Skip to content
GitLab
Commit b9817f75 authored by ralfh's avatar ralfh
Browse files

Reformatting of codes

parent 023c6ad2
Hide whitespace changes
Inline Side-by-side
Assignments/Codes/MatVec/ArrowMatrix/solutions/arrowmatvec.cpp
View file @ b9817f75
#include <iostream>
#include <iomanip>
#include <iostream>
#include <Eigen/Dense>
......@@ -17,31 +17,27 @@ using namespace Eigen;
/* SAM_LISTING_BEGIN_0 */
void arrow_matrix_2_times_x(const VectorXd &d, const VectorXd &a,
const VectorXd &x, VectorXd &y) {
assert(d.size() == a.size() && a.size() == x.size() &&
"Vector size must be the same!");
int n = d.size();
// In this lines, we extract the blocks used to construct the matrix A.
VectorXd d_head = d.head(n-1);
VectorXd a_head = a.head(n-1);
MatrixXd d_diag = d_head.asDiagonal();
MatrixXd A(n,n);
// We build the matrix A using the "comma initialization": each expression separated
// by a comma is a "block" of the matrix we are building.
// d\_diag is the top left (n-1)x(n-1) block
// a\_head is the top right vertical vector
// a\_head.transpose() is the bottom left horizontal vector
// d(n-1) is a single element (a 1x1 matrix), on the bottom right corner
// This is how the matrix looks like:
// A = | D | a |
// |-----+--------|
// | a\^T | d(n-1) |
A << d_diag, a_head,
a_head.transpose(), d(n-1);
y = A*A*x;
assert(d.size() == a.size() && a.size() == x.size() &&
"Vector size must be the same!");
int n = d.size();
// In this lines, we extract the blocks used to construct the matrix A.
VectorXd d_head = d.head(n - 1);
VectorXd a_head = a.head(n - 1);
MatrixXd d_diag = d_head.asDiagonal();
MatrixXd A(n, n);
// We build the matrix A using the "comma initialization": each expression
// separated by a comma is a "block" of the matrix we are building. d\_diag is
// the top left (n-1)x(n-1) block a\_head is the top right vertical vector
// a\_head.transpose() is the bottom left horizontal vector
// d(n-1) is a single element (a 1x1 matrix), on the bottom right corner
// This is how the matrix looks like:
// A = | D | a |
// |-----+--------|
// | a\^T | d(n-1) |
A << d_diag, a_head, a_head.transpose(), d(n - 1);
y = A * A * x;
}
/* SAM_LISTING_END_0 */
......@@ -53,52 +49,47 @@ void arrow_matrix_2_times_x(const VectorXd &d, const VectorXd &a,
* @param[out] y The vector y = A*A*x
*/
/* SAM_LISTING_BEGIN_1 */
void efficient_arrow_matrix_2_times_x(const VectorXd &d,
const VectorXd &a,
const VectorXd &x,
VectorXd &y) {
assert(d.size() == a.size() && a.size() == x.size() &&
void efficient_arrow_matrix_2_times_x(const VectorXd &d, const VectorXd &a,
const VectorXd &x, VectorXd &y) {
assert(d.size() == a.size() && a.size() == x.size() &&
"Vector size must be the same!");
int n = d.size();
// Notice that we can compute (A*A)*x more efficiently using
// A*(A*x). This is, in fact, performing two matrix vector
// multiplications
// instead of a more expensive matrix-matrix multiplication.
// Therefore, as first step, we need a way to efficiently
// compute A*x
// This function computes A*x. you can use it
// by calling A\_times\_x(x).
// This is the syntax for lambda functions: notice the extra
// [variables] code. Each variable written within [] brackets
// will be captured (i.e. seen) inside the lambda function.
// Without \&, a copy of the variable will be performed.
// Notice that A = D + H + V, s.t. A*x = D*x + H*x + V*x
// D*x can be rewritten as d*x componentwise
// H*x is zero, except at the last component
// V*x is only affected by the last component of x
auto A_times_x = [&a, &d, n] (const VectorXd & x) {
// This takes care of the diagonal (D*x)
// Notice: we use d.array() to tell Eigen to treat
// a vector as an array. As a result: each operation
// is performed componentwise.
