cantera/Cantera/src/BasisOptimize.cpp
2007-01-27 16:54:48 +00:00

818 lines
25 KiB
C++

/**
* @file basopt.cpp
*
* $Author$
* $Date$
* $Revision$
*/
#include "ct_defs.h"
#include "ThermoPhase.h"
#include "MultiPhase.h"
using namespace Cantera;
using namespace std;
#ifdef DEBUG_HKM
namespace Cantera {
int Cantera::BasisOptimize_print_lvl = 0;
static char sbuf[1024];
}
static void print_stringTrunc(const char *str, int space, int alignment);
#endif
static int amax(double *x, int j, int n);
static void switch_pos(vector_int &orderVector, int jr, int kspec);
static int mlequ(double *c, int idem, int n, double *b, int m);
#ifndef MIN
#define MIN(x,y) (( (x) < (y) ) ? (x) : (y))
#endif
/**
* Choose the optimum basis for the calculations. This is done by
* choosing the species with the largest mole fraction
* not currently a linear combination of the previous components.
* Then, calculate the stoichiometric coefficient matrix for that
* basis.
*
* Calculates the identity of the component species in the mechanism.
* Rearranges the solution data to put the component data at the
* front of the species list.
*
* Then, calculates SC(J,I) the formation reactions for all noncomponent
* species in the mechanism.
*
* Input
* ---------
* mphase Pointer to the multiphase object. Contains the
* species mole fractions, which are used to pick the
* current optimal species component basis.
* orderVectorElement
* Order vector for the elements. The element rows
* in the formula matrix are
* rearranged according to this vector.
* orderVectorSpecies
* Order vector for the species. The species are
* rearranged according to this formula. The first
* nCompoments of this vector contain the calculated
* species components on exit.
* doFormRxn If true, the routine calculates the formation
* reaction matrix based on the calculated
* component species. If false, this step is skipped.
*
* Output
* ---------
* usedZeroedSpecies = If true, then a species with a zero concentration
* was used as a component. The problem may be
* converged.
* formRxnMatrix
*
* Return
* --------------
* returns the number of components.
*
*
*/
int Cantera::BasisOptimize(int *usedZeroedSpecies, bool doFormRxn,
MultiPhase *mphase, vector_int & orderVectorSpecies,
vector_int & orderVectorElements,
vector_fp & formRxnMatrix) {
int j, jj, k, kk, l, i, jl, ml;
bool lindep;
std::string ename;
std::string sname;
/*
* Get the total number of elements defined in the multiphase object
*/
int ne = mphase->nElements();
/*
* Get the total number of species in the multiphase object
*/
int nspecies = mphase->nSpecies();
doublereal tmp;
doublereal const USEDBEFORE = -1;
/*
* Perhaps, initialize the element ordering
*/
if ((int) orderVectorElements.size() < ne) {
orderVectorElements.resize(ne);
for (j = 0; j < ne; j++) {
orderVectorElements[j] = j;
}
}
/*
* Perhaps, initialize the species ordering
*/
if ((int) orderVectorSpecies.size() != nspecies) {
orderVectorSpecies.resize(nspecies);
for (k = 0; k < nspecies; k++) {
orderVectorSpecies[k] = k;
}
}
#ifdef DEBUG_HKM
double molSave = 0.0;
if (BasisOptimize_print_lvl >= 1) {
writelog(" "); for(i=0; i<77; i++) writelog("-"); writelog("\n");
writelog(" --- Subroutine BASOPT called to ");
writelog("calculate the number of components and ");
writelog("evaluate the formation matrix\n");
if (BasisOptimize_print_lvl > 0) {
writelog(" ---\n");
writelog(" --- Formula Matrix used in BASOPT calculation\n");
writelog(" --- Species | Order | ");
for (j = 0; j < ne; j++) {
jj = orderVectorElements[j];
writelog(" ");
ename = mphase->elementName(jj);
print_stringTrunc(ename.c_str(), 4, 1);
sprintf(sbuf,"(%1d)", j); writelog(sbuf);
}
writelog("\n");
for (k = 0; k < nspecies; k++) {
kk = orderVectorSpecies[k];
writelog(" --- ");
sname = mphase->speciesName(kk);
print_stringTrunc(sname.c_str(), 11, 1);
sprintf(sbuf," | %4d |", k); writelog(sbuf);
for (j = 0; j < ne; j++) {
jj = orderVectorElements[j];
double num = mphase->nAtoms(kk,jj);
sprintf(sbuf,"%6.1g ", num); writelog(sbuf);
}
writelog("\n");
}
writelog(" --- \n");
}
}
#endif
/*
* Calculate the maximum value of the number of components possible
* It's equal to the minimum of the number of elements and the
* number of total species.
