a
../../../lib/libPSI_ipv1.a(ip_error.o): In function `ip_error':
ip_error.cc:(.text+0x240): undefined reference to `showpos'
../../../lib/libPSI_ipv1.a(ip_read.o): In function `ip_append':
ip_read.cc:(.text+0x32f): undefined reference to `yyin'
ip_read.cc:(.text+0x39d): undefined reference to `yyin'
../../../lib/libPSI_ipv1.a(ip_read.o): In function `ip_initialize':
ip_read.cc:(.text+0x3d3): undefined reference to `yyin'
ip_read.cc:(.text+0x47d): undefined reference to `yyin'
../../../lib/libPSI_ipv1.a(y.tab.o): In function `yyparse':
y.tab.c:(.text+0x3b5): undefined reference to `yylex'
collect2: ld returned 1 exit status
make[3]: *** [../../../bin/psi3] Error 1

g++ --version
g++ (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
Copyright (C) 2010 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

make --version
GNU Make 3.81
Copyright (C) 2006 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.

This program built for x86_64-pc-linux-gnu

Maybe you can help me? I looked for yyin in all files under lib but ip_read is the only file containing any references to yyin and I don't know where it should be defined.

RPA calculation

Anonymous — Wed, 17 Nov 2010 04:05:04 -0000

How can I carry out RPA calculation for excitation energy? I cannot find RPA module applicable for EEs in the source.

Gentoo patches - manpages, parallel make & LDFLAGS

Anonymous — Tue, 06 Jul 2010 11:30:43 -0000

Hi,

I would like to offer and ask you to accept the patchset we (Gentoo Linux) use for psi-3.4.0.

http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/sci-chemistry/psi/files/

3.4.0-man_paths.patch & 3.4.0-destdir.patch : fixing installation of manpages respecting DESTDIR (see bug 2796858)
3.4.0-ldflags.patch: Respecting LDFLAGS from user enviroment (LDFLAGS are for linker optimisation not for libraries)
psi-3.4.0-parallel-make.patch & 3.4.0-parallel_fix.patch: allow parallel building witk make -j# (see bug 2629322
3.4.0-gcc-4.3.patch: missing prototype inclusion ( see bug 2819566 & 2819566)
3.4.0-dont-build-libint.patch & use-external-libint.patch: Rely on systems libint. That is what downstream probably wants more (see bug 2796844)

Thanks justin

Problems with large systems

Susi Lehtola — Tue, 29 Jun 2010 21:45:31 -0000

There is a big problem in psi3 with large systems. I reported a similar problem privately already a year ago, which had to do with the energy of the water dimer going below the HF limit with big basis sets (e.g. aug-cc-pV6Z and aug-cc-pV7Z):

Using inexpensive integrals
iter total energy delta E delta P diiser
1 -119.7624814769 1.576713e+02 0.000000e+00 0.000000e
+00
2 -127.8631581108 8.100677e+00 2.438166e+00 2.085692e
+00
3 -339.2258218641 2.113627e+02 2.154100e+00 2.155418e
+00
4 -330.9916500150 -8.234172e+00 2.577503e-02 2.829565e
+01
5 -330.1706733707 -8.209766e-01 3.413397e-03 2.844459e
+01
6 -329.6459138830 -5.247595e-01 1.952113e-03 2.862588e
+01
7 -329.2214636048 -4.244503e-01 1.848766e-03 2.883868e
+01
8 -477.6281288804 1.484067e+02 1.697961e+00 2.900804e
+01
9 -793.3899544816 3.157618e+02 2.219406e+00 3.030855e
+01

The same kind of mathematically impossible behavior is seen with small basis sets but a largeish system. For instance a cluster of 4 propanol molecules with the aug-cc-pVDZ basis set, having 48 centers, 688 AOs and 656 SOs, the solution clearly isn't working:

Using inexpensive integrals
iter total energy delta E delta P diiser
1 -514.7469213453 1.645142e+03 0.000000e+00 0.000000e+00
2 6.2407446392 -5.209877e+02 2.703199e-01 8.992384e-01
3 -444.3812893279 4.506220e+02 5.182276e-01 1.624830e+00
4 -486.5960997134 4.221481e+01 4.031823e-01 6.351848e-01
5 -409.3582185574 -7.723788e+01 3.646935e-01 7.309628e-01
6 -741.9444447401 3.325862e+02 4.537851e-01 1.218576e+00
7 -697.0429433284 -4.490150e+01 3.929403e-01 4.792830e-01

Also, I think the sign convention for delta E is a bit odd.

Change error reporting to be more precise

Susi Lehtola — Tue, 29 Jun 2010 20:58:09 -0000

Currently if there is not enough memory allocated for a calculation, psi3 fails with for instance:

$ psi3
input
cscf
St9exception
cints failed due to errors
CINTS error: St9exception
PSIO_ERROR: Can't find TOC Entry SO-basis Overlap Ints
PSIO_ERROR: unit = 35
PSIO_ERROR: 13 (no such TOC entry)

Command cscf has returned a fail status.Aborted

This is not very user-friendly; instead of a generic "Exception" cints should throw a "Out of memory" exception, and the rest of the calculation should not be performed at all.

Locally-correlated CIS calculation crashes on LINUX

Anonymous — Wed, 02 Jun 2010 16:36:48 -0000

I have found cis calculation with options "local=Ture" and "subgroup =c1" always crashes saying segmentation fault at least on Ubuntu 9.10, 10.04 and OpenSuSE 11.2. But on Mac OS X, the calculation always ends up with good result. How could I come up with an idea in order to get over this problem on LINUX?