VectorXd Ax = ( d.array() * x.array() ).matrix();
// H*x only affects the last component of A*x
// This is a dot product between a and x with the last
// component removed
Ax(n-1) += a.head(n - 1).dot(x.head(n-1));
// V*x is equal to the vector
// (a(0)*x(n-1), ..., a(n-2)*x(n-1), 0)
Ax.head(n-1) += x(n-1) * a.head(n-1);
return Ax;
};
// <=> y = A*(A*x)
y = A_times_x(A_times_x(x));
int n = d.size();
// Notice that we can compute (A*A)*x more efficiently using
// A*(A*x). This is, in fact, performing two matrix vector
// multiplications
// instead of a more expensive matrix-matrix multiplication.
// Therefore, as first step, we need a way to efficiently
// compute A*x
// This function computes A*x. you can use it
// by calling A\_times\_x(x).
// This is the syntax for lambda functions: notice the extra
// [variables] code. Each variable written within [] brackets
// will be captured (i.e. seen) inside the lambda function.
// Without \&, a copy of the variable will be performed.
// Notice that A = D + H + V, s.t. A*x = D*x + H*x + V*x
// D*x can be rewritten as d*x componentwise
// H*x is zero, except at the last component
// V*x is only affected by the last component of x
auto A_times_x = [&a, &d, n](const VectorXd &x) {
// This takes care of the diagonal (D*x)
// Notice: we use d.array() to tell Eigen to treat
// a vector as an array. As a result: each operation
// is performed componentwise.
VectorXd Ax = (d.array() * x.array()).matrix();
// H*x only affects the last component of A*x
// This is a dot product between a and x with the last
// component removed
Ax(n - 1) += a.head(n - 1).dot(x.head(n - 1));
// V*x is equal to the vector
// (a(0)*x(n-1), ..., a(n-2)*x(n-1), 0)
Ax.head(n - 1) += x(n - 1) * a.head(n - 1);
return Ax;
};
// <=> y = A*(A*x)
y = A_times_x(A_times_x(x));
}
/* SAM_LISTING_END_1 */
......@@ -106,48 +97,41 @@ void efficient_arrow_matrix_2_times_x(const VectorXd &d,
* Repeat tests 10 times, and output the minimal runtime
* amongst all times. Test both the inefficient and the efficient
* versions.
*/
*/
void runtime_arrow_matrix() {
/* SAM_LISTING_BEGIN_3 */
std::cout << std::setw(8) << "n"
<< std::setw(15) << "original"
<< std::setw(15) << "efficient"
<< std::endl;
for (unsigned n = 4; n <= 2048; n *= 2) {
// Number of repetitions
unsigned int repeats = 10;
Timer timer, timer_eff;
// Repeat test many times
for (unsigned int r = 0; r < repeats; ++r) {
// Create test input using random vectors
Eigen::VectorXd a = Eigen::VectorXd::Random(n),
d = Eigen::VectorXd::Random(n),
x = Eigen::VectorXd::Random(n),
y;
// Compute times
timer.start();
arrow_matrix_2_times_x(d, a, x, y);
timer.stop();
// Compute times for efficient implementation
timer_eff.start();
efficient_arrow_matrix_2_times_x(d, a, x, y);
timer_eff.stop();
}
// Print results
std::cout << std::setw(8) << n
<< std::scientific << std::setprecision(3)
<< std::setw(15) << timer.min()
<< std::setw(15) << timer_eff.min()
<< std::endl;
/* SAM_LISTING_BEGIN_3 */
std::cout << std::setw(8) << "n" << std::setw(15) << "original"
<< std::setw(15) << "efficient" << std::endl;
for (unsigned n = 4; n <= 2048; n *= 2) {
// Number of repetitions
unsigned int repeats = 10;
Timer timer, timer_eff;
// Repeat test many times
for (unsigned int r = 0; r < repeats; ++r) {
// Create test input using random vectors
Eigen::VectorXd a = Eigen::VectorXd::Random(n),
d = Eigen::VectorXd::Random(n),
x = Eigen::VectorXd::Random(n), y;
// Compute times
timer.start();
arrow_matrix_2_times_x(d, a, x, y);
timer.stop();
// Compute times for efficient implementation
timer_eff.start();
efficient_arrow_matrix_2_times_x(d, a, x, y);
timer_eff.stop();
}
/* SAM_LISTING_END_3 */
// Print results
std::cout << std::setw(8) << n << std::scientific << std::setprecision(3)
<< std::setw(15) << timer.min() << std::setw(15)
<< timer_eff.min() << std::endl;
}
/* SAM_LISTING_END_3 */
}
// Rename long variable name to duration_t (easy to change)
......@@ -161,29 +145,28 @@ using duration_t = std::chrono::nanoseconds;
* \param[in] repeats Number of repetitions.