*/
int nComponents = MIN(ne, nspecies);
int nNonComponents = nspecies - nComponents;
/*
* Set this return variable to false
*/
*usedZeroedSpecies = false;
/*
* Create an array of mole numbers
*/
vector_fp molNum(nspecies,0.0);
mphase->getMoles(DATA_PTR(molNum));
/*
* Other workspace
*/
vector_fp sm(ne*ne, 0.0);
vector_fp ss(ne, 0.0);
vector_fp sa(ne, 0.0);
if ((int) formRxnMatrix.size() < nspecies*ne) {
formRxnMatrix.resize(nspecies*ne, 0.0);
}
#ifdef DEBUG_HKM
/*
* For debugging purposes keep an unmodified copy of the array.
*/
vector_fp molNumBase(molNum);
#endif
int jr = -1;
/*
* Top of a loop of some sort based on the index JR. JR is the
* current number of component species found.
*/
do {
++jr;
/* - Top of another loop point based on finding a linearly */
/* - independent species */
do {
/*
* Search the remaining part of the mole number vector, molNum
* for the largest remaining species. Return its identity.
* kk is the raw number. k is the orderVectorSpecies index.
*/
kk = amax(DATA_PTR(molNum), 0, nspecies);
for (j = 0; j < nspecies; j++) {
if (orderVectorSpecies[j] == kk) {
k = j;
break;
}
}
if (j == nspecies) {
throw CanteraError("BasisOptimize", "orderVectorSpecies contains an error");
}
if (molNum[kk] == 0.0) *usedZeroedSpecies = true;
/*
* If the largest molNum is negative, then we are done.
*/
if (molNum[kk] == USEDBEFORE) {
nComponents = jr;
nNonComponents = nspecies - nComponents;
goto L_END_LOOP;
}
/*
* Assign a small negative number to the component that we have
* just found, in order to take it out of further consideration.
*/
#ifdef DEBUG_HKM
molSave = molNum[kk];
#endif
molNum[kk] = USEDBEFORE;
/* *********************************************************** */
/* **** CHECK LINEAR INDEPENDENCE WITH PREVIOUS SPECIES ****** */
/* *********************************************************** */
/*
* Modified Gram-Schmidt Method, p. 202 Dalquist
* QR factorization of a matrix without row pivoting.
*/
jl = jr;
for (j = 0; j < ne; ++j) {
jj = orderVectorElements[j];
sm[j + jr*ne] = mphase->nAtoms(kk,jj);
}
if (jl > 0) {
/*
* Compute the coefficients of JA column of the
* the upper triangular R matrix, SS(J) = R_J_JR
* (this is slightly different than Dalquist)
* R_JA_JA = 1
*/
for (j = 0; j < jl; ++j) {
ss[j] = 0.0;
for (i = 0; i < ne; ++i) {
ss[j] += sm[i + jr*ne] * sm[i + j*ne];
}
ss[j] /= sa[j];
}
/*
* Now make the new column, (*,JR), orthogonal to the
* previous columns
*/
for (j = 0; j < jl; ++j) {
for (l = 0; l < ne; ++l) {
sm[l + jr*ne] -= ss[j] * sm[l + j*ne];
}
}
}
/*
* Find the new length of the new column in Q.
* It will be used in the denominator in future row calcs.
