⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 mbptr12.h

📁 大型并行量子化学软件;支持密度泛函(DFT)。可以进行各种量子化学计算。支持CHARMM并行计算。非常具有应用价值。
💻 H
字号:
//// mbptr12.h//// Copyright (C) 2001 Edward Valeev//// Author: Edward Valeev <edward.valeev@chemistry.gatech.edu>// Maintainer: EV//// This file is part of the SC Toolkit.//// The SC Toolkit is free software; you can redistribute it and/or modify// it under the terms of the GNU Library General Public License as published by// the Free Software Foundation; either version 2, or (at your option)// any later version.//// The SC Toolkit is distributed in the hope that it will be useful,// but WITHOUT ANY WARRANTY; without even the implied warranty of// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the// GNU Library General Public License for more details.//// You should have received a copy of the GNU Library General Public License// along with the SC Toolkit; see the file COPYING.LIB.  If not, write to// the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.//// The U.S. Government is granted a limited license as per AL 91-7.//#ifndef _chemistry_qc_mbptr12_mbptr12_h#define _chemistry_qc_mbptr12_mbptr12_h#ifdef __GNUC__#pragma interface#endif#include <util/misc/compute.h>#include <util/group/memory.h>#include <util/group/message.h>#include <util/group/thread.h>#include <chemistry/qc/basis/obint.h>#include <chemistry/qc/basis/tbint.h>#include <chemistry/qc/scf/scf.h>#include <chemistry/qc/mbpt/mbpt.h>#include <chemistry/qc/mbptr12/linearr12.h>#include <chemistry/qc/mbptr12/vxb_eval.h>#include <chemistry/qc/mbptr12/vxb_eval_info.h>#include <chemistry/qc/mbptr12/mp2r12_energy.h>namespace sc {// //////////////////////////////////////////////////////////////////////////class R12IntEval;class R12IntEvalInfo;class MP2R12Energy;  /** The MBPT2_R12 class implements several linear R12 second-order perturbation theorymethods. */class MBPT2_R12: public MBPT2 {    Ref<R12IntEval> r12eval_;           // the R12 intermediates evaluator    /** These are MP2-R12 energy objects for each MP2-R12 method, since several different energies        can be evaluated with the same set of intermediates */    Ref<MP2R12Energy> r12a_energy_;    Ref<MP2R12Energy> r12ap_energy_;    Ref<MP2R12Energy> r12b_energy_;    Ref<GaussianBasisSet> aux_basis_;    Ref<SCVector> epair_0_, epair_1_;   // Singlet/triplet pair energies if spin-adapted                                        // Alpha-beta/alpha-alpha pair energies if spin-orbital#define ref_to_mp2r12_acc_ 100.0    double mp2_corr_energy_;    double r12_corr_energy_;    LinearR12::StandardApproximation stdapprox_;    R12IntEvalInfo::StoreMethod r12ints_method_;    char* r12ints_file_;    bool spinadapted_;    void init_variables_();    // This checks if the integral factory is suitable for R12 calculations    void check_integral_factory_();    /* calculate the MP2-R12 energy in std approximations A and A' */    void compute_energy_a_();  protected:    // implement the Compute::compute() function,    // overrides MBPT2::compute()    void compute();  public:    MBPT2_R12(StateIn&);    /** The KeyVal constructor.        <dl>        <dt><tt>stdapprox</tt><dd> This gives a string that must take on one        of the values below.  The default is A.        <dl>          <dt><tt>A</tt><dd> Use second order M\o{}ller-Plesset perturbation theory	  with linear R12 terms in standard approximation A (MP2-R12/A).          Only energies can be computed with the MP2-R12/A method.          <dt><tt>A'</tt><dd> Use second order M\o{}ller-Plesset perturbation theory	  with linear R12 terms in standard approximation A' (MP2-R12/A').          This will cause MP2-R12/A energies to be computed also.          Only energies can be computed with the MP2-R12/A' method.          <dt><tt>B</tt><dd> Use second order M\o{}ller-Plesset perturbation theory	  with linear R12 terms in standard approximation B. 	  This method is not implemented yet.        </dl>	<dt><tt>spinadapted</tt><dd> This specifies whether to compute spin-adapted	or spin-orbital pair energies. Default is to compute spin-adapted energies.	<dt><tt>aux_basis</tt><dd> This specifies the auxiliary basis to be used for the resolution	of the identity. Default is to use the same basis as for the orbital expansion.	<dt><tt>r12ints</tt><dd> This specifies how to store transformed MO integrals.	Valid values are:	<dl>	  <dt><tt>mem-posix</tt><dd> Store integrals in memory for single-pass situations	  and in a binary file on task 0's node using POSIX I/O for multipass situations.	  <tt>posix</tt> is usually less efficient than <tt>mpi</tt> for distributed	  parallel multipass runs since the I/O is performed by one task only. However, this method guaranteed to          work in all types of environments, hence <tt>mem-posix</tt> is the default.	  <dt><tt>posix</tt><dd> Store integrals in a binary file on task 0's node using POSIX I/O.	  This method is different from <tt>mem-posix</tt> in that it forces the integrals out to disk          even if they could be stored in memory. <tt>posix</tt> should only be used for benchmarking          and testing purposes.	  <dt><tt>mem-mpi</tt><dd> Store integrals in memory for single-pass situations	  and in a binary file using MPI-I/O for multipass situations. This method	  assumes the availability of MPI-I/O. <tt>mem-mpi</tt> is the preferred choice	  in distributed environments which have MPI-I/O available.	  <dt><tt>mpi</tt><dd> Store integrals in a binary file using MPI-I/O. This method	  is different from <tt>mem-mpi</tt> in that it forces the integrals out to disk	  even if they could be stored in memory. <tt>mpi</tt> should only be used for benchmarking	  and testing purposes.	  <dt><tt>mem</tt><dd> Store integrals in memory. Can only be used with single-pass	  transformations. This method should only be used for testing purposes	</dl>	If <tt>r12ints</tt> is not specified, then <tt>mem-posix</tt> method will be used.	If user wishes to use MPI-I/O, pending its availability, for higher parallel efficiency,	<tt>r12ints</tt> should be explicitly set to <tt>mem-mpi</tt>.        <dt><tt>r12ints_file</tt><dd> This specifies which file to use to store transformed	MO integrals if <tt>r12ints=posix-io</tt> or <tt>r12ints=mpi-io</tt> is used.	Default is "./<inputbasename>.r12ints.dat", where <inputbasename> is the name of the input	file without ".in". If MPI-I/O is used then it is user's responsibility to ensure	that the file resides on a file system that supports MPI-I/O.        </dl> */    MBPT2_R12(const Ref<KeyVal>&);    ~MBPT2_R12();    void save_data_state(StateOut&);    Ref<GaussianBasisSet> aux_basis() const;    LinearR12::StandardApproximation stdapprox() const;    bool spinadapted() const;    R12IntEvalInfo::StoreMethod r12ints_method() const;    char* r12ints_file() const;    double corr_energy();    double r12_corr_energy();    RefSymmSCMatrix density();    void obsolete();    int gradient_implemented() const;    int value_implemented() const;    void print(std::ostream&o=ExEnv::out0()) const;};}#endif// Local Variables:// mode: c++// c-file-style: "CLJ"// End:

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -