Home

User manual for the Uppsala Quantum Chemistry

1. random f90 Module for different random number generators to be used in the diffusion quantum monte Carlo calculations Afunc f90 This is the function A_lri 11 12 Ax Bx Cx gamma defined at the top of page 245 in the Cook Book 5 RHF f90 Subroutine for performing restricted Hartree Fock calculations Fnurec f90 This subroutine calculates the values F_nu X see Cook Book p 280 5 for nu 0 1 2 NUMAX Through the recursion formula given on the bottom of p 280 of the Cook Book Fnu f90 The function F_nu x described in the Cook Book 5 on pages 244 and 280 is here cal culated from the Kummer confluent hypergeometric function See WIRE s Computational Molecular Science 2012 2 290 303 CISD 90 Subroutine for the construction and diagonalization of the configuration interaction Hamil tonian constructed from singles and doubles excitations getK f90 Subroutine calculating the the exchange matrix Ka by contracting the electron electron in tegrals ij kl with the density matrix Py getKv f90 Vectorized version of getK tf90 getJ f90 Subroutine calculating the the Hartree matrix J by contracting the electron electron inte grals ij kl with the density matrix Py getJv f90 Vectorized version of getJ f90 7 1 FILE MAP 39 gammaf f90 calculates the value of the GAMMA function T I 0 5 gfortan has the gamma function as an intrinsic function whereas ifort on osx does not seem to have it fj f90 The
2. R given the eigen vector C Le the value W r gradhforbitalval f90 Subroutine that calculates the gradient of a Hartree Fock orbital or a Kohn Sham orbital at the spatial point r R3 given the eigen vector C Le the gradient VV r laplacehforbitalval f90 Subroutine that calculates the laplacian of a Hartree Fock orbital or a Kohn Sham orbital at the spatial point r R3 given the eigen vector C Le the laplacian V W r det f90 This function calculates the determinant of a NxN matrix A by first transforming it into upper diagonal form 7 1 FILE MAP 43 stoexponent f90 Finds the exponent A and coeficient A of a slater type orbital Ae by fiting it to the gaussian basis function gto r at the inflexion point re i e at the point where gto r min The exponent A and coeficient A are calculated from the continuity con ditions 1 gto r Ae e 2 gto re Are gt e leading to gto r gto re and A gto r Je 9 e e 9t0 re slaterdet f90 This fuction calculates the slater determiant from a N Total Number of electrons b BAS Hartree Fock basis c C Fockian eigenvectors d r positions the electrons 3N long array e UP TRUE Spin up slater determinant is calculated otherwise spin down slater de terminant is to be calculated NS slater determinant dimension gradslaterdet f90 This subroutine calculates the gradient of a slater de
3. RUN33X RUN33Y RUN33Z contains the output and input three TDDFT calcu lations for H20 using the PBE functional and the 6 31G basis set See figure 5 1 for the resulting absorption spectrum 34 CHAPTER 5 EXAMPLE CALCULATIONS 1 25 T T T T T iL H O PBE 6 31G t 0 05 a u j 0 75 S O 0 5 0 25 eV Figure 5 1 Absorption spectrum calculated with uquantchem s TDDFT implementation Here 3 Dirac pulses where used to produce the spectrum A 6 31G basis set together with the PBE functional was used The corresponding input files are INPUTFILE RUN33X INPUTFILE RUN33Y and INPUTFILE RUN33Z The time step used was 0 05 a u Chapter 6 Utility Matlab programs for plotting In the sub directory V 31 UTILITYPROGRAMS there are a couple of simple matlab scripts that can help you visualize charge densities and atomic orbitals calculated with UQUANTCHEM 35 Chapter 7 For the developer In this chapter the different subdirectories of the directory V 31 and their respective content will be briefly discussed 7 1 File map The following subdirectories are included in the V 31 package 7 1 1 BASIS This directory contains the files of all the different gaussian basis sets that accompany this release of uquantchem 7 1 2 BLAS This directory contains the basic linear algebra sub programs BLAS attached to this release of uquantchem to enable the compilation of the code in the event that the
4. BE BE Bee eg A Be a es 15 Ah EDIR SLs plete on eo fh tye eh eth ge Bae Goa fee eat a a was ae haa ew 15 AD S SOMEGR erei m a a phd Oye ae he 15 4 2 9 NCHEBGAUSS ccc eos be A tea eee aan oe 15 42 10 NLEBEDEV ermitas Sie da bk BE DAL A badd ae 15 AD VIeiMOEDYN lata ct Beaded te O eGo Ske Ss Suet Bly eB Bray as ds is 16 42 12 RUBOMD gt o Auv id a Aa AREER SR Pe A 16 DAS SORTS TART a as a a a ete Ta 16 42 14 SDORDER 2 a A eee ee e A 16 AA SE A E AR 16 CONTENTS 4 2 16 4 2 17 4 2 18 4 2 19 4 2 20 4 2 21 4 2 22 4 2 23 4 2 24 4 2 25 4 2 26 4 2 27 4 2 28 4 2 29 4 2 30 4 2 31 4 2 32 4 2 33 4 2 34 4 2 35 4 2 36 4 2 37 4 2 38 4 2 39 4 2 40 4 2 41 4 2 42 4 2 43 4 2 44 4 2 45 4 2 46 4 2 47 4 2 48 4 2 49 4 2 50 4 2 51 4 2 52 4 2 53 4 2 54 4 2 55 4 2 56 4 2 57 4 2 58 4 2 59 4 2 60 4 2 61 4 2 62 3 KAPPA rnd Bud ty a AR Bh eas bate Micki Nive BS Bets Gs te at 16 LER SOF eae Png e AS cee aah Be Dy haere in UG e Thee analy E 17 ZEROSGCE TYPE 0 ia a hate Pers Ss ae Ss a e ee Oe a A 17 PIXANSGE Gasta es ds oh tlle bode Se cee be Seek oe aed 17 MOVIE r e 23 Moto tha oR oO Bg NE Goalie E o ot MMR AEBS BREE 17 WRITEONELY gt soeu Boao es a hie ae Se ee ee Ge ete a A 18 TEMPERATURE coco a OA OS et we gs de 18 GEORGE sama es ek Rd e A he nol A ae te a Bae ee A 18 PUGAYA 2 2 i028 Pe Rbk Ek a da e eee e Poh AA we AS 18 TORT Si NE ALD ede oy A 19 RELAXN s ai Goad Cd Bee bl OS tae we A 19
5. For the developer 7 1 File map BASIS feck ee oa tee ee see oe A BLAS it ts sera id Boe lapack 3 4 0 UTILITYPROGRAMS TESIS tudea a o pb eee README ji sis fi fo dre ae 7 1 1 7 1 2 7 1 3 7 1 4 7 1 5 7 1 6 7 1 7 7 1 8 7 1 9 7 1 10 manual pdf SERIALVERSION OPENMPVERSION MPI_VERSION CONTENTS Chapter 1 Introduction The Uppsala Quantum Chemistry UQUANTCHEM project was started in order to build a transparent from the point of view of a physicist development platform for implementing new computational and theoretical ideas in quantum chemistry The package is written in fortran90 Due to the ambition of transparency i e being able to see the physics in bedded within the source code the code might not be as fast as other similar codes Some of the initial inefficiency as been dealt with by parallelization which hopefully has resulted in a proof of principle level of computational speed which will be enough to test new ideas on medium sized molecules The ultimate goal is to use the platform of UQUANTCHEM to develop new computa tional schemes within the context of quantum Monte Carlo The UQUANTCHEM software is published under the General Public License Version 3 0 GPLv3 0 Thus any one is free to use the UQUANTCHEM software It is my sincere wish that you will enjoy using the UQUANTCHEM package as much as I have enjoyed writing it and that it might help any freshman to better understand the te
6. P G P E D LDA Here G 5 J 4K in the case of a RHF calculation and G 4J in the case of a LDA calculation P density matrix propagated by XLBOMD scheme D Ofu F P F P Fockian obtained using the density matrix P O step function u chemical potential h one electron hamiltonian and Eze exchange correlation energy functional 4 2 19 FIXNSCF Type Integer Default 1 This is a parameter that if set such that FIXNSCF gt 0 in a molecular dynamics calculation the number of scf cycles are kept constant and equal to FIXNSCF However for the first 10 time steps the total energies are fully converged with respect to the energy tolerance TOL 4 2 20 MOVIE Type Logical Default False Flag specifying weather or not to save the atomic positions of every every SAMPLERATE th time step in a Molecular Dynamics Calculatio If set to TRUE the atomic positions of every SAMPLERATE th time step will be saved to the file MOLDYNMOVIE xsf 18 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 2 21 WRITEONFLY Type Logical Default False If true the total energy kinetic energy potential energy and temperature will be written on the screen and into the file MOLDYNENERGY dat during the course of a Molecular Dynamics Calculation If FALSE the energies will be saved to the same file in the end of the calculation and no output will be written to the screen 4 2 22 TEMPERATURE Type Double P
7. Pasta Cu IORBNR 1 ds 4 13 Pa YU pIORBNR 2 Oia 4 14 4 15 After this a self consistent calculation is performed by minimizing the total energy with respect to the density matrix P PHOLE pEXC where PHOLE and PEXC are updated throughout the self consistent calculation and where the density matrix P is calculated from the eigenvectors of the Kohn Sham Fockian Hamiltonian F P PHOLE pexo 24 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 2 48 WHOMOLUMO Type Logical Default False Flag specifying weather or not to save the charge density and the orbital values of the highest occupied molecular orbital homo and the lowest un occupied molecular orbital lumo to file or not If WHOMOLUMO TRUE the the charge densities are saved to the files HOMODENS dat and LUMODENS dat and the orbital values are saved to HOMO xsf and LUMO xsf The output format for the HOMODENS dat is the same as for the CHARGEDENS dat file and the output format of the HOMO xsf and LUMO xsf files are that of the zrysden software package Assuming that the zcrysden software package have been installed on your machine you only have to give the following command on the command line in order to plot the HOMO iso surface V 31 SERIALVERSION gt xcrysden xsf HOMO xsf after which you go to the Tools drag down menu of the zcrysden program and select the iso surface alternative Then choose the Data Grid alternative click ok Finally
8. Double Precision Default None Four numbers on the same row specifying the atomic number of an atom in the molecule and its corresponding position in cartesian coordinates The first number equals the atomic number and the three following numbers are the cartesian coordinates x y and z of the atom The entries containing information on the atomic numbers and positions must follow immediately after the NATOMS entry 22 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 2 40 WRITECICOEF Type Logical Default False Flag specifying weather or not to save the expansion coefficients of the Configuration inter action CI wave function to file or not If WRITECICOEF TRUE then the Cl wave function expansion coefficients are saved to the file CIEXPANSIONCOEFF dat 4 2 41 WRITEDENS Type Logical Default False Flag specifying weather or not to save the charge density to file or not If WRITEDENS TRUE the the charge density is saved to the file CHARGEDENS dat The format of the file is the following At each row of the file the rightmost number corresponds to the charge density at a spatial point r x y 2 specified by the three preceding integers on the same row These integers lets call them I J and K correspond to cartesian coordinates by the following mapping LIMITS 1 I 1 x I LIMITS 1 2 MESH 1 1 E LIMITS 2 I 1 y J LIMITS 2 2 ESHO I 7 LIMITS 3 I 1
