From owner-nwchem-users@emsl.pnl.gov Thu Nov 22 06:24:48 2007 Received: from odyssey.emsl.pnl.gov (localhost [127.0.0.1]) by odyssey.emsl.pnl.gov (8.14.1/8.14.1) with ESMTP id lAMEOl1Y005884 for ; Thu, 22 Nov 2007 06:24:48 -0800 (PST) Received: (from majordom@localhost) by odyssey.emsl.pnl.gov (8.14.1/8.14.1/Submit) id lAMEOlrm005883 for nwchem-users-outgoing-0915; Thu, 22 Nov 2007 06:24:47 -0800 (PST) X-Authentication-Warning: odyssey.emsl.pnl.gov: majordom set sender to owner-nwchem-users@emsl.pnl.gov using -f X-Ironport-SG: OK_Domains X-Ironport-SBRS: 5.2 X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: AgAAAM8fRUfRVaKymGdsb2JhbACPMAEBBwQEK4EP X-IronPort-AV: E=Sophos;i="4.21,451,1188802800"; d="scan'208";a="38088760" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=hBAn3DMgjA3qs+i/Ibd+ARiv+/zmbBUcRiaCDMuyMlM=; b=nhxFqYH7bJ4gkAe5utq7liy9Ker349wP1QpEc4AtMMF2HWBU4MDUe1ubnbI8wLHEsenN1sLzBLAHXiZaIreP4oHe/dPINeFxzi/WYmfxAB1yRJkA5G4kwh9cMN+rgVLF4DLDmrxlb0OUsBYZcj5Nh+UMDyDswv3GSTyGZ3YGMNQ= DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=RwU4Nvv10rIEzUdD2LWeGOZNjChWILFmb+xabA7E/pzqTJT7aEMBMaWCjT7tEK+wEXintlbef0IHS54i8JxDOPJV1upz1pfeiUcmGp9h5+4sFwttKdrv5ZMznUPhBFYdmm5qztYqutbMdmMOZFMBsEOUoY5q3cj/iDGigt1ZIXw= Message-ID: <96f4bb620711220624x4f452a50v1f3421899aab6d82@mail.gmail.com> Date: Thu, 22 Nov 2007 08:24:42 -0600 From: "Jeff Hammond" To: "Erik McNellis" , "NWChem Users" Subject: Re: [NWCHEM] Some NWChem problem examples In-Reply-To: <47454DDE.6040706@fhi-berlin.mpg.de> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <47454DDE.6040706@fhi-berlin.mpg.de> Sender: owner-nwchem-users@emsl.pnl.gov Precedence: bulk Erik, One cannot apply a pair theory (MP2) to a molecule with only one electron. When there are no beta electrons, the code will try to create null arrays. The MP2 NO problem is because the relevant subroutine is called only when task=3D=3Dgradient. You have found one good way to get MP2 NOs for the time being. Jeff On Nov 22, 2007 3:37 AM, Erik McNellis wrote: > Dear List, > > in the course of recent work I have come across a few examples of input t= hat at least my NWChem 5.0 binaries seem to have problems with, and I would= very much appreciate if any of you could test the > attached input files on your machines and NWChem binaries, and / or scan = the input for user errors. I have the corresponding output for each input, = but limitations on attachment sizes won't let me > send it with this email. > > I'd like to begin by commenting on the NWChem FAQ: > The MP2 FAQ section states, that in order to do an MP2 property analysis,= one must give the MP2 natural orbitals as input to the property module, sp= ecifying > > "task mp2 > > property > > vectors .mp2nos > end > > task property" > > in the input file. The only syntax that works in my tests is > > " > > task mp2 gradient; task mp2 property" > > Perhaps the FAQ needs to be updated? > > In the attached input files, I have used the 'def-TZVP' AO basis set of A= hlrichs et al, explicitly specified for reproducibility. It is a triple zet= a basis set with a single set of polarization > functions, and no diffuse functions. The first example is a DFT/PBE calcu= lation of a Cu dimer, with OEP Self Interaction Correction. The integration= grid is set to 'xfine' in accord with the > documentation. The program exits after three SCF iterations with the erro= r > > "0:0:wrong what value:: 0 > Last System Error Message from Task 0:: Numerical result out of range > 0: ARMCI aborting 0 (0). > system error message: Numerical result out of range" > > The second is an MP2 calculation of a single hydrogen atom, which complet= es the UHF nicely, but fails on the MP2 with the message > > "0:0:nga_create_config:ga_set_data:wrong dimension specified:: 0 > Last System Error Message from Task 0:: No such file or directory > 0: ARMCI aborting 0 (0). > system error message: No such file or directory" > > The third file is a larger calculation (~ 10 h on 16 Power5+ CPUs), compu= ting the RI-MP2 error for a 28 Cu cluster. The AO basis used is the same as= above, and the ri-mp2 basis is the corresponding > optimized RI-MP2 auxiliary basis set as implemented in Turbomole (see F. = Weigend, M. H=E4ser, H. Patzelt and R. Ahlrichs; Chem. Phys. Letters 294, 1= 43 (1998)). This auxiliary basis performs excellently > with NWChem for small organic molecules, with RI-MP2 errors in the ~ 20 m= eV range as expected from literature. The attached input file runs perfectl= y fine, returning no errors, but reports the > following MP2 and RI-MP2 energies: > > SCF energy -45890.312126220204 > Correlation energy -9.977321223375 > Total MP2 energy -45900.289447443582 > > Reference energy: -45890.312126220203936 > Correlation energy: -227.482320499693515 > Total RI-MP2 energy: -46117.794446719897678 > > That is, the RI error in this case is approximately 217.5 Ha, or 5918.5 e= V. The RI-MP2 calculation also uses ~ 20% more RAM than allowed (actual tot= al memory usage =3D 1.2 * (no. processes) * (set total > memory)). I have verified that this excess memory and enormous RI error a= re independent of input options, NWChem versions 4.7 or 5.0, operating syst= em and architecture, building with vendor or > reference BLAS and fortran I/O buffers. > Of course, a possible explanation is that RI-MP2 simply cannot do large t= ransition metal clusters. However, I am able to do smaller clusters (6 - 22= Cu atoms) contained within this structure, with > vanishing RI errors, with the same basis set specification. Asking around= in the community also leads me to believe that no such behavior is observe= d with this basis set and similar systems in Turbomole. > > Finally, I have a question regarding the MP2 routines: > As I understand the present state of the MP2 methods (NWChem 5.0), one ca= n do direct MP2 on RHF ground states, regular MP2 on RHF or UHF ground stat= es, and RI-MP2 on RHF, UHF or ROHF ground states. A > user computing an open shell system then has the choice of suffering poss= ible spin contamination in a regular MP2 calculation, or the RI error in an= ROHF + RI-MP2 calculation. > Are there any plans to implement ROHF references in the regular and direc= t MP2 methods? > > Thank you very much, > > Erik McNellis > > -- > --------------------------------------------------------------------- > Erik McNellis, Ph.D. student > > Fritz-Haber-Institut, Abt. Theorie PHONE: +49 30 8413 4817 > Faradayweg 4-6 FAX: +49 30 8413 4701 > D-14195 Berlin (Germany) http://www.fhi-berlin.mpg.de/th/th.html > ----------------------key:1-0.0735-11600-23.05:fhi------------------- > > > > > > > echo > > start Cu_dimer_oep > > title "Cu dimer sic OEP" > > dft > xc xpbe96 cpbe96 > grid xfine > sic oep > end > > basis "ao basis" spherical noprint > Cu S > 377518.799230000 0.00022812 > 56589.984311000 0.00176880 > 12878.711706000 0.00919935 > 3645.375214300 0.03741102 > 1187.007294500 0.12189874 > 426.464219020 0.28983901 > 165.706601640 0.41531872 > 65.598942707 0.21905799 > Cu S > 414.412658110 -0.02468253 > 128.320560390 -0.11716827 > 20.622089750 0.55301316 > 8.782122604 0.52242719 > Cu S > 13.741372006 -0.22736062 > 2.243124683 0.71761211 > Cu S > 0.893705491 1.00000000 > Cu S > 0.108366995 1.00000000 > Cu S > 0.038806178 1.00000000 > Cu P > 2034.759669200 0.00235248 > 481.904681060 0.01913407 > 154.674829630 0.09017111 > 57.740576969 0.26063285 > 23.099052811 0.42093486 > 9.388247859 0.24344615 > Cu P > 37.596171210 -0.02899109 > 5.124069081 0.54919084 > 2.011999609 0.93793330 > Cu P > 0.738606860 1.00000000 > Cu D > 74.129460637 0.01436322 > 21.359842587 0.08662818 > 7.499524054 0.25631431 > 2.760139417 0.40374062 > Cu D > 0.953620612 0.39427042 > Cu D > 0.284208625 0.23091147 > Cu P > 0.155065000 1.00000000 > end > > geometry noautoz noautosym > Cu 1.282867195 0.000000000 0.000000000 > Cu -1.282867206 0.000000000 0.000000000 > end > > task dft > > echo > > start H_atom_mp2 > > title "MP2 of single H atom" > > scf > nopen 1; uhf > end > > basis "ao basis" spherical noprint > H S > 34.061341000 0.00602520 > 5.123574600 0.04502109 > 1.164662600 0.20189726 > H S > 0.327230410 1.00000000 > H S > 0.103072410 1.00000000 > H P > 0.800000000 1.00000000 > end > > geometry noautoz noautosym > H 0.000000000 0.000000000 0.000000000 > end > > task mp2 > > echo > > start 28_Cu_rimp2; title "28_Cu_rimp2" > > memory global 850 mb total 2550 mb > > scf > maxiter 100 > semidirect memsize 63750000 filesize 0 > tol2e 1e-12 > adapt off > end > > mp2 > freeze atomic > tight > end > > geometry noautosym noautoz noprint > Cu 2.773767105 0.360323264 -2.082346434 > Cu -1.074834502 -0.380340451 0.000000557 > Cu -1.074834513 -1.861667881 -2.082346434 > Cu 0.208032715 1.841650694 0.000000557 > Cu -2.357701686 1.841650694 0.000000557 > Cu -1.074834480 2.582314409 -2.082346434 > Cu 4.056634289 -1.861667881 -2.082346434 > Cu 0.208032682 -2.602331596 0.000000557 > Cu 0.208036519 -4.083665691 -2.082346434 > Cu -2.357701697 0.360323264 -2.082346434 > Cu -3.640576611 -1.861667881 -2.082346434 > Cu 0.208032704 0.360323264 -2.082346434 > Cu 1.490899899 -0.380340451 0.000000557 > Cu 0.208032693 -1.121004166 -4.177252816 > Cu 0.208028889 4.804312219 -2.082346434 > Cu 1.490899921 2.582314409 -2.082346434 > Cu -3.640568881 2.582314409 -2.082346434 > Cu 1.490899888 -1.861667881 -2.082346434 > Cu 1.490899910 1.100986979 -4.177252816 > Cu 1.490896084 4.063648504 0.000000557 > Cu 2.773767116 1.841650694 0.000000557 > Cu -1.074834491 1.100986979 -4.177252816 > Cu 2.773767083 -2.602331596 0.000000557 > Cu -1.074838317 4.063648504 0.000000557 > Cu -2.357709416 -2.602331596 0.000000557 > Cu 4.056634300 -0.380340451 0.000000557 > Cu 4.056642019 2.582314409 -2.082346434 > Cu -3.640576600 -0.380340451 0.000000557 > end > > basis "ao basis" spherical noprint > Cu S > 377518.799230000 0.00022812 > 56589.984311000 0.00176880 > 12878.711706000 0.00919935 > 3645.375214300 0.03741102 > 1187.007294500 0.12189874 > 426.464219020 0.28983901 > 165.706601640 0.41531872 > 65.598942707 0.21905799 > Cu S > 414.412658110 -0.02468253 > 128.320560390 -0.11716827 > 20.622089750 0.55301316 > 8.782122604 0.52242719 > Cu S > 13.741372006 -0.22736062 > 2.243124683 0.71761211 > Cu S > 0.893705491 1.00000000 > Cu S > 0.108366995 1.00000000 > Cu S > 0.038806178 1.00000000 > Cu P > 2034.759669200 0.00235248 > 481.904681060 0.01913407 > 154.674829630 0.09017111 > 57.740576969 0.26063285 > 23.099052811 0.42093486 > 9.388247859 0.24344615 > Cu P > 37.596171210 -0.02899109 > 5.124069081 0.54919084 > 2.011999609 0.93793330 > Cu P > 0.738606860 1.00000000 > Cu D > 74.129460637 0.01436322 > 21.359842587 0.08662818 > 7.499524054 0.25631431 > 2.760139417 0.40374062 > Cu D > 0.953620612 0.39427042 > Cu D > 0.284208625 0.23091147 > Cu P > 0.155065000 1.00000000 > end > > task direct_mp2 > > basis "ri-mp2 basis" spherical noprint > Cu S > 76.618887000 0.22127009 > 35.844205000 1.10771089 > 17.435980000 -1.23596756 > Cu S > 8.790108000 1.00000000 > Cu S > 4.575692000 1.00000000 > Cu S > 2.449357000 1.00000000 > Cu S > 1.342247000 1.00000000 > Cu S > 0.749382000 1.00000000 > Cu S > 0.424079000 1.00000000 > Cu S > 0.241964000 1.00000000 > Cu S > 0.138436000 1.00000000 > Cu P > 214.965070000 -0.38582926 > 49.560551000 1.55957840 > Cu P > 21.134817000 -1.17913540 > Cu P > 5.354381200 1.89982860 > Cu P > 2.197932500 1.00000000 > Cu P > 1.366181900 1.00000000 > Cu P > 0.600377880 1.00000000 > Cu P > 0.332775310 1.00000000 > Cu P > 0.146738560 1.00000000 > Cu D > 106.442650000 2.09841882 > Cu D > 23.194836000 10.00024900 > Cu D > 8.841183800 0.94747908 > Cu D > 4.571022700 1.26590806 > Cu D > 1.822179900 1.00000000 > Cu D > 0.850462610 1.00000000 > Cu D > 0.387499580 1.00000000 > Cu F > 52.133642000 -1.27232142 > 13.246909000 1.15816535 > 6.284814800 2.20111275 > Cu F > 3.050580800 0.70886158 > Cu F > 1.470033500 1.00000000 > Cu F > 0.506644560 1.00000000 > Cu G > 17.805104000 1.25317348 > Cu G > 6.023657900 1.31975806 > Cu G > 2.062614900 -0.16398042 > Cu G > 0.702209360 1.00000000 > end > > task rimp2 > > --=20 Jeff Hammond The University of Chicago