1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
|
#if (defined(YADE_POTENTIAL_PARTICLES) or defined(YADE_POTENTIAL_BLOCKS))
#if defined(YADE_REAL_BIT) and (YADE_REAL_BIT != 64)
#include "LapackCompatibility.hpp"
#include <lib/high-precision/Real.hpp>
#include <vector>
// when compiling with non-double Real we need the converting wrapper functions.
// TODO: of course all these functions should be replaced to use Eigen, MatrixXr and VectorXr.
// Here are only some workaround/wrappers to use (double precision only) lapack-fortran.
//
/* This is an example which should help in accomplishing this task, it works with Eigen and solves a system of linear equations.
https://eigen.tuxfamily.org/dox/group__MatrixfreeSolverExample.html
// example: #include <iostream>
// example: #include <Eigen/Core>
// example: #include <Eigen/Dense>
// example: #include <Eigen/IterativeLinearSolvers>
// example: #include <unsupported/Eigen/IterativeSolvers>
// example: int main()
{
int n = 10;
Eigen::SparseMatrix<double> A = MatrixXr::Random(n,n).sparseView(0.5,1);
A = A.transpose()*A;
VectorXr b(n), x;
b.setRandom();
// Solve Ax = b using various iterative solver with matrix-free version:
{
Eigen::ConjugateGradient<Eigen::SparseMatrix<double>, Eigen::Lower|Eigen::Upper, Eigen::IdentityPreconditioner> cg;
cg.compute(A);
x = cg.solve(b);
std::cout << "CG: #iterations: " << cg.iterations() << ", estimated error: " << cg.error() << std::endl;
}
{
Eigen::BiCGSTAB<Eigen::SparseMatrix<double>, Eigen::IdentityPreconditioner> bicg;
bicg.compute(A);
x = bicg.solve(b);
std::cout << "BiCGSTAB: #iterations: " << bicg.iterations() << ", estimated error: " << bicg.error() << std::endl;
}
{
Eigen::GMRES<Eigen::SparseMatrix<double>, Eigen::IdentityPreconditioner> gmres;
gmres.compute(A);
x = gmres.solve(b);
std::cout << "GMRES: #iterations: " << gmres.iterations() << ", estimated error: " << gmres.error() << std::endl;
}
{
Eigen::DGMRES<Eigen::SparseMatrix<double>, Eigen::IdentityPreconditioner> gmres;
gmres.compute(A);
x = gmres.solve(b);
std::cout << "DGMRES: #iterations: " << gmres.iterations() << ", estimated error: " << gmres.error() << std::endl;
}
{
Eigen::MINRES<Eigen::SparseMatrix<double>, Eigen::Lower|Eigen::Upper, Eigen::IdentityPreconditioner> minres;
minres.compute(A);
x = minres.solve(b);
std::cout << "MINRES: #iterations: " << minres.iterations() << ", estimated error: " << minres.error() << std::endl;
}
}
*/
std::vector<double> toDoubleVec(const ::yade::Real* from, int size)
{
std::vector<double> ret(size, 0);
for (int i = 0; i < size; i++) {
ret[i] = static_cast<double>(from[i]);
}
return ret;
}
void toRealArrPtr(const std::vector<double>& from, ::yade::Real* to, int size)
{
for (int i = 0; i < size; i++) {
to[i] = static_cast<::yade::Real>(from[i]);
}
}
// TODO: all these functions should be replaced to use Eigen, MatrixXr and VectorXr.
// Here are only some workaround/wrappers to use (double precision only) lapack-fortran.
void dgesv_(const int* N, const int* nrhs, ::yade::Real* Hessian_Real, const int* lda, int* ipiv, ::yade::Real* gradient_Real, const int* ldb, int* info)
{
// DGESV computes the solution to system of linear equations A * X = B for GE matrices
const int leading_dimension_of_the_array_A = *lda;
const int leading_dimension_of_the_array_B = *ldb;
int total_sq_size = leading_dimension_of_the_array_A * leading_dimension_of_the_array_A;
std::vector<double> Hessian_double = toDoubleVec(Hessian_Real, total_sq_size);
std::vector<double> gradient_double = toDoubleVec(gradient_Real, leading_dimension_of_the_array_B);
dgesv_(N, nrhs, Hessian_double.data(), lda, ipiv, gradient_double.data(), ldb, info);
toRealArrPtr(Hessian_double, Hessian_Real, total_sq_size);
toRealArrPtr(gradient_double, gradient_Real, leading_dimension_of_the_array_B);
}
void dpbsv_(
const char* uplo, const int* n, const int* kd, const int* nrhs, ::yade::Real* AB_Real, const int* ldab, ::yade::Real* B_Real, const int* ldb, int* info)
{
// DPBSV computes the solution to system of linear equations A * X = B for OTHER matrices
const int side = *ldab;
int total_size = side * side;
std::vector<double> AB_double = toDoubleVec(AB_Real, total_size);
std::vector<double> B_double = toDoubleVec(B_Real, side);
dpbsv_(uplo, n, kd, nrhs, AB_double.data(), ldab, B_double.data(), ldb, info);
toRealArrPtr(AB_double, AB_Real, total_size);
toRealArrPtr(B_double, B_Real, side);
}
void dgemm_(
const char* transA,
const char* transB,
const int* m,
const int* n,
const int* k,
const ::yade::Real* alpha_Real,
::yade::Real* A_Real,
const int* lda,
::yade::Real* B_Real,
const int* ldb,
const ::yade::Real* beta_Real,
::yade::Real* C_Real,
const int* ldc)
{
// DGEMM performs one of the matrix-matrix operations - FIXME use Eigen instead.