9. RELALGO eis sa TAD i A ees Mey Ba a oe 19 ETOD sib sist E 19 NS TERS a5 S08 srt so cate oe AS Fe Sts op te hash a OG ra Be eas 19 DRE ha A ee eh Re tear Seth ae ee o Lea 19 NLSPOINTS 4 La ie oe ae ae ee ee Bie eS Bh Se g 20 PORDER 0020 4a Joe ot e ce ee ae a ot ten ly ae ete BA cag 20 MERA iarr Bb Pine eth E TI oS ier eas ea Nd E AN 20 ETEMP Only used for CORRLEVEL URHF LDA PBE B3LYP 20 DISORD 25 40 is sso Se fel se A AA a ae oe Get e 21 DITSSTART ai Sayarat Gee tate es a Set Bak AL Gh a war a Ae 21 N Ce fbn Bb Rea Ee ah ete ie ate ee Patna EY ERD AGS te 21 NATOMS Shia GAL e Maye HG ed deed hd doh Hh Dade ay 4 21 ATOM 3 3 ee y ae a ta ew ee AS epee BALLS 21 WRITECT CORE 2 0 To sete Gece e ae adh Sokal a Ee BR ee 22 WR ITEDENSS yf bl deh a AGT eluate Be A A BA UE de 22 MESA 2m a ot a it idee atte ae ae a SA 1s Gate ash Bie Gol Sn Be BG 22 LIMITES a Se eee E E at Weep aon ae AAA hoa To e 22 HEORBWRITE 5 2 lee ae ae ea A ee A 22 TOSA ik cae A a ashore Cate VA dees Gt dardo Lhe 23 ADRBS gt 2 520 q di Ghent oh ae eens ah IA ao at DAA IDAS 23 TORBNR ir a a de E A eh ek ae E 23 WHOMOLUMO siaii ai aa Ban eye sd we gl oe Ge 24 APPROXBEE satu Save dr de a e Mol Al ae te da wae ht A 24 BETOL 14002 a dd e lo e A A 24 BETA 8 8 aio a o MG at dot dit e di baie 24 NREPEICAS gt 3 2 std e sd de bode a a a 8 25 SAMPLERATE 3 ro a te ee dete a A a dl 25 NPERSIST o sia rre es IN a a ee wo a 25 NRECAL G a S af ns a as
10. of atoms when using some of the BASISFILE files stored in the V 31 BASIS subdirectory So take care If there is a basis not provided with the current UQUANTCHEM distribution you might find it on the basis set cite https bse pnl1 gov bse portal Here you can by means of cut and past create more basis set files This is done by the following procedure 1 Go to https bse pnl gov bse portal 2 On the leftmost scroll menu there are a plethora of basis sets defined Scroll down to the basis set you want to use and mark it 3 On the periodic table in the middle of the page there will now appear orange color ings in the left bottom corners of the atom for which a basis exists 4 Mark the atoms for which you want to use the basis 5 Select the turbomole format and click on the bottom marked Get Basis Set 6 Now there will appear a window containing the basis set information you require Copy this and past it into a file BASNAME raw located in the same directory as the perl script rawtofortranformat pl located in the the sub directory V 31 BASIS 7 replace the third line in the perl script reading header aug pcS 4 with header BASNAME 8 run rawtofortranformat pl and the file BASNAME dat will be created This new file can be used by UQUANTCHEM 4 1 3 The MOLDYNRESTART dat file In this file contains the information needed to continue a Molecular Dynamics Calculation The file contains the followi
11. REDISTRIBUTIONFREQ 0 then the walkers are redistributed only if there is a threat of all walkers being killed on one computational node thread processor or if there is a risk of overpopulation of walkers at one node Observe that in order to enable a restart of a DQMC calculation one has to set REDISTRIBUTIONFREQ gt 0 This will force uquantchem to save all the information needed to continue a DQMC calculation to the file DQMCRESTART dat every REDISTRIBUTIONFREQ th time step If the file DQMCRESTART dat exists uquantchem will continue the DQMC calcu lation from the time step at which the file DOMCRESTART dat was latest updated 4 2 57 TEND Type Double Precision Default 10 0 Specifies the run time of the Diffusion Quantum Monte Carlo calculation aMolecular Dy namics Calculation or a TDDFT calculation 4 2 58 TSTART Type Double Precision Default TIMESTEP Specifies at which time one will start calculating the mean value of the local energy EL i e TEND TIMESTEP Et TEND TSTART 2 Ext 4 22 t TSTART 4 2 59 TIMESTEP Type Double Precision Default 0 0025 Specifies the time step of the Diffusion Quantum Monte Carlo calculation a Molecular Dynamics Calculation or a TDDFT calculation 4 2 60 CUTTOFFFACTOR Type Double Precision Default 1 0 Specifies the cut off for the local energy Ez in a QDMC calculation Is used to discard pathological configurations If the the positions of a walker generation I gener
12. Thijssen Computational Physics Page 330 Cambridge P Puley Chem Phys Lett 73 393 1980 P Puley J Comp Chem 3 556 1982 S E Koonin and D C Meredith Computational Physics Page 213 214 West wide Press 1990 C J Umrigar M P Nightingale K J Runge J Chem Phys 99 2865 1993 S Manten and A L chow J Chem Phys 115 5362 2001 A M N Niklasson C J Tymczak and M Challacombe Phys Rev Lett 100 123004 2008 Phys Rev Lett 97 123001 2006 J A Pople R Krishnan H B Schlegel and J S Binkley Int J Quantum Chemistry Quantum Chemistry Symposium 13 225 241 see eqn 21 p 229 H Bernhard Schlegel Theor Chem Acc 103 294 296 here see eqn 3 7 Anders M N Niklasson and Marc J Cawkwell Phys Rev B 86 174308 2012 Anders M N Niklasson Peter Steneteg Anders Odell Nicolas Bock Matt Challacombe C J Tymczak Erik Holmstrm Guishan Zheng and Valery Weber J Chem Phys 130 214109 2009 17 B Mielich A Savin H Stoll and H Preuss Chem Phys Lett 157 200 1989 18 A D Becke Phys Rev A bf 38 3098 1988 51 52 BIBLIOGRAPHY 19 P J Stephens F J Devlin C F Chabalowski and M J Frisch J Phys Chem bf 98 11623 1994 20 B G Johnsonm P M W Gill and J A Pople J Chem Phys 98 5612 1993
13. as a function of time isodens m This matlab script uses the out put file CHARGEDENS dat to calculate and plot an iso density surface of the charge density isohomolumo m This matlab script uses the out put files LUMODENS dat and HOMODENS dat to calculate and plot an iso density surface corresponding to the LUMO and HOMO orbitals 7 1 5 TESTS This subdirectory contains the subdirectories RUN1 RUN2 RUN7 which in turn contains input and output files of seven different test calculations performed with the openmp version of the uquantchem code A more detailed specification of these test calculations can be found in the README file 7 1 6 README This file contains basic information on how to compile the uquantchem code and information on the different test calculations located in the subdirectory TESTS 7 1 7 manual pdf This manual that you are now reading 7 1 8 SERIALVERSION This subdirectory contains all the fortran source files f90 of the serial version of the uquantchem code together with different Makefiles Here follows a complete list of the dif ferent fortran source files and a short description 38 CHAPTER 7 FOR THE DEVELOPER uquantchem f90 Main program datatypemodule f90 Module in which the different datatypes specific to the uquantchem code have been defined exchcorrmodule f90 Module in which the different exchange correlation functionals have been defined in terms of subroutines and functions
14. eh A ns a io Site A Ge ap ees See 26 REDISTRIBUTIONFREQ Only used in the MPI version 26 TEND oe 60 Gh do hak eet hee Roa as tes a eee te i s Re See Bee 26 TSTART of ce do A cee ea as new ten lay ae edd 4 As cpg ly 26 TIMESTEP r da eb dont Banton al Roads ee MOE he he ee an toh te te hank eS ates hy 26 CUT TOFF FACTOR ma Sos on de and E di 26 CUSPCORR cnc geo aes hat Bee eh econ ay a Sea ML te wae a al le a 27 TOO oh i Stet on ee Be ee ddd eae Sole bab we amp Ue atk Ae 27 4 2 63 4 2 64 4 2 65 4 2 66 4 2 67 4 2 68 4 2 69 4 2 70 4 2 71 4 3 1 4 3 2 4 3 3 4 3 4 4 3 5 4 3 6 4 3 7 4 3 8 4 3 9 4 3 10 4 3 11 4 3 12 4 3 13 4 3 14 4 3 15 4 3 16 4 3 17 CORRALCUSP BJASTROW si u aaa Se ee eee CIASTROW tase ta cit a BS NVME snk bid ae e ese a Ge ee 8 LEXGSP aalala d Bee ce Kh ase ho ENEXCM 2 22 kee da NEEXC narniai Gee ke Aaa ae ee SPINCONSERVE RESTRICT emmm we es 4 3 The output files The CHARGEDENS dat file The RHFEIGENVALUES dat file The UHFEIGENVALUES dat file The ORBITAL dat file The ABSSPECTRUM dat file The TDFTOUT dat file The OCCUPATIONUP dat file The OCCUPATIONDOWN dat file The HOMODENS dat file The LUMODENS dat file The ENERGYDQMC dat file The DQMCRESTART dat file The MOLDYNENERGY dat file The MOLDYNRESTART dat file The MOLDYNMOVIE xsf file The ATOMPOSITIONS dat file The HOMO xsf file and the LUMO xsf file Example Calculations Utility Matlab programs for plotting
15. to TRUE The atomic positions of SAMPLERATE th time step will be used as a movie frame and saved to the MOLDYNMOVIE xsf file To create at gif movie you need to use the program xcrysden in the following way gt gt xcrysden xsf MOLDYNMOVIE xsf 4 2 54 NPERSIST Type Integer Default 50 NPERSIST is the number of generations a random walker in the diffusion quantum Monte Carlo algorithm is permitted to stay in the same position If a walker stays more than NPERSIST generations in one place the acceptance probability is increased by a factor of 1 1 Vs NPERSIST 4 21 Here N is the number of generations time steps a walker has been rejected to move This is to avoid a population catastrophe For more details see Umrigar et al 10 26 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 2 55 NRECALC Type Integer Default Int 1 TIMESTEP Integer number deciding how often the mean total energy estimate Ex should be updated so that the number of walkers is kept close to the initial number of walkers NREPLICAS See the the description of the related parameter BETA for a more detailed description 4 2 56 REDISTRIBUTIONFREQ Only used in the MPI version Type Integer Default 0 The frequency in which the random walkers in a diffusion Monte Carlo calculation are redis tributed evenly over the MPI threads processors For example REDISTRIBUTIONFREQ 3 results in a redistribution of walkers every third time step If
16. w w wt 2T 2T e t EFTELDMAX NEPERIOD 2 2 if NEPERIOD 1 lt t lt NEPERIOD 2 T w Ww 2 e t 0 0 if t gt NEPERIOD 2 W Here is the propagation direction of the Electric field see EDIR and c the speed of light in vacuum If EPROFILE DP the electric field has been assumed to be a Dirac pulse and equal to EFIELDMAX for t 0 Just at the precise moment when the TDFT TDHF time propa gation starts and zero for t gt 0 This modulation is used to calculate absorption spectra since it corresponds to a Dirac perturbation which is a superposition of all frequencies Here w OMEGA 4 2 INPUT PARAMETERS 15 4 2 5 NEPERIOD Type Integer Default 1 Number of periods that the amplitude modulation e t is equal to EFIELDMAX 4 2 6 EFIELDMAX Type Double Default 0 030 a u The strength of the electric dipole field used in a TDFT or a TDHF calculation 4 2 7 EDIR Type Integer Default 1 Defines the propagation direction and polarization of the electric dipole field used in a TDFT THF calculation If EDIR 1 then the field propagates in the x direction and is polarized along the y direction if EDIR 2 then the field propagates in the y direction and is polarized along the z direction and if EDIR 3 then the field propagates in the z direction and is polarized along the x direction The direction of travel is only important in the case when EPROFILE AC Inhomogeneous fiel