*/
sa[jr] = 0.0;
for (ml = 0; ml < ne; ++ml) {
tmp = sm[ml + jr*ne];
sa[jr] += tmp * tmp;
}
/* **************************************************** */
/* **** IF NORM OF NEW ROW .LT. 1E-3 REJECT ********** */
/* **************************************************** */
if (sa[jr] < 1.0e-6) lindep = true;
else lindep = false;
} while(lindep);
/* ****************************************** */
/* **** REARRANGE THE DATA ****************** */
/* ****************************************** */
if (jr != k) {
#ifdef DEBUG_HKM
if (BasisOptimize_print_lvl >= 1) {
kk = orderVectorSpecies[k];
sname = mphase->speciesName(kk);
sprintf(sbuf," --- %-12.12s", sname.c_str()); writelog(sbuf);
jj = orderVectorSpecies[jr];
ename = mphase->speciesName(jj);
sprintf(sbuf,"(%9.2g) replaces %-12.12s", molSave, ename.c_str());
writelog(sbuf);
sprintf(sbuf,"(%9.2g) as component %3d\n", molNum[jj], jr);
writelog(sbuf);
}
#endif
switch_pos(orderVectorSpecies, jr, k);
}
/* - entry point from up above */
L_END_LOOP: ;
/*
* If we haven't found enough components, go back
* and find some more. (nc -1 is used below, because
* jr is counted from 0, via the C convention.
*/
} while (jr < (nComponents-1));
if (! doFormRxn) return nComponents;
/* ****************************************************** */
/* **** EVALUATE THE STOICHIOMETRY ********************** */
/* ****************************************************** */
/*
* Formulate the matrix problem for the stoichiometric
* coefficients. CX + B = 0
* C will be an nc x nc matrix made up of the formula
* vectors for the components. Each component's formular
* vector is a column. The rows are the elements.
* n rhs's will be solved for. Thus, B is an nc x n
* matrix.
*
* BIG PROBLEM 1/21/99:
*
* This algorithm makes the assumption that the
* first nc rows of the formula matrix aren't rank deficient.
* However, this might not be the case. For example, assume
* that the first element in FormulaMatrix[] is argon. Assume that
* no species in the matrix problem actually includes argon.
* Then, the first row in sm[], below will be indentically
* zero. bleh.
* What needs to be done is to perform a rearrangement
* of the ELEMENTS -> i.e. rearrange, FormulaMatrix, sp, and gai, such
* that the first nc elements form in combination with the
* nc components create an invertible sm[]. not a small
* project, but very doable.
* An alternative would be to turn the matrix problem
* below into an ne x nc problem, and do QR elimination instead
* of Gauss-Jordon elimination.
* Note the rearrangement of elements need only be done once
* in the problem. It's actually very similar to the top of
* this program with ne being the species and nc being the
* elements!!
*/
for (k = 0; k < nComponents; ++k) {
kk = orderVectorSpecies[k];
for (j = 0; j < nComponents; ++j) {
jj = orderVectorElements[j];
sm[j + k*ne] = mphase->nAtoms(kk, jj);
}
}
for (i = 0; i < nNonComponents; ++i) {
k = nComponents + i;
kk = orderVectorSpecies[k];
for (j = 0; j < nComponents; ++j) {
jj = orderVectorElements[j];
formRxnMatrix[j + i * ne] = mphase->nAtoms(kk, jj);
}
}
/*
* Use Gauss-Jordon block elimination to calculate
* the reaction matrix
*/
j = mlequ(DATA_PTR(sm), ne, nComponents, DATA_PTR(formRxnMatrix), nNonComponents);
if (j == 1) {
writelog("ERROR: mlequ returned an error condition\n");
throw CanteraError("basopt", "mlequ returned an error condition");
}
#ifdef DEBUG_HKM
if (Cantera::BasisOptimize_print_lvl >= 1) {
writelog(" ---\n");
sprintf(sbuf," --- Number of Components = %d\n", nComponents);
writelog(sbuf);
writelog(" --- Formula Matrix:\n");
writelog(" --- Components: ");
for (k = 0; k < nComponents; k++) {
kk = orderVectorSpecies[k];
sprintf(sbuf," %3d (%3d) ", k, kk); writelog(sbuf);
}
writelog("\n --- Components Moles: ");
for (k = 0; k < nComponents; k++) {
kk = orderVectorSpecies[k];
sprintf(sbuf,"%-11.3g", molNumBase[kk]); writelog(sbuf);
}
writelog("\n --- NonComponent | Moles | ");
for (i = 0; i < nComponents; i++) {
kk = orderVectorSpecies[i];
sname = mphase->speciesName(kk);
sprintf(sbuf,"%-11.10s", sname.c_str()); writelog(sbuf);
}
writelog("\n");
for (i = 0; i < nNonComponents; i++) {
k = i + nComponents;
kk = orderVectorSpecies[k];
sprintf(sbuf," --- %3d (%3d) ", k, kk); writelog(sbuf);
sname = mphase->speciesName(kk);
sprintf(sbuf,"%-10.10s", sname.c_str()); writelog(sbuf);
sprintf(sbuf,"|%10.3g|", molNumBase[kk]); writelog(sbuf);
/*
* Print the negative of formRxnMatrix[]; it's easier to interpret.
