Gromacs源码收获（四）

mdrunner函数在runner.c和runner_openmm.c中均有，是MD模拟的主要程序之一。但是对另一个do_md函数的调用一直是个谜，今天终于看出了点头绪。

这里以runner.c中的mdrunner作为例子：

先看一个Interator的定义，在这里我们发现了do_md的影子。

Interator有积分器，综合者的意思

/* The array should match theeI array in include/types/enums.h */

const gmx_intp_t Interator[eiNR] = { {do_md},{do_steep}, {do_cg}, {do_md}, {do_md}, {do_nm}, {do_lbfgs}, {do_tpi}, {do_tpi},{do_md}, {do_md}, {do_md}};

就好奇gmx_intp_t是什么类型的结构，看一下：

typedef struct {

gmx_integrator_t *func;

} gmx_intp_t;

位置就在其定义上面，然后看下gmx_integrator_t的定义：

typedef doublegmx_integrator_t (FILE *log, t_commrec *cr,

int nfile,const t_filenm fnm[],

const output_env_toenv, gmx_bool bVerbose,

gmx_boolbCompact, int nstglobalcomm,

gmx_vsite_t*vsite, gmx_constr_t constr,

int stepout,

t_inputrec*inputrec,

gmx_mtop_t*mtop, t_fcdata *fcd,

t_state*state,

t_mdatoms*mdatoms,

t_nrnb *nrnb, gmx_wallcycle_t wcycle,

gmx_edsam_ted,

t_forcerec*fr,

intrepl_ex_nst, int repl_ex_nex, int repl_ex_seed,

gmx_membed_t membed,

realcpt_period, real max_hours,

const char*deviceOptions,

unsigned longFlags,

gmx_runtime_t*runtime);

原来是一个方法的typedef声明，那integrator就是eiNR个包含gmx_integrator_t函数的gmx_intp_t结构体数组。有点绕。

通过integrator定义前面的注释，我们找到了eiNR在enums.h中的定义

enum {

eiMD, eiSteep, eiCG, eiBD,eiSD2, eiNM, eiLBFGS, eiTPI, eiTPIC, eiSD1, eiVV, eiVVAK, eiNR

};

但是不理解里面说的match是怎么match的。

下一次对主要嫌疑人的integrator见面在这：

if (integrator[inputrec->eI].func ==do_md)

{

/* Turn on signal handling on all nodes*/

* (A user signal from the PME nodes(if any)

* is communicated to the PP nodes.

signal_handler_install();

}

这里好像注册了一个信号，具体是干什么的，还未知。

然后就是：

/* Now do whatever the userwants us to do (how flexible...) */

integrator[inputrec->eI].func(fplog,cr, nfile, fnm,

oenv,bVerbose, bCompact,

nstglobalcomm,

vsite,constr,

nstepout,inputrec, mtop,

fcd,state,

mdatoms,nrnb, wcycle, ed, fr,

repl_ex_nst, repl_ex_nex, repl_ex_seed,

membed,

cpt_period, max_hours,

deviceOptions,

Flags,

&runtime);