17. you fill in your Isovalue usually 0 1 is a good default to start with click the Render isovalue and click on the Submit button in the lower right corner of the pop up menu and you will have created a isosurface of your HOMO orbital 4 2 49 APPROXEE Type Logical Default TRUE if APPROXEE TRUE then elements of the four electron tensor k 1 for which the fol lowing condition is satisfied v i jli kK DP max lt EETOL 4 16 Pmax Maz 21 Pl 21 Prol Pirl Pjr Puls Pol 4 17 will not be calculated and approximately set to zero This is to enable a more effective alter native to calculating the electron repulsion tensor i k l between orbitals that are centered far from each other If APPROXEE FALSE all elements of the tensor will be calculated 4 2 50 EETOL Type Double Precision Default 1 0E 10 The threshold used in the approximation of the four electron tensor j k 1 See above definition of APPROXEE for more details 4 2 51 BETA Type Double Precision Default 1 0 Parameter used for updating the total energy estimate Er used in quantum Monte Carlo calculations Whenever the time step number J satisfies MO D I NRECALC 4 0 the energy estimate is updated according to Er 1 Er 1 1 4 18 BETA a O Tef fNRECALC 2 Wr 4 2 INPUT PARAMETERS 25 here J is the time step number W Number of replicas of the current time step and NRECALC Integer numbe
18. 2 K LIMITS 3 2 1 4 2 42 MESH Type Integer Integer Integer Default 100 100 100 Specifies the mesh grid to be used when saving the charge density the homo density and the lumo density to file For more details on the use of integers in MESH see the section about the WRITEDENS parameter 4 2 43 LIMITS Type Double Precision Double Precision Double Precision Default 5 0 5 0 5 0 Specifies the the part of space in which the charge density the homo density and the lumo density is to be calculated and saved to file The part of space in which these densities are to be calculated is defined by x E LIMITS 1 LIMITS 1 y LIMITS 2 LIMITS 2 z LIMITS 3 LIMITS 3 4 2 44 HFORBWRITE Type Logical Default False If true the Hartree Fock eigen function orbital with index TOSA is calculated on the same mesh as defined by MESH and limits defined by LIMITS The values of the orbital is saved to the file ORBITAL dat 4 2 INPUT PARAMETERS 23 4 2 45 IOSA Type INTEGER Default Ne MOD Ne 2 2 MOD Ne 2 THE HOMO orbital The index of the orbital which corresponding charge density being saved to the file ORBITAL dat if HFORBWRITE TRUE 4 2 46 AORBS Type INTEGER Default 0 If AORBS 0 this corresponds to the index of the orbital which is being saved to the file ARBORB xsf If AORBS gt 0 the spin up orbital is saved and if AORBS lt 0 the spin down orbital is saved
19. Assuming that the zcrysden software package have been installed on your machine you only have to give the following command on the command line in order to plot the orbital iso surface V 31 SERIALVERSION gt xcrysden xsf AORBS xsf For more details on how to obtain the iso surface see the section describing the WHOMOLUMO input flag 4 2 47 IORBNR Type INTEGER ARRAY Default 0 0 Indexes of orbital being ionized excited and treated as a hole and the excited orbital If IORBNR 0 then no ionization excitation will be performed If TORBNR 1 gt 0 then the spin up orbital with index IORBNR 1 will be treated as a hole if TORBNR 1 lt 0 then the spin down orbital with index IORBNR 1 will be treated as a hole If IORBNR 2 gt 0 then the spin up orbital with index IORBNR 2 will be treated as an excitation provided IORBNR 2 gt N 2 if IORBNR 2 lt 0 then the spin down orbital with index IORBNR 2 will be treated as a excitation provided IORBNR 2 gt N 2 Le the orbital with index IORBNR 1 is excited to the orbital IORBNR 2 or treated as a hole if IORBNR 2 0 This feature is only available for CORRLEVEL LDA PBE B3LYP i e DF T type of calculations The hole calculation is performed as follows First a self consistent calculation with all Kohn Sham orbitals present is performed From this calculation the density matrix PYOLE and PEXC corresponding to the hole and excitation are constructed
20. EL second column the estimate of the ground state energy Ep third column and the num ber of random walkers fourth column of the Diffusion Quantum Monte Carlo DQMC calculation are printed Don t forget to check the resulting DQMC energy by running the utility program dqmc_check This program located at V 31 UTILITYPROGRAMS takes the ENERGYDQMC dat file as input and checks the correlation between walker populations separated by m time steps m 5 default and is set by editing the perl script dqmc_check p1 The dqmc_check pl script will also calculate the mean value of the local energies Ez separated by m time steps Remem ber that uquantchem has no offset when sampling the local energy Ez i e m 1 Thus the mean value for the local energy EL given as output directly from uquantchem might very well have been calculated over correlated populations 4 3 12 The DQMCRESTART dat file In this file all the information needed to restart a DQMC calculation is stored If this file is present in the running directory of uquantchem then the DQMC calculation will be continued from the time step where the DQMCRESTART dat file was most recently up dated Observe that the condition REDISTRIBUTIONFREQ gt 0 must be satisfied in order for the DQMCRESTART dat to be created The DQMCRESTART dat file is updated every REDISTRIBUTIONFREQ th time step The possibility to continue a DQMC calculation only exists for the MPI version of the
21. KAPPA Type Double Precision Default depending on DORDER see Table I in 16 Parameter used in the XLBOMD time propagation of the density matrix Only used if XLBOMD is TRUE 4 2 INPUT PARAMETERS 17 4 2 17 ZEROSCF Type Logical Default False If this flag is set to TRUE then when performing a Molecular Dynamics Calculation self consistent calculation to obtain the density matrix is only performed for the initial 10 first time steps This feature is only to be used together with XLBOMD set to TRUE since other wise the density matrix will be constant throughout the MD calculation The combination of setting both the flags ZEROSCF and XLBOMD equals the method of Fast Quantum Mechanical Molecular Dynamics FQMMD as is described by Niklasson et al 15 4 2 18 ZEROSCFTYPE Type Integer Default 1 Only available in the OMP and MPI version In the case of ZEROSCF TRUE this parameter selects which of two possible energy expres sions to use when calculating the total energy and the interatomic forces when performing a molecular dynamics calculation using the XLBOMD scheme with only 1 scf cycle per time step For simplicity we here assume a non spin polarized RHF calculation or a non spin polarized LDA calculation to exemplify the two energy expressions ZEROSCFTYPE 1 E 2Tr hD Tr DG D RHF E 2Tr hD Tr DG D Esc D LDA ZEROSCFTYPE 2 E 2Tr hD Tr 2D P G P RHF E 2Tr hD Tr 2D
22. LDA PBE B3LYP Used for setting the level of electron correlation employed in the calculation URHF Un restricted Hartree Fock RHF Restricted Hartree Fock CISD Configuration Interaction Calculation singles and Doubles MP2 Moller Plesset second order perturbation theory VMC Variational Monte Carlo DQMC Quantum Diffusion Monte Carlo LDA DFT calcula tion using the Local Density Approximation LDA in the spirit of Vosko Wilk and Nusair 1 PBE DFT calculations using the gradient corrected functional of Perdew Burke and Ernzerhof PBE 2 B3LYP DFT 1 17 18 19 calculations using the hybrid functional of B3LYP 4 2 2 ADEF Type Logical Default FALSE If true then all the time independent calculations will be performed with a static efield present In the case of a molecular dynamics calculation depenpending on the param eter EPROFILE the field will have different time evolutions during the molecular dynam ics run see section below describing EPROFILE The option EPROFILE DP does not work 14 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION when ADEF TRUE There is also an extra option for EPROFILE not available when when DOTDFT TRUE i e for TDDFT THF calculations and that is EPROFILE ST which will result in all calculations being perfomed in the presence of a static electric field of strength EFTELDMAX When doing molecular dynamics calculations in the presence of a time dependent ex ternal electri
23. P PF9S75V8 4 3 If PULAY is equal to 4 then the following expression is used to calculate the Pulay contribution to the force 14 Fpulay 2Tr FP VS P 4 4 4 2 INPUT PARAMETERS 19 Here F is the Fockian e and C are the eigenvalues respectively eigenvectors of the matrix equation FC SC S the overlap matrix and P equals the density matrix Puv 5 CriCri 4 5 4 2 25 TOL Type Double Precision Default 1 0E 6 Used to set the convergence criterion for the self consistent field calculations i e the self consistent Hartree Fock calculations are terminated when the difference between the total energy of two consecutive iterations is lt TOL 4 2 26 RELAXN Type Logical Default False Flag specifying weather or not to perform a structure relaxation calculation If RELAXN TRUE uquantchem performs a structure relaxation calculation 4 2 27 RELALGO Type Integer Default 1 Only used if RELAXN TRUE Selects which type of relaxation algorithm to be used when optimizing the atomic positions If RELALGO 1 then the optimization will be done by searching for the atomic positions for which all the forces are zero Here caution must be taken since there is no guarantee that the atomic configuration might end up in a molecular structure corresponding to a local energy maximum Thus even though this is by far the most effective algorithm as compared to RELALGO 2 always keep an eye on the total ene