//extern void dgemm_(const char *transA, const char *transB, const int *m, const int *n, const int *k, const double *alpha, double *A, const int *lda, double *B, const int *ldb, const double *beta, double *C, const int *ldc);
const double alpha_double = static_cast<double>(*alpha_Real);
const double beta_double = static_cast<double>(*beta_Real);
std::vector<double> A_double = toDoubleVec(A_Real, (*m) * (*k));
std::vector<double> B_double = toDoubleVec(B_Real, (*k) * (*n));
std::vector<double> C_double = toDoubleVec(C_Real, (*n) * (*m));
dgemm_(transA, transB, m, n, k, &alpha_double, A_double.data(), lda, B_double.data(), ldb, &beta_double, C_double.data(), ldc);
toRealArrPtr(A_double, A_Real, A_double.size());
toRealArrPtr(B_double, B_Real, B_double.size());
toRealArrPtr(C_double, C_Real, C_double.size());
}
void dgemv_(
const char* trans,
const int* m,
const int* n,
const ::yade::Real* alpha_Real,
::yade::Real* A_Real,
const int* lda,
::yade::Real* x_Real,
const int* incx,
const ::yade::Real* beta_Real,
::yade::Real* y_Real,
const int* incy)
{
// DGEMV performs one of the matrix-vector operations - FIXME use Eigen instead.
if (*incx != 1) throw std::runtime_error("dgemv_ wrapper: incx should be 1");
if (*incy != 1) throw std::runtime_error("dgemv_ wrapper: incy should be 1");
const double alpha_double = static_cast<double>(*alpha_Real);
const double beta_double = static_cast<double>(*beta_Real);
bool blasN = false;
if (((*trans) == 'n') or ((*trans) == 'N')) { blasN = true; }
std::vector<double> A_double = toDoubleVec(A_Real, (*m) * (*n));
std::vector<double> x_double = toDoubleVec(x_Real, blasN ? (*n) : (*m));
std::vector<double> y_double = toDoubleVec(y_Real, blasN ? (*m) : (*n));
dgemv_(trans, m, n, &alpha_double, A_double.data(), lda, x_double.data(), incx, &beta_double, y_double.data(), incy);
toRealArrPtr(A_double, A_Real, A_double.size());
toRealArrPtr(x_double, x_Real, x_double.size());
toRealArrPtr(y_double, y_Real, y_double.size());
}
void dcopy_(const int* N, ::yade::Real* from_x, const int* incx, ::yade::Real* to_y, const int* incy)
{
// DCOPY copies a vector, x, to a vector, y.
if (*incx != 1) throw std::runtime_error("dcopy_ wrapper: incx should be 1");
if (*incy != 1) throw std::runtime_error("dcopy_ wrapper: incy should be 1");
// FIXME: that should be just a simple assignment
// Y=X;
// with normal VectorXr
for (int i = 0; i < *N; ++i) {
to_y[i] = from_x[i];
}
}
::yade::VectorXr toVectorXr(const ::yade::Real* from, int size)
{
::yade::VectorXr ret(size);
for (int i = 0; i < size; i++) {
ret[i] = from[i];
}
return ret;
}
::yade::Real ddot_(const int* N, const ::yade::Real* x, const int* incx, const ::yade::Real* y, const int* incy)
{
// DDOT forms the dot product of two vectors.
if (*incx != 1) throw std::runtime_error("ddot_ wrapper: incx should be 1");
if (*incy != 1) throw std::runtime_error("ddot_ wrapper: incy should be 1");
::yade::VectorXr xx = toVectorXr(x, *N);
::yade::VectorXr yy = toVectorXr(y, *N);
return xx.dot(yy);
}
void daxpy_(const int* N, const ::yade::Real* da_Real, ::yade::Real* dx_Real, const int* incx, ::yade::Real* dy_Real, const int* incy)
{
if (*incx != 1) throw std::runtime_error("daxpy_ wrapper: incx should be 1");
if (*incy != 1) throw std::runtime_error("daxpy_ wrapper: incy should be 1");
const double da_double = static_cast<double>(*da_Real);
std::vector<double> dx_double = toDoubleVec(dx_Real, *N);
std::vector<double> dy_double = toDoubleVec(dy_Real, *N);
daxpy_(N, &da_double, dx_double.data(), incx, dy_double.data(), incy);
toRealArrPtr(dx_double, dx_Real, dx_double.size());
toRealArrPtr(dy_double, dy_Real, dy_double.size());
}
void dscal_(const int* N, const ::yade::Real* alpha_Real, ::yade::Real* x_Real, const int* incx)
{
if (*incx != 1) throw std::runtime_error("dscal_ wrapper: incx should be 1");
// FIXME - this should be just an Eigen multiplication by scalar, like: blasQAs *= scaleFactor; // it unrolls loops, and does everything you want.
for (int i = 0; i < *N; i++) {
x_Real[i] *= *alpha_Real;
}
}
void dsyev_(
const char* jobz,
const char* uplo,
const int* N,
::yade::Real* A_Real,
const int* lda,
::yade::Real* W_Real,
::yade::Real* work_Real,
int* lwork,
int* info)
{
// DSYEV computes the eigenvalues and, optionally, the left and/or right eigenvectors for SY matrices
std::vector<double> A_double = toDoubleVec(A_Real, (*lda) * (*N));
std::vector<double> W_double = toDoubleVec(W_Real, *N);
std::vector<double> work_double = toDoubleVec(work_Real, *lwork);
dsyev_(jobz, uplo, N, A_double.data(), lda, W_double.data(), work_double.data(), lwork, info);
toRealArrPtr(A_double, A_Real, A_double.size());
toRealArrPtr(W_double, W_Real, W_double.size());
toRealArrPtr(work_double, work_Real, work_double.size());
}
#endif
#endif
|