*/
for (j = 0; j < nComponents; j++) {
sprintf(sbuf," %6.2f", - formRxnMatrix[j + i * ne]);
writelog(sbuf);
}
writelog("\n");
}
writelog(" "); for (i=0; i<77; i++) writelog("-"); writelog("\n");
}
#endif
return nComponents;
} /* basopt() ************************************************************/
#ifdef DEBUG_HKM
static void print_stringTrunc(const char *str, int space, int alignment)
/***********************************************************************
* vcs_print_stringTrunc():
*
* Print a string within a given space limit. This routine
* limits the amount of the string that will be printed to a
* maximum of "space" characters.
*
* str = String -> must be null terminated.
* space = space limit for the printing.
* alignment = 0 centered
* 1 right aligned
* 2 left aligned
***********************************************************************/
{
int i, ls=0, rs=0;
int len = strlen(str);
if ((len) >= space) {
for (i = 0; i < space; i++) {
sprintf(sbuf,"%c", str[i]); writelog(sbuf);
}
} else {
if (alignment == 1) {
ls = space - len;
} else if (alignment == 2) {
rs = space - len;
} else {
ls = (space - len) / 2;
rs = space - len - ls;
}
if (ls != 0) {
for (i = 0; i < ls; i++) writelog(" ");
}
sprintf(sbuf,"%s", str); writelog(sbuf);
if (rs != 0) {
for (i = 0; i < rs; i++) writelog(" ");
}
}
}
#endif
/*
* Finds the location of the maximum component in a double vector
* INPUT
* x(*) - Vector to search
* j <= i < n : i is the range of indecises to search in X(*)
*
* RETURN
* return index of the greatest value on X(*) searched
*/
static int amax(double *x, int j, int n) {
int i;
int largest = j;
double big = x[j];
for (i = j + 1; i < n; ++i) {
if (x[i] > big) {
largest = i;
big = x[i];
}
}
return largest;
}
static void switch_pos(vector_int &orderVector, int jr, int kspec) {
int kcurr = orderVector[jr];
orderVector[jr] = orderVector[kspec];
orderVector[kspec] = kcurr;
}
/*
* vcs_mlequ:
*
* Invert an nxn matrix and solve m rhs's
*
* Solve C X + B = 0;
*
* This routine uses Gauss elimination and is optimized for the solution
* of lots of rhs's.
* A crude form of row pivoting is used here.
*
*
* c[i+j*idem] = c_i_j = Matrix to be inverted: i = row number
* j = column number
* b[i+j*idem] = b_i_j = vectors of rhs's: i = row number
* j = column number
* (each column is a new rhs)
* n = number of rows and columns in the matrix
* m = number of rhs to be solved for
* idem = first dimension in the calling routine
* idem >= n must be true
*
* Return Value
* 1 : Matrix is singluar
* 0 : solution is OK
*
* The solution is returned in the matrix b.
*/
static int mlequ(double *c, int idem, int n, double *b, int m) {
int i, j, k, l;
double R;
/*
* Loop over the rows
* -> At the end of each loop, the only nonzero entry in the column
* will be on the diagonal. We can therfore just invert the
* diagonal at the end of the program to solve the equation system.