呵呵，找到函数调用了，看一下。首先看看inputrec->eI：

t_inputrec *inputrec; //mdrunner函数中的变量

//结构定义

typedef struct {

int eI; /* Integration method */

gmx_large_int_t nsteps; /* number of steps to be taken */

int simulation_part; /* Used in checkpointing to separatechunks */

gmx_large_int_t init_step; /* start at a stepcount >0 (usedw. tpbconv) */

int nstcalcenergy; /* frequency of energy calc. and T/Pcoupl. upd. */

int cutoff_scheme; /* group or verlet cutoffs */

int ns_type; /* which ns method should weuse? */

int nstlist; /* number of steps beforepairlist is generated */

int ndelta; /* number of cells per rlong */

int nstcomm; /* number of steps after which center of mass */

/* motion is removed */

int comm_mode; /* Center of mass motion removalalgorithm */

int nstcheckpoint; /* checkpointing frequency */

int nstlog; /* number of steps after whichprint to logfile */

int nstxout; /* number of steps after which Xis output */

int nstvout; /* id. for V */

int nstfout; /* id. for F */

int nstenergy; /* number of steps after whichenergies printed */

int nstxtcout; /* id. for compressed trj (.xtc) */

double init_t; /* initial time (ps) */

double delta_t; /* time step (ps) */

real xtcprec; /* precision of xtc file */

real fourier_spacing; /*requested fourier_spacing, when nk? not set */

int nkx, nky, nkz; /* number of k vectors in each spatialdimension*/

/*for fourier methods for long range electrost.*/

int pme_order; /* interpolation order for PME */

real ewald_rtol; /* Real space tolerance for Ewald,determines */

/*the real/reciprocal space relative weight */

int ewald_geometry; /* normal/3d ewald, or pseudo-2d LRcorrections */

real epsilon_surface; /* Epsilon for PME dipole correction */

gmx_bool bOptFFT; /* optimize the fft plan atstart */

int ePBC; /* Type of periodic boundaryconditions */

int bPeriodicMols; /* Periodic molecules */

gmx_bool bContinuation; /* Continuation run: starting state iscorrect */

int etc; /* temperature coupling */

int nsttcouple; /* interval in steps for temperaturecoupling */

gmx_bool bPrintNHChains; /*whether to print nose-hoover chains */

int epc; /* pressure coupling */

int epct; /* pressure coupling type */

int nstpcouple; /* interval in steps for pressurecoupling */

real tau_p; /* pressure coupling time (ps) */

tensor ref_p; /* reference pressure (kJ/(molnm^3)) */

tensor compress; /* compressability ((mol nm^3)/kJ) */

int refcoord_scaling; /* How to scale absolute referencecoordinates */

rvec posres_com; /* The COM of the posres atoms */

rvec posres_comB; /* The B-state COM of the posresatoms */

int andersen_seed; /* Random seed for Andersen thermostat(obsolete) */

real verletbuf_drift; /* Max. drift (kJ/mol/ps/atom) for listbuffer */

real rlist; /* short range pairlist cut-off(nm) */

real rlistlong; /* long range pairlist cut-off (nm) */

int nstcalclr; /* Frequency of evaluating directspace long-range interactions */

real rtpi; /* Radius for test particleinsertion */

int coulombtype; /* Type of electrostaticstreatment */

int coulomb_modifier; /* Modify the Coulomb interaction */

real rcoulomb_switch; /* Coulomb switch range start (nm) */

real rcoulomb; /* Coulomb cutoff (nm) */

real epsilon_r; /* relative dielectricconstant */

real epsilon_rf; /* relative dielectric constant ofthe RF */

int implicit_solvent; /* No (=explicit water), or GBSA solventmodels */

int gb_algorithm; /* Algorithm to use for calculationBorn radii */

int nstgbradii; /* Frequency of updating GeneralizedBorn radii */

real rgbradii; /* Cutoff for GB radiicalculation */

real gb_saltconc; /* Salt concentration (M) for GBSAmodels */

real gb_epsilon_solvent; /* dielectric coeff. of implicitsolvent */

real gb_obc_alpha; /* 1st scaling factor for Bashford-CaseGB */

real gb_obc_beta; /* 2nd scaling factor forBashford-Case GB */

real gb_obc_gamma; /* 3rd scaling factor forBashford-Case GB */

real gb_dielectric_offset; /* Dielectricoffset for Still/HCT/OBC */

int sa_algorithm; /* Algorithm for SA part of GBSA */

real sa_surface_tension; /* Energy factor for SA part of GBSA */

int vdwtype; /* Type of Van der Waalstreatment */

int vdw_modifier; /* Modify the VdW interaction */

real rvdw_switch; /* Van der Waals switch range start(nm) */

real rvdw; /* Van der Waals cutoff (nm) */

int eDispCorr; /* Perform Long range dispersioncorrections */

real tabext; /* Extension of the table beyondthe cut-off, *

* as well asthe table length for 1-4 interac. */

real shake_tol; /* tolerance for shake */

int efep; /* free energycalculations */

t_lambda *fepvals; /* Data for the FEP state */

gmx_bool bSimTemp; /* Whether to do simulatedtempering */

t_simtemp *simtempvals; /* Variables for simulated tempering */

gmx_bool bExpanded; /* Whether expanded ensembles areused */