24. User manual for the Uppsala Quantum Chemistry package UQUANTCHEM Vol by Petros Souvatzis Department of Physics AS aes sa Uppsala University 2012 Contents 1 Introduction 5 2 Compiling the code 6 3 What Can be done with UQUANTCHEM 8 3 1 Hartree Fock Calculations 0 2000020002 ee 8 3 2 Configurational Interaction Calculations o o 8 3 3 Moller Plesset Calculations MP2 o o o 8 3 4 Density Functional Theory Calculations DET 8 3 5 Time Dependent Density Functional Theory Calculations TDDFT 9 3 6 Quantum Montecarlo Calculations o e o 9 3 7 Born Oppenheimer Molecular Dynamics 04 9 4 Setting up a UQANTCHEM calculation 11 401 The put Mes aida a fab SR ee Ge eh ee ee AS 11 4 L Lo Phe INPUTEILLE le estena ri as Sh baa RA 11 4 1 2 The BASISFILE file e e 12 4 1 3 The MOLDYNRESTART dat file 12 4 1 4 The INITVELO dat fle 20 0000 eee 13 4 1 5 Running Uquantchem e e 13 4 2 Inputs parameters os 202 de A a ot BAe Ad 13 4 2 1 CORRGEVEL o osa aot a A ate a eo A 13 ud De GRADER a satiate ay RS facie Berean WO tach oe oc pce teks 13 423 DOTDET culata ni ete Gt Be A ie A a A 14 AVDA CEPROFTILE 44 4 a a bh ee I ke dt 14 4 20 NEPERTOD ote ok duc a a ea SR NRA EE A HS 15 4 2 0 SEF TELDMAK O ausip Gels a Men
25. Where 1 nw E si Aa 4 9 4 2 INPUT PARAMETERS 21 4 2 35 DIISORD Type INTEGER Default 3 MAX 25 Mixing parameter equal to half of the order of the direct inversion of the iterative subspace method DIIS 7 8 If DIISORD 0 DEFAULT the DIIS mixing scheme is not used The number 2DIISORD equals the number of density matrices mixed together according to 2DIISORD 2DIISORD P X aP Y asl 4 10 i l a such that Ae 0 in the mean square sense where 2DIISORD Ae Y aye 4 11 1 and e FPS SPF e ST1MPes 1 4 12 Here F is the Fockian P the density matrix used to calculate the Fockian F S the overlap matrix and S71 the L wdin ortogonalization matrix The index refers to the number of iterations performed in the scf cycle 4 2 36 DIISSTART Type INTEGER Default 50 The DIIS mixing is used if AP lt 1 or until the number of scf iterations gt DIISSTART 4 2 37 Ne Type Integer Default Total Nuclear charge of Molecule Total number of electrons in the molecule 4 2 38 NATOMS Type Integer Default None Specifies the total number of atoms in the molecule After this entry there must follow an equal number of rows specifying the atomic species and there spatial positions given in cartesian coordinates The only entries allowed after the NATOMS entry are the rows specifying the atomic species and positions 4 2 39 ATOM Type Integer Double Precision Double Precision
26. YEVD to diagonalize the CISD hamilto nian matrix by the divide and conquer algorithm gridsetup f90 This subroutine factorizes the number of processors nproc into nprow and npcol that are the sizes of the 2 d processors mesh countee f90 Calculates the diagonal electron electron integrals ij i7 and counts the number of integrals that give a considerable contribution to the exchange and Hartree matrices Kj and Jij Here we use the Schwartz inequality i j k 1 lt 1 3 5 4 5 k 1 k D 2 See Eqn 9 12 25 p 404 in T Helgaker et al s Molecular Electronic Structure Theory together with Eqn 56 and 57 page 86 in Haettig Multiscale Simulation Methods in Molecular Sciences J Grotendorst N Attig S Blugel D Marx Eds Institute for Advanced Simulation Forschungszentrum Julich NIC Series Vol 42 ISBN 978 3 9810843 8 2 pp 77 120 2009 to decide wich integrals give considerable contributions Bibliography O N QOQ 10 11 12 13 14 15 16 S H Vosko L Wilk and M Nusair Can J Phys 58 1200 1980 John P Perdew Kieron Burke Matthias Ernzerhof Phys Rev Lett 77 3865 1996 A Szabo and N S Ostlund Modern Quantum Chemistry Dover Mineola New York 1996 X Li S M Smith A N Markevitch D A Romanov R J Lewis and H B Schlegel Phys Chem Chem Phys 7 233 2005 D B Cook Handbook of Computational Chemistry Dover Mineola New York 2005 J M
27. aries that comes with the ugantchem package together with uquantchem by following these steps 1 V 31 gt cd SERIALVERSION 2 V 31 SERIALVERSION gt cp Makefile gfortran nolapack noblas serial Makefile 3 In the Makefile edit the line LAPACKPATH Users petros UQUANTCHEM Src V 21 lapack 3 4 0 so it reads LAPACKPATH Where uquantchem is located on my machine lapack 3 4 0 4 In the Makefile edit the line BLASPATH Users petros UQUANTCHEM Src V 21 BLAS so it reads BLASPATH Where uquantchem is located on my machine BLAS 5 V 31 SERIALVERSION gt make all And similarly if you want to install the openmp or the MPI version of the code you do just move in to the OPENMPVERSION directory or the MPI_VERSION directory and perform the steps 2 5 In the case of the openmp gfortran version the simplest way to install is to use the pre made make file named Makefile gfortran openmp or if you have access to any of the Swedish super computer clusters Lindgren or Neolith there are also pre made Make files for the MPI version of the code to make life easier Chapter 3 What Can be done with UQUANTCHEM 3 1 Hartree Fock Calculations There are to types of Hartree Fock types of Hartree Fock implemented in UQUANTCHEM The restricted Hartree Fock RHF and unrestricted Hartree Fock URHF The implemen tation is based on expanding the molecular orbitals of the slater determinant in a basis set consisting of contracted gaussian primitiv
28. ates a local energy EL 1 such that Er EL 1 Fal gt CUTTOFFF ACTOR 4 23 R 4 2 INPUT PARAMETERS 27 4 2 61 CUSPCORR Type LOGICAL Default TRUE If CUSPCORR TRUE then the basis functions are corrected so that they have the correct nuclear cusp behavior close to the nuclei This correction is done in the spirit of S Manten et al 11 This correction is only used for quantum Monte Carlo calculations 4 2 62 rc Type Double Precision Default 0 10 If nuclear cusp correction is used the basis functions are corrected at distances lt rc from the nuclei at which they are centered 4 2 63 CORRALCUSP Type LOGICAL Default TRUE If true all basis functions will be cusp corrected If false only basis functions constructed from contractions of more than 1 primitive gaussiam will be cusp corrected 4 2 64 BJASTROW Type Double Precision Default 1 0 Parameter used in the Jastrow factor 7 containing the explicit electron correlation in the trial function Yr DD J 4 24 Where BJASTROW Tij da 0 1 CIASTROW riz 4 25 Tij r rj is the distance between electron i and j D and D are the slater determinants created from spin up respectively spin down orbitals The orbitals are constructed from the URHF self consistent solution Here 6 0 25 if the spin of the electrons and j are identical otherwise if the spins are opposite 0 5 Observe that BJASTROW should be kept e
29. aves the charge density on disk 40 CHAPTER 7 FOR THE DEVELOPER exciteornot f90 Function used to decide which excitations to use when performing a CISD calculation hforbitalsave f90 Subroutine that calculates the values of a Hartree Fock or Kohn sham orbital on a mesh and saves the result to disk binomfac f90 N N calculates the binomial factor M MNM basfunkval f90 Calculates the value of a real basis function 4 at the spatial point r R3 Le y r URHF 90 Subroutine for performing unrestricted Hartree Fock calculations primeeintegral f90 Function that calculates the electron repulsion integrals uv KA over the primitive gaussian functions Qu Ov Ox Pr Le i fe fa Sala jr Y potential f90 Subroutine that calculates the potential matrix V and the nuclear gradients of the poten tial matrix Vr Vi Where Vij is defined by AS wilt wy vr E e Ral P Here Rx are the positions of the nuclei in the molecule and Zz are the respective atomic numbers overlap f90 Subroutine that calculates the overlap matrix S and the nuclear gradients of the overlap matrix Vr Sij Where Sij is defined by Sy route oneint f90 This function calculates the Ix ta term on p 974 in J Comp Chem 11 972 977 1990 which is used to calculate the two electron integrals through Rys quadrature 7 1 FILE MAP 41 normalize f90 This subroutine calculates the normalization constants for al
30. c field the electrons are assumed to follow the field adiabatically Therefore great care should be taken to make sure that the system evolves as closely as possible to the ground state of the system The validity of the adiabatic approximation might be eval uated by calculating the occupation numbers for the hartree fock Kohn Sham orbitals as a function of time by performing a TDDFT THF calculation with the same field that is going to be used in the MD run If the fluctuations of the orbital occupation numbers are not to severe the adiabatic approximation can be used at least with some confidence 4 2 3 DOTDFT Type Logical Default FALSE If true then a time dependent density functional calculation TDDFT is performed if CORRLEVEL LDA PBE B3LYP otherwise if CORRLEVEL URHF then a time dependent Hartree Fock calculation is performed This has only been implemented in the MPI and OPENMP version of the code 4 2 4 EPROFILE Type Character Default HO Defines weather or not the electric field used in a TDDFT TDHF calculation is homogeneous or not and the modulation e t of the field If the EPROFILE HO then the amplitude of the electric field is defined by E t r e t sin wt or if EPROFILE AC then the electric field is defined by E t e t sin wt c wW h r where the modulation e t is defined by wt 2T e t EFIELDMAX if t lt 2T w 27 2T e t EFIELDMAX if lt t lt NEPERIOD 1
31. chniques used in modern quantum chemistry Petros Souvatzis Chapter 2 Compiling the code To compile the code you need to have a fortran compiler installed on your machine together with Lapack If you don t have Lapack installed the latest version of Lapack is provided to gether with the UQUANTCHEM package There are several pre made Make files provided with the UQUANTCHEM package which can be used as templates to create a Make file that corresponds to the specifications of your particular system Here an example of a more or less generic installation procedure will be given ALTERNATIVE 1 Let s assume you have the gfortran compiler available on your system and that you also have lapack and blas preinstalled on your machine Now assume you want to compile the serial version of the the code then follow the following steps 1 V 31 gt cd SERIALVERSION 2 V 31 SERIALVERSION gt cp Makefile gfortran serial Makefile 3 In the Makefile edit the line LAPACKPATH Users petros UQUANTCHEM Src V 21 lapack 3 4 0 so it reads LAPACKPATH Path where I have my lapack lib 4 In the Makefile edit the line BLASPATH Users petros UQUANTCHEM Src V 21 BLAS so it reads BLASPATH Path where I have my blas lib 5 V 31 SERIALVERSION gt make ALTERNATIVE 2 Let s assume you have the gfortran compiler available on your system and that you do not have lapack and blas preinstalled on your machine Then you can compile the lapack and blas libr
32. code 4 3 13 The MOLDYNENERGY dat file The energies and temperature are saved to the file MOLDYNENERGY dat first column time step index second column total energy third column kinetic energy fourth column pl 4 3 THE OUTPUT FILES 31 potential energy fifth column temperature and sixth column number of self consistent cycles performed 4 3 14 The MOLDYNRESTART dat file In this file contains the information needed to continue a Molecular Dynamics Calculation The file contains the following information 1 st line of file contains the time step index integer the following NATOMS lines contain three columns with the atomic positions corre sponding to the time step index of the first line the NATOMS lines following the atomic posi tions contain three columns with the velocities of the atoms corresponding to the time step index of the first line finally the last NATOMS lines contain three colums with the interatomic forces of the atoms corresponding to the time step index of the first line IF THE FILE MOLDYNRESTART dat EXISTS IN THE RUNNING DIRECTORY OF UQUANTCHEM THE CALCULATION WILL BE AUTOMATICALLY CONTINUED TO RESTART A MOLECULAR DYNAMICS CALCULATION FROM SCRATCH BE SURE TO REMOVE THIS FILE FROM THE RUNNING DIRECTORY 4 3 15 The MOLDYNMOVIE xsf file This file contains atomic positions in A sampled from a Molecular Dynamics Calculation and can be used by the xcrysden program to create a gif movie of the mot
33. cular Dynamics Calculation If MOLDYN TRUE a Molecular Dynamics Calculation will be performed 4 2 12 XLBOMD Type Logical Default False Flag specifying weather or not to perform a Molecular Dynamics Calculation using the time reversible propagation algorithm by A M Niklasson 12 for updating the density matrix at each time step If XLBOMD is TRUE a Molecular Dynamics Calculation using the A M Niklasson scheme will be performed 4 2 13 SOFTSTART Type Logical Default False This flag if true specifies the following start up scheme when doing XL BOMD calculations 1 FIRST 50 time steps full scf XL BOMD with high level low order of dissipation DORDER 4 2 Next 50 time steps fast QMMD still with high level low order of dissipation DORDER 4 3 Rest of the calculation is done with XL BOMD or fast QMMD together with the order of dissipation DORDER that has been specified by the user in the INPUTFILE Here depending on the user input XL BOMD or fast QMMD is run 4 2 14 DORDER Type Integer Default 8 Determines the order of the dissipative force term used in the time propagation of the density matrices 16 The order of the dissipative force equals DORDER 1 Is only used if XLBOMD is TRUE 4 2 15 ALPHA Type Double Precision Default depending on DORDER see Table I in in 16 Parameter used in the XLBOMD time propagation of the density matrix Only used if XLBOMD is TRUE 4 2 16
34. ds and the wavelength is comparable to the molecule In the case of EPROFILE DC or EPROFILE HO EDIR only defines the polarization of the Electric field 4 2 8 OMEGA Type Double Default 0 076 a u The frequency of the electric field used in a TDFT or a TDHF calculation if EPROFILE AC or EPROFILE HO The default frequency corresponds to the vacuum wavelength of A 600 nm 4 2 9 NCHEBGAUSS Type Integer Default 100 The number of radial mesh points used in the Chebushev Gauss quadrature used to integrate the exchange correlation energy and exchange correlation potential matrix elements in the radial direction 4 2 10 NLEBEDEV Type Integer Default 3 This integer is used to choose the angular mesh employed by the Lebedev quadrature which is used to integrate the exchange correlation energy and exchange correlation potential ma trix elements in on a spherical surface The integer and its corresponding number of mesh points are the following NLEBEDEV 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Number of mesh points 110 170 194 230 266 302 350 434 590 770 974 1202 1454 1730 2030 2354 2702 3074 3470 3890 4334 4802 5294 5810 Thus the default num ber of angular mesh points is 194 16 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 2 11 MOLDYN Type Logical Default False Flag specifying weather or not to perform a Mole
35. e functions The implementation is of text book style based on the the book of Szabo Ostlund 3 and the book of Cook 5 To calculate the electron electron repulsion integrals ij kl rys quadrature is used 3 2 Configurational Interaction Calculations Configuration interaction calculations is possible to perform with UQUANTCHEM with a basis set constructed by double and single excitations of the original Hartree Fock slater determinant CISD For mor details see Szabo and Ostlund p 231 269 3 3 3 Moller Plesset Calculations MP2 Standard many body perturbation theory calculations up to second order so called Moller Plesset Calculations MP2 are also possible For more details see Szabo and Ostlund p 350 353 3 3 4 Density Functional Theory Calculations DFT Density functional theory calculations are possible to perform with UQUANTCHEM The following functionals are available LDA PBE and B3LYP 8 3 5 TIME DEPENDENT DENSITY FUNCTIONAL THEORY CALCULATIONS TDDFT 9 3 5 Time Dependent Density Functional Theory Calcu lations TDDFT It is also possible to perform time dependent density functional calculations by real time propagation of the density matrix P Here the quantum Liouville equation i RPI 3 1 is solved by the modified midpoint algorithm 4 P t At e OA pa Ape EME 3 2 Here F is the Kohn Sham Fockian Hamiltonian of the system 3 6 Quantum Montecarlo Calculations It is possibl
36. e of the second kind used for the integration along the radial direction 48 CHAPTER 7 FOR THE DEVELOPER sofmu f90 This is the function s u 3 1 f u described by Eqn 3 in A D Becke J Chem Phys 88 2547 1988 Here the recommended third order k 3 is used pvoronoi f90 This function is given by Eqn 13 in A D Becke J Chem Phys 88 2547 1988 getvxc f90 This subroutine calculates the exchange correlation potential matrix elements i VuelJ and the nuclear gradients of these matrix elements Vr i Vrc j using the resolution of the identity into fuzzy polyheadra prescribed by A D Becke J Chem Phys 88 2547 1988 together with Gauss Chebysshev and Lebedev quadrature excdens f90 This function calculates the exchange correlation energy density excdr f90 This function calculates the exchange correlation energy density exc f90 This function calculates the exchange correlation energy from the exchange correlation en ergy density by using the resolution of the identity into fuzzy polyheadra prescribed by A D Becke J Chem Phys 88 2547 1988 together with Gauss Chebysshev and Lebedev quadrature DFT f90 Subroutine performing density functional theory calculations quadcheck f90 Calculates the total number of electrons by integrating the charge density p r by using the resolution of the identity into fuzzy polyheadra prescribed by A D Becke J Chem Phys 88 2547 1988 together w
37. e tensor Vr VYV r The output is stored in the 3x3 matrix nggrad according to the following ordering 2 2 2 nggrad 1 1 arog nggrad 1 2 sR Se nggrad 1 3 apap 2 2 2 nggrad 2 1 o nggrad 2 2 Bn by nggrad 2 3 Bn oy 2 2 2 nggrad 3 1 akaz nggrad 3 2 aR gz nggrad 3 3 5 02 nglaplacebasfunkval f90 Calculates the value of nuclear gradient of the laplacian of a real basis function at the spatial point r R Le The function Vr V VW r The output is stored in the array nglapl according to the following ordering nglapl 1 dV W r dR nglapl 2 dV WU r dR nglapl 3 dV2U r dR getvxcr f90 This subroutine calculate the r dependent matrix elements r V r r and the nu clear gradient of these matrix elements After integration over the entire 3 DIM space with some quadrature these will become the exchange correlation potential matrix elements to be plugged into the Kohn sham hamiltonian and the corresponding force expression for the exchange correlation contribution to the interatomic forces atomicradii f90 This function takes the atomic number as input and returns the empirical radii of Bragg and Slater J C Slater J Chem Phys 41 3199 1964 lebedev f90 This subroutine provides the weights and roots of the Lebedev quadrature used for the in tegration on the surface of a sphere chebgauss f90 This subroutine calculates the weights and roots of the Chebyshev Gauss quadratur
38. e to perform two types of quantum Monte Carlo calculations Variational Monte Carlo VMC which is mainly used to generate good initial random walker configurations for the Diffusion Quantum Monte Carlo DQMC calculations The implementation of DQMC in UQUANTCHEM follows closely the algorithm outlined in the work of Umrigar and Nightingale 10 However in UQUANTCHEM the trial function is constructed with a much simpler Jastrow factor 7 and slater determinants are constructed from cusp cor rected gaussian orbitals The implementation of the cusp correction in UQUANTCHEM follows that described by S Manten and A L chow 11 The explicit form of the trial function used for the importance sampling in the DQMC of UQUANTCHEM is given by Yr DD J 3 3 Where BJASTROW Tij Aa eap 1 CIASTROW riz 3 4 Tij r rj is the distance between electron i and j D and D are the slater determinants created from spin up respectively spin down orbitals The orbitals are constructed from the URHF self consistent solution Here 0 25 if the spin of the electrons and j are identical otherwise if the spins are opposite 6 0 5 The jastrow parameters BJASTROW and CJASTROW are described the section below discussing the input parameters 3 7 Born Oppenheimer Molecular Dynamics Molecular dynamics calculations can be performed within the computational framework provided by uquantchem The nuclei are here propagated with Newton
39. ere the energy is sampled The data DR E DR PE SPOXS is least squares fitted to a polynomial of degree PORDER 1 4 2 32 PORDER Type Integer Default 7 Only used if RELAXN TRUE Here PORDER equals the degree 1 of the polynomial used in the least squares fitting employed in the line search of the conjugate gradient relaxation of the nuclei 4 2 33 MIX Type Double Precision Default 0 0 Mixing parameter used to linearly mix the density matrix of the previous P 1 iteration with the density matrix of the current iteration P in a RHF or a URHF calculation I e P P 1 MIX P MIX 4 6 4 2 34 ETEMP Only used for CORRLEVEL URHF LDA PBE B3LYP Type Double Precision Default 1 0 Only used for CORRLEVEL LDA PBE B3LYP Electronic temperature used to calculate the electronic free energy in the case of a DFT calculation and a URHF calculation If ETEMP gt 0 the density matrix P v will be calculated by occupying the states according to Fermi Dirac statistics i e Ne 1 P y Gah Peo 4 7 i 1 Here 6 1 KBETEMP p is the chemical potential and e are the energy eigenvalues of the eigenvalue equation FC e SC where F is the Kohn Sham fockian hamiltonian Note that if ETEMP gt 0 the parameter PULAY will be automatically set to 2 regardless of the specification of PULAY made in the INPUFILE The entropy S is calculated according to S Kp niln ns 1 nj In 1 n 4 8
40. function f_j 1 m a b on page 237 in the Cook book 5 fac f90 simple factorial function n gto f90 Calculates the value at argument for a spherically symmetric gto basis function consisting of contracted primitive gaussians gtop f90 Calculates the radial derivative at argument r of a spherically symmetric gto basis function consisting of contracted primitive gaussians gtopp f90 Calculates the second radial derivative at argument r of a spherically symmetric gto basis function consisting of contracted primitive gaussians gtoppp f90 Calculates the third radial derivative at argument r of a spherically symmetric gto basis function consisting of contracted primitive gaussians eeints f90 Subroutine that calculates the electron electron integrals ij kl by Rys quadrature diaghHF 90 Subroutine that calculates the eigenvalues e and eigenvectors C to the eigenvalue prob lem FC eSC Here F corresponds to the Fockian matrix and S the overlap matrix diagh f90 Subroutine that calculates the eigenvalues e and eigenvectors C to the eigenvalue prob lem HC eC Here H typically corresponds to the Hamiltonian matrix dfac f90 Function that calculates the double factorial n checkinputfile f90 This subroutine checks wheater or not the file INPUTFILE exists and if so how many lines NLINES it contains before the entry of the number of atoms NATOMS chargedenssave f90 This subroutine calculates and s
41. he overlap matrix when all the basis functions have been shifted with the vector r gt R r ie r gt r R relaxf f90 This subroutine searches for the atomic positions at which all nuclear forces are lt FTOL mulliken f90 Subroutine that calculates the Mulliken charges of the different atoms and prints the result in the stout 7 1 9 OPENMPVERSION This subdirectory contains all the fortran source files f90 of the opemp version of the uquantchem code together with different Makefiles The source files within this directory have the same names and functionality as the f90 files located in the SERIALVERSION directory For a complete listing of the fortran source files contained in the OPENMPVERSION directory se the preceding subsection 7 1 10 MPI_VERSION This subdirectory contains all the fortran source files f90 of the mpi version of the uquantchem code together with different Makefiles The source files within this directory have the same names and functionality as the f90 files located in the SERIALVERSION directory For a listing of the fortran source files contained in the MPI_VERSION directory se the above section describing the SERIALVERSION However there are a couple of source files 50 CHAPTER 7 FOR THE DEVELOPER in the MPI_VERSION directory that are unique to the mpi version of the code Below follow a complete listing of these files diagscalapack f90 Subroutine that uses the ScaLapack routine PDS
42. ion of the molecule To do this just gt gt xcrysden xsf MOLDYNMOVIE xsf then once xcrysden starts the rest is pretty self explanatory The resulting movie is most easily viewed by any web browser 4 3 16 The ATOMPOSITIONS dat file This file contains the latest update of the atomic positions during and after a structural relaxation calculation 4 3 17 The HOMO xsf file and the LUMO xsf file Contain the HOMO and LUMO orbital values printed on a 3D mesh and can be used by the xcrysden program to plot iso surfaces of the respective orbitals C hapter 5 Example Calculations In the sub directories V 31 EXAMPLE_INPUT_OUTPUT one can find the results of different uquantchem calculations RU ing RU N1 contains the output and input of a URHF calculation of a water molecule us a 6 31G basis set N2 contains the output and input of a MP2 calculation of a He atom using a cc pVQZ basis set RU N3 contains the output and input of a CISD calculation of a H molecule atom using a cc pVDZ basis set RU a6 RU N4 contains the output and input of a VMC calculation of a Hz molecule atom using 31G basis set N5 contains the output and input of a DQMC calculation of a H molecule atom using a 6 31G basis set RU N6 contains the output and input of a DQMC calculation of a He atom using a cc pVTZ basis set RU N7 contains the output and input of a RHF charge density calculatio
43. ith Gauss Chebysshev and Lebedev quadrature This is used to check the quality of the quadrature mesh TRACE f90 Calculates the trace of a square matrix hessianbasfunk f90 Calculates the hessian of a real basis function Y at the spatial point r R3 Le the matrix _v hij z Ox 01 hessianrho f90 This subroutine calculates the hessian matrix of the charge density p r i e 2 hi j k AORT where R k i th coordinate of the nucleus positioned at R k and r the j th coordinate of the spatial electron coordinate r 7 1 FILE MAP 49 dftforcedens f90 This subroutine calculates the exchange correlation force density at the point r R3 in space getxcforce f90 This subroutine calculates the exchange correlation contribution to the interatomic forces by integration of the exchange correlation force density This is done by using the resolution of the identity into fuzzy polyheadra prescribed by A D Becke J Chem Phys 88 2547 1988 together with Gauss Chebysshev and Lebedev quadrature pnonorm 90 This function is given by Eqn 13 in A D Becke J Chem Phys 88 2547 1988 tmu 90 This function is the function t j described in appendix B Eqn B9 In J Chem Phys 98 5612 1993 gradpvoronoi f90 This subroutine calculates all the nuclear gradients of the Becke weight function centered at atom I See Appendix B in J Chem Phys 98 5612 1993 gradoverlap f90 Here we calculate the gradient of t
44. l of the basis functions ROOTWEIGHT5 f90 Subroutine that calculates the roots and weights of 5 th order Rys polynomials The cal culation is done by polynomial fits of pree calculated roots and weights The fits are taken from the GAMES package Same as in Pyquante a very fast process leastsq f90 Performs a least square fit of the function f r Ae C to the N data points r i woa i 1 y i GTO r i 1 2 N makedens f90 Subroutine that calculates the density matrix Pav from the eigen vectors of the Fockian Khon Sham Hamiltonian C i e Noce Puy y CuiCl i l kinetic f90 Subroutine that calculates the Kinetic energy matrix Tij and the nuclear gradients of the kinetic energy matrix VrTij Where Si is defined by Ty 5 Cro V r ijkl 90 Function that contacts the 4 component index i j k l used to label the different electron electron repulsion integrals ij kl into one single index homolumosave f90 Subroutine that calculates the densities and orbital values of the HOMO and LUMO orbitals on a mesh and writes the result to the files HOMODENS dat LUMODENS dat HOMO xsf and LUMO xsf gettotalbasis f90 If the logical input variable COUNTER TRUE this subroutine calculates the size of the total basis set from the local basis sets centered at the different atoms If COUNTER FALSE this subroutine returns the total basis set rysquad f This subroutine calculates the roots and
45. lates the potential energy integral given on the bottom of p 244 in the Cook Book 5 primoverlap f90 This function calculates the overlap integral S S 5 5 as given by the expressions on the pages 234 238 in the Cook book 5 primkinetic f90 This function returns the integral GTO V GTO2 as given by the cook Book on pages 234 239 5 Here GTO are primitive gaussian functions MP2 f90 Subroutine that performs second order Mgller Plesset calculations gradprimoverlap f90 This subroutine calculates the nuclear gradients of the overlap integral S Sy 5 40 as given by the expressions on the pages 234 238 in the Cook book 5 46 CHAPTER 7 FOR THE DEVELOPER gradprimkinetic f90 This subroutine calculates the nuclear gradients of the integral GTO1 V GTO2 as given by the cook Book on pages 234 239 5 Here GTO are primitive gaussian functions gradprimpotential f90 This subroutine calculates the nuclear gradients of the potential energy integral given on the bottom of p 244 in the Cook Book 5 gradprimeeintegral f90 This subroutine calculates the nuclear gradients of the two electron integral between primi tive gaussians forces f90 This subroutine calculates the interatomic forces on all atoms of the molecule dAfunc f90 This function calculates the derivative of the function Aj 11 l2 Az Br Ca y defined at the top of page 245 in the Cook Book 5 with respect to Cy relax f90 This subrouti
46. lations if SPINCONSERVE FALSE then spin flips will be allowed and the spin of the system is not conserved when creating the CISD basis set or calculating the MP2 correctons 4 2 71 RESTRICT Type LOGICAL Default FALSE If true then the CISD and MP2 calculations will be based un slater determinants created from a restricted Hartree Fock RHF calculation 4 3 The output files 4 3 1 The CHARGEDENS dat file In this file the charge density calculated with Hartree Fock or Diffusion Quantum Monte Carlo is saved The file consists of four columns The first three columns counting from left to right contain integer numbers corresponding to the coordinate mesh indexes I J K described together with the input parameters WRITEDENS and MESH The fourth column contains the calculated values of the charge density p x I y J z K plz y x 4 3 2 The RHFEIGENVALUES dat file In this file the energy eigenvalues of the restricted Hartree Fock calculation RHF are stored First column eigenvalue index and second column eigenvalues of the spin down orbitals 4 3 THE OUTPUT FILES 29 4 3 3 The UHFEIGENVALUES dat file In this file the energy eigenvalues of the unrestricted Hartree Fock calculation URHF are stored First column eigenvalue index second column eigenvalue of spin up orbitals and third column eigenvalues of the spin down orbitals 4 3 4 The ORBITAL dat file In this file the Hartree Fock eigen function Wrosa ca
47. lculated on the mesh specified by the input parameters WRITEDENS and MESH is saved The file consists of Five columns The first three columns counting from left to right contain integer numbers corresponding to the coordinate mesh indexes I J K described together with the input parameters WRITEDENS and MESH The fourth column contains the calculated values of the Hartree Fock eigen func tion rosa x L J 2 K Vrosar 2 y 2 and the fifth column contains the calculated values of the Hartree Fock eigen function Viosa 1 y J 2 K Yiosa y 4 3 5 The ABSSPECTRUM dat file Contains the absorption spectrum i e the time Fourier transform of the dipole moment Hp t gt 5 Ratom P Later Tr P t dp 4 26 atom Here p is the polarization direction of the electric field P the density matrix and dp the dipole tensor in the polarization direction 4 3 6 The TDFTOUT dat file Contains information from the TDDFT TDHF calculation First column time step index second column time third column dipole moment fourth column expectation value of the effective hamiltonian Fockian fifth column total electron charge sixth column electric field 4 3 7 The OCCUPATIONUP dat file First column contains the time and the second column contains the orbital occupation numbers n t of the spin up orbitals ni t CHO P 901 0 4 27 projected out from the density matrix P t of a TDDFT TDHF calc
48. lectrons N number of electrons 7 1 FILE MAP 45 laplacetrialfnk f90 This function calculates the value of laplacian of the trial function used in the DQMC importance sampling at a specific point in the 3Ne dimensional coordinate space of the electrons N number of electrons EL f90 This function calculates the Local energy of the trial function at a specific point in the 3Ne dimensional coordinate space of the electrons N number of electrons Used in the DQMC importance sampling dqmc 90 This routine performs the diffusion quantum Monte Carlo or the Variational Quantum Monte Carlo calculations This is the routine that is at the top of the hierarchy program tree i e the control routine of all the quantum Monte Carlo calculations preparedifusion f90 Subroutine that performs a series of calculations in order to enable a diffusion step in the DQMC calculation For details see J Chem Phys 99 2865 2890 1993 and the com ments embedded in the source of this routine ROOTWEIGHTMAX3 f90 Subroutine that calculates the roots and weights of Rys polynomials of order lt 3 The cal culation is done by polynomial fits of pree calculated roots and weights The fits are taken from the GAMES package Same as in Pyquante a very fast process readbasis f90 This subroutine reads in the expansion coefficients and exponents of the gaussian basis set stored in the file BASISFILE primpotential f90 This function calcu
49. n for a H20 molecule using a 6 31G basis set RU set RU N8 contains the output and input of a RHF of a Cr atom using a 6 31 basis Here the one of the d orbitals is plotted in the z 0 plane N9 contains the output and input of a DQMC calculation of a He atom using a cc pVTZ basis set Here the calculation is done with a slightly smaller time step than in RUN6 Also the charge density has been accumulated from the walker configurations at the different time steps 32 33 RUN10 contains the output and input of a VMC calculation of a H20 using a 6 31G basis set Also the charge density has been calculated from the generated Metropolis con figurations RUN11 20 I have forgotten the exact type of calculations performed but I think most of them are DQMC or VMC calculations RUN21 contains the output and input of a structure relaxation calculation for CH using URHF and a 3 21G basis set RUN22 26 contains the output and input of a structure relaxation calculation for H20 using URHF and a 6 31G basis set RUN27 contains the output and input of a structure relaxation calculation for H us ing URHF and a cc pVDZ basis set RUN31 contains the output and input of a structure relaxation calculation for H20 using the PBE functional and a cc pVTZ basis set RUN31 contains the output and input of a structure relaxation calculation for H20 using the B3LYP functional and a STO 3G basis set
50. ne calculates the atomic positions of the molecule that corresponds to the min imal energy linesearchmin f90 This subroutine fits a number of NPOINTS energies and moves to 4 th order polynomial in the interval 0 DR and calculates the position of the minimum of this polynomial in the interval 0 DR This routine is used by relax f90 in order to find the move involved a steepest decent or a conjugate gradient calculation massa f90 This function just give the atomic mass as output given the atomic number as input The atomic masses have been taken from Aschcroft and Mermins Solid State Physics moleculardynamics f90 Subroutine that performs molecular dynamics calculations moleculardynamicssoft f90 Subroutine that performs molecular dynamics calculations with a so called soft start See the description of the input parameter SOFTSTART invert f90 Subroutine that inverts a N x N real matrix vxc f90 This subroutine calculates the exchange correlation potential 7 1 FILE MAP 47 vxcalt f90 Subroutine not used for the moment dvxc f90 Subroutine not used for the moment ngbasfunkval f90 Calculates the value of the gradient of a real basis function at the spatial point r R with respect to the nuclear coordinates at which the basis function is centered nggradbasfunkval Calculates the value of nuclear gradient of the gradient of a real basis function Y at the spatial point r x y z Le the elements of th
51. ng information 1 st line of file contains the time step index 4 2 INPUT PARAMETERS 13 integer the following NATOMS lines contain three columns with the atomic positions corre sponding to the time step index of the first line the NATOMS lines following the atomic posi tions contain three columns with the velocities of the atoms corresponding to the time step index of the first line finally the last NATOMS lines contain three colums with the interatomic forces of the atoms corresponding to the time step index of the first line IF THE FILE MOLDYNRESTART dat EXISTS IN THE RUNNING DIRECTORY OF UQUANTCHEM THE CALCULATION WILL BE AUTOMATICALLY CONTINUED TO RESTART A MOLECULAR DYNAMICS CALCULATION FROM SCRATCH BE SURE TO REMOVE THIS FILE FROM THE RUNNING DIRECTORY 4 1 4 The INITVELO dat file This file if provided is used in a Molecular Dynamics Calculation to specify the initial velocities of the atoms The file should contain three columns and NATOMS rows of real numbers specifying the initial velocities of the atoms in au This file is not read if the file MOLDYNRESTART dat is present in the running directory 4 1 5 Running Uquantchem To run for instance the serial version of UQUANTCHEM just simply run the command RUN_DIR gt uquantchem s in the same directory where the INPUTFILE and BASISFILE are located 4 2 Input parameters 4 2 1 CORRLEVEL Type Character Default None CORRLEVEL URHF RHF CISD MP2 VMC DQMC
52. qual to one since this help avoid infinities in the local energy Ez when r rj Thus the only variational parameter that can be used in variational Monte Carlo Calculations CORRLEVEL VMC is CJASTROW 4 2 65 CJASTROW Type Double Precision Default 0 5 See above description of BJASTROW 4 2 66 NVMC Type INTEGER Default 1000000 Number of Metropolis configurations used in the variational Monte Carlo calculation Note that there are SAMPLERATE number of metropolis moves in between every Metropolis config urations used Thus there are a total of NVMC SAMPLERATE Metropolis moves performed 28 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 2 67 LEXCSP Type LOGICAL Default FALSE If true the slater determinants used in the configuration interaction calculation CORRLEVEL CISD will be limited to the slater determinants created by exchanging the NEEXC highest occupied Hartree Fock orbitals with unoccupied Hartree Fock orbitals that have energy eigen values lt ENEXCM 4 2 68 ENEXCM Type Double Precision Default None See description of LEXCSP 4 2 69 NEEXC Type INTEGER Default Ne if LEXCSP FALSE See description of LEXCSP 4 2 70 SPINCONSERVE Type LOGICAL Default TRUE If SPINCONSERVE TRUE then the total spin of the system is conserved for all excitations used to create the basis set of slater determinants used in the configuration interaction calculations CISD or MP2 calcu
53. r set by the user Default Int 1 TIMESTEP which defines the frequency in which the energy update instead is done by following prescription NREPLICAS Erl By 4 19 RU Er r 1 Tep P NRECALC I Wr e i e whenever MOD I NRECALC 0 Here tae Na 1 Ex n Eslj 2 L R Teff TIMESTEP 4 20 Ex n gt 2 j z Teff W 4 20 Here N 1 is the number of walkers accepted to move See for instance Umrigar et al 10 4 2 52 NREPLICAS Type Integer Default 2000 Total number of initial random walkers replicas used in the Diffusion Monte Carlo calcula tion 4 2 53 SAMPLERATE Type Integer Default 10 First use of this parameter is the sample rate employed in the variational Monte Carlo cal culation used to calculate the initial distribution of random walkers Since the Metropolis algorithm is used for the generation of this initial distribution only random walkers sep arated by a number of SAMPLERATE Metropolis moves are used for the initial distribution This is to avoid that the walkers are correlated 9 Second use of this parameter is to specify how often uquantchem is to save the file MOLDYNRESTART dat This file contains the information necessary for a continuation of a Molecular Dynamics Calculation The file will be saved every SAMPLERATE th time step Third use of this parameter is to specify the sampling rate of the atomic positions used to create the file MOLDYNMOVIE xsf if the flag MOVIE is set
54. re is no native BLAS on the machine at which uquantchem is to be installed 7 1 3 lapack 3 4 0 This directory contains the linear algebra package LAPACK attached to this release of uquantchem to enable the compilation of the code in the event that there is no native LAPACK on the machine at which uquantchem is to be installed 7 1 4 UTILITYPROGRAMS This directory contains a number of utility scripts written in perl or to be used with matlab Here follows a list of the included programs and a short description of how the are utilized velautocorr pl This script takes the uquantchem movie file MOLDYNMOVIE xsf and calculates the velocity auto correlation function The result is saved in the file VELAUTO dat and performs the tem poral fouriertransform of the very same function and saves it to the files SPECTRUM_THz dat 36 7 1 FILE MAP 37 and SPECTRUM_cm dat velautocorr2 pl A slightly modified version of velautocorr pl pdbtoau pl For converting the atomic coordinates in pdb files to a coordinate file suitable for the uquantchem code dqmc_check pl Calculates the mean local energy Ez by using every the local energy at every m th inte ger specified by user inside of script time step from the out put file of a dqme calculation ENERGYDQMC dat It also calculates the correlation function distance pl This script takes the uquantchem movie file MOLDYNMOVIE xsf and calculates the inter atomic distance between atom A and B