*/
for (i = 0; i < n; ++i) {
if (c[i + i * idem] == 0.0) {
/*
* Do a simple form of row pivoting to find a non-zero pivot
*/
for (k = i + 1; k < n; ++k) {
if (c[k + i * idem] != 0.0) goto FOUND_PIVOT;
}
#ifdef DEBUG_HKM
sprintf(sbuf,"vcs_mlequ ERROR: Encountered a zero column: %d\n", i);
writelog(sbuf);
#endif
return 1;
FOUND_PIVOT: ;
for (j = 0; j < n; ++j) c[i + j * idem] += c[k + j * idem];
for (j = 0; j < m; ++j) b[i + j * idem] += b[k + j * idem];
}
for (l = 0; l < n; ++l) {
if (l != i && c[l + i * idem] != 0.0) {
R = c[l + i * idem] / c[i + i * idem];
c[l + i * idem] = 0.0;
for (j = i+1; j < n; ++j) c[l + j * idem] -= c[i + j * idem] * R;
for (j = 0; j < m; ++j) b[l + j * idem] -= b[i + j * idem] * R;
}
}
}
/*
* The negative in the last expression is due to the form of B upon
* input
*/
for (i = 0; i < n; ++i) {
for (j = 0; j < m; ++j)
b[i + j * idem] = -b[i + j * idem] / c[i + i*idem];
}
return 0;
} /* mlequ() *************************************************************/
/**
*
* ElemRearrange:
*
* This subroutine handles the rearrangement of the constraint
* equations represented by the Formula Matrix. Rearrangement is only
* necessary when the number of components is less than the number of
* elements. For this case, some constraints can never be satisfied
* exactly, because the range space represented by the Formula
* Matrix of the components can't span the extra space. These
* constraints, which are out of the range space of the component
* Formula matrix entries, are migrated to the back of the Formula
* matrix.
*
* A prototypical example is an extra element column in
* FormulaMatrix[],
* which is identically zero. For example, let's say that argon is
* has an element column in FormulaMatrix[], but no species in the
* mechanism
* actually contains argon. Then, nc < ne. Unless the entry for
* desired elementabundance vector for Ar is zero, then this
* element abundance constraint can never be satisfied. The
* constraint vector is not in the range space of the formula
* matrix.
* Also, without perturbation
* of FormulaMatrix[], BasisOptimize[] would produce a zero pivot
* because the matrix
* would be singular (unless the argon element column was already the
* last column of FormulaMatrix[].
* This routine borrows heavily from BasisOptimize algorithm. It
* finds nc constraints which span the range space of the Component
* Formula matrix, and assigns them as the first nc components in the
* formular matrix. This guarrantees that BasisOptimize has a
* nonsingular matrix to invert.
*/
int Cantera::ElemRearrange(int nComponents, const vector_fp & elementAbundances,
MultiPhase *mphase,
vector_int & orderVectorSpecies,
vector_int & orderVectorElements) {
int j, k, l, i, jl, ml, jr, ielem, jj, kk;
bool lindep = false;
int nelements = mphase->nElements();
std::string ename;
/*
* Get the total number of species in the multiphase object
*/
int nspecies = mphase->nSpecies();
double test = -1.0E10;
#ifdef DEBUG_HKM
if (BasisOptimize_print_lvl > 0) {
writelog(" "); for(i=0; i<77; i++) writelog("-"); writelog("\n");
writelog(" --- Subroutine ElemRearrange() called to ");
writelog("check stoich. coefficent matrix\n");
writelog(" --- and to rearrange the element ordering once\n");
}
#endif
/*
* Perhaps, initialize the element ordering
*/
if ((int) orderVectorElements.size() < nelements) {
orderVectorElements.resize(nelements);
for (j = 0; j < nelements; j++) {
orderVectorElements[j] = j;
}
}
/*
* Perhaps, initialize the species ordering. However, this is
* dangerous, as this ordering is assumed to yield the
* component species for the problem
*/
if ((int) orderVectorSpecies.size() != nspecies) {
orderVectorSpecies.resize(nspecies);
for (k = 0; k < nspecies; k++) {
orderVectorSpecies[k] = k;
}
}
/*
* If the elementAbundances aren't input, just create a fake one
* based on summing the column of the stoich matrix.
* This will force elements with zero species to the
* end of the element ordering.
*/
vector_fp eAbund(nelements,0.0);
if ((int) elementAbundances.size() != nelements) {
for (j = 0; j < nelements; j++) {
eAbund[j] = 0.0;
for (k = 0; k < nspecies; k++) {
eAbund[j] += fabs(mphase->nAtoms(k, j));
}
}
} else {
copy(elementAbundances.begin(), elementAbundances.end(),
eAbund.begin());
}
vector_fp sa(nelements,0.0);
vector_fp ss(nelements,0.0);
vector_fp sm(nelements*nelements,0.0);
/*
* Top of a loop of some sort based on the index JR. JR is the
* current number independent elements found.