t_expanded *expandedvals; /* Expanded ensemble parameters */

int eDisre; /* Type of distancerestraining */

real dr_fc; /* force constant for ta_disre */

int eDisreWeighting; /* type of weighting of pairs in onerestraints */

gmx_bool bDisreMixed; /* Use comb of time averaged andinstan. viol's */

int nstdisreout; /* frequency of writing pairdistances to enx */

real dr_tau; /* time constant for memoryfunction in disres */

real orires_fc; /* force constant for orientationalrestraints */

real orires_tau; /* time constant for memory functionin orires */

int nstorireout; /* frequency of writing tr(SD) toenx */

real dihre_fc; /* force constant for dihedralrestraints (obsolete) */

real em_stepsize; /* The stepsize for updating */

real em_tol; /* The tolerance */

int niter; /* Number of iterations forconvergence of */

/*steepest descent in relax_shells */

real fc_stepsize; /* Stepsize for directionalminimization */

/* inrelax_shells */

int nstcgsteep; /* number of steps after which asteepest */

/*descents step is done while doing cg */

int nbfgscorr; /* Number of corrections to thehessian to keep */

int eConstrAlg; /* Type of constraint algorithm */

int nProjOrder; /* Order of the LINCS ProjectionAlgorithm */

real LincsWarnAngle; /* If bond rotates more than %g degrees,warn */

int nLincsIter; /* Number of iterations in the finalLincs step */

gmx_bool bShakeSOR; /* Use successive overrelaxationfor shake */

real bd_fric; /* Friction coefficient for BD(amu/ps) */

int ld_seed; /* Random seed for SD and BD */

int nwall; /* The number of walls */

int wall_type; /* The type of walls */

real wall_r_linpot; /* The potentail is linear forr<=wall_r_linpot */

int wall_atomtype[2]; /* The atom type for walls */

real wall_density[2]; /* Number density for walls */

real wall_ewald_zfac; /*Scaling factor for the box for Ewald */

int ePull; /* Type of pulling: no,umbrella or constraint */

t_pull *pull; /* The data for center of masspulling */

gmx_bool bRot; /* Calculate enforced rotationpotential(s)? */

t_rot *rot; /* The data for enforcedrotation potentials */

real cos_accel; /* Acceleration for viscositycalculation */

tensor deform; /* Triclinic deformationvelocities (nm/ps) */

int userint1; /* User determined parameters */

int userint2;

int userint3;

int userint4;

real userreal1;

real userreal2;

real userreal3;

real userreal4;

t_grpopts opts; /* Group options */

t_cosines ex[DIM]; /* Electric field stuff (spatial part) */

t_cosines et[DIM]; /* Electric field stuff (time part) */

gmx_bool bQMMM; /* QM/MM calculation */

int QMconstraints; /* constraints onQM bonds */

int QMMMscheme; /* Scheme: ONIOM or normal */

real scalefactor; /* factor for scaling the MM charges in QMcalc.*/

/* parameterneeded for AdResS simulation */

gmx_bool bAdress; /* Is AdResS enabled ? */

t_adress *adress; /* The data for adress simulations */

} t_inputrec;

然后调用对应的func函数完成函数调用，但是还有一个疑问inputrec的初始化和值的标定在哪里？继续找

snew(inputrec, 1);这里是申请空间。

这里

* Read (nearly) all datarequired for the simulation */

read_tpx_state(ftp2fn(efTPX, nfile,fnm), inputrec, state, NULL, mtop);

看下t_filenm的定义：

typedef struct {

int ftp; /* File type (see enum above) */

const char *opt; /* Command line option */

const char *fn; /* File name (as set insource code) */

unsigned long flag; /* Flag for all kinds of info (see defs)*/

int nfiles; /* number of files */

char **fns; /* File names */

} t_filenm;

总之是读取一个类型的文件，然后通过文件初始化。

通过Gromacs网站了解到调用的方法取决于mdp文件中的整型量。但是一般情况是调用的do_md方法。

Gromacs源码收获（四）相关推荐

【投屏】Scrcpy源码分析四（最终章 - Server篇）
Scrcpy源码分析系列 [投屏]Scrcpy源码分析一(编译篇) [投屏]Scrcpy源码分析二(Client篇-连接阶段) [投屏]Scrcpy源码分析三(Client篇-投屏阶段) [投屏]Sc ...
一点一点看JDK源码（四）java.util.ArrayList 中篇
一点一点看JDK源码(四)java.util.ArrayList 中篇 liuyuhang原创,未经允许禁止转载本文举例使用的是JDK8的API 目录:一点一点看JDK源码(〇) 1.综述在前篇中 ...
mybatis源码阅读(四)：mapper(dao)实例化
转载自 mybatis源码阅读(四):mapper(dao)实例化在开始分析之前,先来了解一下这个模块中的核心组件之间的关系,如图: 1.MapperRegistry&MapperPro ...
39 网络相关函数(七)——live555源码阅读(四)网络
39 网络相关函数(七)--live555源码阅读(四)网络 39 网络相关函数(七)--live555源码阅读(四)网络简介 14)readSocket从套接口读取数据 recv/recvfrom ...
ABP源码分析四十七：ABP中的异常处理
ABP源码分析四十七:ABP中的异常处理参考文章: (1)ABP源码分析四十七:ABP中的异常处理 (2)https://www.cnblogs.com/1zhk/p/5538983.html (3 ...
gSOAP 源码分析(四)
gSOAP 源码分析(四) 2012-6-2 邵盛松前言本文主要说明gSOAP中对Client的认证分析 gSOAP中包含了HTTP基本认证,NTLM认证等,还可以自定义SOAP Heard实现认 ...
hdl_graph_slam源码解读(四)：关键帧
目录 1. 整体介绍 2. 详细介绍 2.1 KeyFrame类 2.2 KeyFrameSnapshot类 2.3 KeyFrameUpdater类参考文献 1. 整体介绍在图优化中,关键帧的作 ...
深入SpringBoot源码（四）初识Environment
深入SpringBoot源码(四)初识Environment new DefaultApplicationArguments prepareEnvironment getOrCreateEnviron ...
Spring 源码分析(四) ——MVC（二）概述
随时随地技术实战干货,获取项目源码.学习资料,请关注源代码社区公众号(ydmsq666) from:Spring 源码分析(四) --MVC(二)概述 - 水门-kay的个人页面 - OSCHINA ...

Gromacs源码收获（四）

Gromacs源码收获（四）相关推荐

最新文章

热门文章