55. recision Default 300 The temperature used to set the initial velocities in a Molecular Dynamics Calculation However if the file INITVELO dat is present in the run directory of the calculation the initial velocities will be set to those velocities specified in the file INITVELO dat and the value TEMPERATURE will be ignored 4 2 23 CFORCE Type Logical Default False Flag specifying weather or not to calculate the interatomic forces If CFORCE TRUE the interatomic forces will be calculated For the interested user the exchange correlation contribution of the force has been calculated in a similar fashion as described by Johnson et al 20 4 2 24 PULAY Type Integer Default 4 By setting this parameter the user can select between 4 different ways to calculate the Pulay contribution to the interatomic forces This feature is mainly present to enable tests of the numerical stability of the time reversible XLBOMD algorithm All four different approaches are equivalent and only differ due to numerical noise If PULAY is equal to 1 then the following expression is used to calculate the Pulay contribution to the force 13 Frulay 25 EC VS C 4 1 If PULAY is equal to 2 then the following expression is used to calculate the Pulay contribution to the force Fpulay 2Tr S7 FPVS 4 2 If PULAY is equal to 3 then the following expression is used to calculate the Pulay contribution to the force 15 Fpuiay Tr S7 F
56. rgy change during the relaxation run If RELALGO 2 then the optimization will be done by searching for a minimum of the total energy with respect to the atomic positions 4 2 28 FTOL Type Double Precision Default 1 0E 4 Used for structure relaxation calculations RELAXN TRUE The relaxation of the struc ture stops when the mean force is lt FTOL or if the number of relaxation moves gt NSTEPS Be careful with using FTOL lt 1 0E 5 since it might require a huge number of iterations before the structure is relaxed within the prescribed force tolerance 4 2 29 NSTEPS Type Integer Default 500 Used for structure relaxation calculations RELAXN TRUE NSTEPS is the maximum number of relaxation moves permitted 4 2 30 DR Type Double Precision Default 0 5 Only used if RELAXN TRUE If RELALGO 2 the parameter DR defines the interval 0 DR 20 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION along the gradient which the energy is minimized on Upon this interval the energy is cal culated at NLSPOINTS number of points and fitted least squares to a polynomial of degree PORDER 1 Only used for structure relaxation calculations RELAXN TRUE If RELALGO 1 pa rameter DR is the maximum distance in au that the atoms are allowed to move in one Newton Raphson cycle 4 2 31 NLSPOINTS Type Integer Default 10 Only used if RELAXN TRUE Here NLSPOINTS number of points in the interval 0 DR wh
57. s set to be used by UQUANTCHEM More specifically the BASISFILE file contains informa tion about the orbital quantum numbers of the basis functions the gaussian exponents and the contraction coefficients 4 1 1 The INPUTFILE file In the the sub directory V 31 EXAMPLE_INPUT_OUTPUT several example input files can be found Here we only give an example of a INPUT file specifying a URHF calculation of a water molecule CORRLEVEL URHF TOL 1 0E 8 Ne 10 NATOMS 3 ATOM 1 0 453548746355979 1 751220869758844 0 0000000 ATOM 8 0 000000000000000 0 000000000000000 0 0000000 ATOM 1 1 809000000000000 0 000000000000000 0 0000000 11 12 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 1 2 The BASISFILE file In the sub directory V 31 BASIS several basis files can be found To use a specific type of gaussian basis set just copy the file containing the basis you want to use in your calculation from the V 13 BASIS directory to the BASISFILE file For example if you want to use the 6 31G basis set just type cp V 31 BASIS 6 31GST ST dat BASISFILE A word of caution should be said about the BASISFILE file IN ORDER FOR THE CORRECT BASIS TO BE USED FOR A MOLECULE CONSISTING OF ATOMS WITH ATOMIC NUMBERS 11 lt S Z2 lt lt Zn ALL ATOMS WITH ATOMIC NUMBERS UP TO Zn MUST BE SPECIFIED IN THE BASISFILE file FOR THE UQUANTCHEM PROGRAM TO WORK CORRECTLY Unfortunately the above specification might not always be fulfilled for some combinations
58. s classical equations of motion in the context of the Born Oppenheimer approximation The inter atomic forces are calculated analytically from the gradients of the Hartree Fock energy RHF or URHF with respect to the nuclear positions The extended Lagrangian Molecular Dynamics formalism XL BOMD 12 15 16 has been implemented providing a total energy almost completely 10 CHAPTER 3 WHAT CAN BE DONE WITH UQUANTCHEM free of drift Furthermore the Fast First Principles Molecular Dynamics formalism FFP MD has also been implemented providing a very stable time propagation without employing self consistency Chapter 4 Setting up a UQANTCHEM calculation 4 1 The input files There are only two input files that one needs to provide the UQUANTCHEM code the INPUTFILE file contains the required information about which atoms are present in the molecule there positions and the level of approximation used to deal with the electron cor relation On top of this basic information the user can provide more detailed information about convergence criteria and parameters that deals with other levels of approximation or details about the calculation These non basic parameters have been given more or less rea sonable default values so that to enable un experienced users to get up in the air quickly without being weighted down with to many technical details The second input file needed is the BASISFILE file containing information about the basi
59. termiant with respect to the electron coordinate I from a N Total Number of electrons b BAS Hartree Fock basis c C Fockian eigenvectors d r positions the electrons 3N long array e UP TRUE if I spin upp orbital else UP FALSE NS Slaterdeterminant dimension laplaceslaterdet f90 This fuction calculates the laplacian of a slater determiant with respect to the electron coordinate I from a N Total Number of electrons b BAS Hartree Fock basis c C Fockian eigenvectors d r positions the electrons 3N long array e NS Slaterdeterminant dimension UP TRUE if I spin upp orbital else UP FALSE jastrowup f90 This function calculates the Jastrow factor contribution from the spin up electrons N total number of electrons r coordinates of all electrons r is a 3 N long array b c jastrow parameters 44 CHAPTER 7 FOR THE DEVELOPER jastrowdown f90 This function calculates the Jastrow factor contribution from the spin down electrons N total number of electrons r coordinates of all electrons r is a 3 N long array b c jastrow parameters jastrowud f90 This function calculates the Jastrow factor contribution from the cross terms between spin up and spin down electrons N total number of electrons r coordinates of all electrons r is a 3 N long array b c jastrow parameters jastrow f90 This function calculates the Total Jastrow factor N
60. total number of electrons r coordinates of all electrons r is a 3 N long array b c jastrow parameters DIIS f90 This is the direct inversion in the iterative subspace method DIIS by P Puley in CHEM Phys Lett 73 393 1984 The routine is used to estimate the self consistent density matrix in the region where the density dependence of the energy can be approximated well up to second order Also see T Halgaker P Jorgensen and J Olsen Molecular Electronic Structure Theory p 460 463 findclosestatom f90 Subroutine that finds the atomic nucleus located closest to a particular electron The mo tivation for the use of this routine can be found in J Chem Phys 99 2865 2890 1993 gradjastrow f90 This subroutine calculates the gradient of the Jastrow factor with respect to the electron coordinate I N total number of electrons r coordinates of all electrons r is a 3 N long array b c jastrow parameters laplacejastrow f90 This function calculates the laplacian of the Jastrow factor with respect to the electron coordinate I N total number of electrons r coordinates of all electrons r is a Nx3 array b c jastrow parameters guideforce f90 This subroutine calculates the guiding force used in the DQMC importance sampling trialfnk f90 This function calculates the value of the trial function used in the DQMC importance sampling at a specific point in the 3Ne dimensional coordinate space of the e
61. ulation 4 3 8 The OCCUPATIONDOWN dat file First column contains the time and the second column contains the orbital occupation numbers n t of the spin down orbitals m t C7 0 PC 0 4 28 projected out from the density matrix P t of a TDDFT TDHF calculation 30 CHAPTER 4 SETTING UP A UQANTCHEM CALCULATION 4 3 9 The HOMODENS dat file In this file the charge density corresponding to the highest occupied molecular orbital HOMO calculated with Hartree Fock is saved The file consists of four columns The first three columns counting from left to right contain integer numbers corresponding to the coordinate mesh indexes I J K described together with the input parameters WRITEDENS and MESH The fourth column contains the calculated values of the charge density puomo t 1 y J 2 K poomo z y 2 4 3 10 The LUMODENS dat file In this file the charge density corresponding to the lowest unoccupied molecular orbital LUMO calculated with Hartree Fock is saved The file consists of four columns The first three columns counting from left to right contain integer numbers corresponding to the coordinate mesh indexes I J K described together with the input parameters WRITEDENS and MESH The fourth column contains the calculated values of the charge density prumo x 1 y J 2 K prumo z Y 2 4 3 11 The ENERGYDQMC dat file In this file the time step index first column the mean value of the local energy
62. weights of a Rys polynomial of arbitrary order 42 CHAPTER 7 FOR THE DEVELOPER rho f90 Function that calculates the charge density p from the basis functions and the density matrix P at one point in space readin f90 Subroutine that reads the user specified data stored in the file INPUTFILE ROOTWEIGHT4 90 Subroutine that calculates the roots and weights of 4 th order Rys polynomials The cal culation is done by polynomial fits of pree calculated roots and weights The fits are taken from the GAMES package Same as in Pyquante a very fast process interpol f90 This routine calculates the interpolating cubic polynomial linking together the second deriva tive of an sto type basis function sto r Ad e gt at the point r re and a the second derivative of a gto type basis function gto r at r re r so that both the second and the third derivative of the polynomial matches that sto at r r and the gto at r re r This is done according to J Chem Phys 115 5362 2001 gradbasfunkval f90 Subroutine that calculates the gradient of a real basis function at the spatial point r R Le the gradient V r laplacebasfunkval f90 Function that calculates the value of the laplacian of a real basis function at the spatial point r R Le the laplacian V7 r hforbitalval f90 Function that calculates the value of a Hartree Fock orbital or a Kohn Sham orbital at the spatial point r

User manual for the Uppsala Quantum Chemistry

Contents

Download Pdf Manuals

Related Search

Related Contents