*/
jr = -1;
do {
++jr;
/*
* Top of another loop point based on finding a linearly
* independent element
*/
do {
/*
* Search the element vector. We first locate elements that
* are present in any amount. Then, we locate elements that
* are not present in any amount.
* Return its identity in K.
*/
k = nelements;
for (ielem = jr; ielem < nelements; ielem++) {
kk = orderVectorElements[ielem];
if (eAbund[kk] != test && eAbund[kk] > 0.0) {
k = ielem;
break;
}
}
for (ielem = jr; ielem < nelements; ielem++) {
kk = orderVectorElements[ielem];
if (eAbund[kk] != test) {
k = ielem;
break;
}
}
if (k == nelements) {
// When we are here, there is an error usually.
// We haven't found the number of elements necessary.
// This is signalled by returning jr != nComponents.
#ifdef DEBUG_HKM
if (BasisOptimize_print_lvl > 0) {
sprintf(sbuf,"Error exit: returning with nComponents = %d\n", jr);
writelog(sbuf);
}
#endif
return jr;
}
/*
* Assign a large negative number to the element that we have
* just found, in order to take it out of further consideration.
*/
eAbund[kk] = test;
/* *********************************************************** */
/* **** CHECK LINEAR INDEPENDENCE OF CURRENT FORMULA MATRIX */
/* **** LINE WITH PREVIOUS LINES OF THE FORMULA MATRIX ****** */
/* *********************************************************** */
/*
* Modified Gram-Schmidt Method, p. 202 Dalquist
* QR factorization of a matrix without row pivoting.
*/
jl = jr;
/*
* Fill in the row for the current element, k, under consideration
* The row will contain the Formula matrix value for that element
* with respect to the vector of component species.
* (note j and k indecises are flipped compared to the previous routine)
*/
for (j = 0; j < nComponents; ++j) {
jj = orderVectorSpecies[j];
kk = orderVectorElements[k];
sm[j + jr*nComponents] = mphase->nAtoms(jj,kk);
}
if (jl > 0) {
/*
* Compute the coefficients of JA column of the
* the upper triangular R matrix, SS(J) = R_J_JR
* (this is slightly different than Dalquist)
* R_JA_JA = 1
*/
for (j = 0; j < jl; ++j) {
ss[j] = 0.0;
for (i = 0; i < nComponents; ++i) {
ss[j] += sm[i + jr*nComponents] * sm[i + j*nComponents];
}
ss[j] /= sa[j];
}
/*
* Now make the new column, (*,JR), orthogonal to the
* previous columns
*/
for (j = 0; j < jl; ++j) {
for (l = 0; l < nComponents; ++l) {
sm[l + jr*nComponents] -= ss[j] * sm[l + j*nComponents];
}
}
}
/*
* Find the new length of the new column in Q.
* It will be used in the denominator in future row calcs.
*/
sa[jr] = 0.0;
for (ml = 0; ml < nComponents; ++ml) {
double tmp = sm[ml + jr*nComponents];
sa[jr] += tmp * tmp;
}
/* **************************************************** */
/* **** IF NORM OF NEW ROW .LT. 1E-6 REJECT ********** */
/* **************************************************** */
if (sa[jr] < 1.0e-6) lindep = true;
else lindep = false;
} while(lindep);
/* ****************************************** */
/* **** REARRANGE THE DATA ****************** */
/* ****************************************** */
if (jr != k) {
#ifdef DEBUG_HKM
if (BasisOptimize_print_lvl > 0) {
kk = orderVectorElements[k];
ename = mphase->elementName(kk);
writelog(" --- ");
sprintf(sbuf,"%-2.2s", ename.c_str()); writelog(sbuf);
writelog("replaces ");
kk = orderVectorElements[jr];
ename = mphase->elementName(kk);
sprintf(sbuf,"%-2.2s", ename.c_str()); writelog(sbuf);
sprintf(sbuf," as element %3d\n", jr); writelog(sbuf);
}
#endif
switch_pos(orderVectorElements, jr, k);
}
/*
* If we haven't found enough components, go back
* and find some more. (nc -1 is used below, because
* jr is counted from 0, via the C convention.
*/
} while (jr < (nComponents-1));
return nComponents;
} /* vcs_elem_rearrange() ****************************************************/