*/
template <class Function>
duration_t timing(const Function & F, int repeats = 10) {
// Shortcut for time_point
using time_point_t = std::chrono::high_resolution_clock::time_point;
duration_t timing(const Function &F, int repeats = 10) {
// Shortcut for time_point
using time_point_t = std::chrono::high_resolution_clock::time_point;
// Loop many times
duration_t min_elapsed;
for(int r = 0; r < repeats; r++) {
// Start clock (MATLAB: tic)
time_point_t start = std::chrono::high_resolution_clock::now();
// Loop many times
duration_t min_elapsed;
for (int r = 0; r < repeats; r++) {
// Start clock (MATLAB: tic)
time_point_t start = std::chrono::high_resolution_clock::now();
// Run function
F();
// Run function
F();
// Stop clock (MATLAB: toc) and measure difference
duration_t elapsed = std::chrono::duration_cast<duration_t>(
std::chrono::high_resolution_clock::now() - start);
// Stop clock (MATLAB: toc) and measure difference
duration_t elapsed = std::chrono::duration_cast<duration_t>(
std::chrono::high_resolution_clock::now() - start);
// Compute min between all runs
min_elapsed = r == 0 ? elapsed : std::min(elapsed, min_elapsed);
}
return min_elapsed;
// Compute min between all runs
min_elapsed = r == 0 ? elapsed : std::min(elapsed, min_elapsed);
}
return min_elapsed;
}
/* \brief Compute timing using chrono
......@@ -191,72 +174,67 @@ duration_t timing(const Function & F, int repeats = 10) {
*/
void runtime_arrow_matrix_with_chrono() {
// Table header
std::cout << std::setw(8) << "n"
<< std::scientific << std::setprecision(3)
<< std::setw(15) << "original"
<< std::setw(15) << "efficient"
// Table header
std::cout << std::setw(8) << "n" << std::scientific << std::setprecision(3)
<< std::setw(15) << "original" << std::setw(15) << "efficient"
<< std::endl;
// Run from $2^5$ to $2^11$ with powers of two
for (unsigned int n = (1 << 5); n < (1 << 12); n = n << 1) {
// Create random vectors
VectorXd d = VectorXd::Random(n);
VectorXd a = VectorXd::Random(n);
VectorXd x = VectorXd::Random(n);
VectorXd y(n);
// Call "timing", using a lambda function for F
// Remember: we cannot pass arrow\_matrix\_2\_times\_x directly to timing
// the timing function expects a n object with operator()(void)
duration_t elapsed =
timing([&a, &d, &x, &y]() { arrow_matrix_2_times_x(d, a, x, y); }, 10);
// Call "timing", using a lambda function for F
duration_t elapsed_efficient = timing(
[&a, &d, &x, &y]() { efficient_arrow_matrix_2_times_x(d, a, x, y); },
10);
// Output timings
std::cout << std::setw(8) << n << std::scientific << std::setprecision(3)
<< std::setw(15) << elapsed.count() * 1e-9 // ns to s
<< std::setw(15) << elapsed_efficient.count() * 1e-9 // ns to s
<< std::endl;
// Run from $2^5$ to $2^11$ with powers of two
for(unsigned int n = (1 << 5); n < (1 << 12); n = n << 1) {
// Create random vectors
VectorXd d = VectorXd::Random(n);
VectorXd a = VectorXd::Random(n);
VectorXd x = VectorXd::Random(n);
VectorXd y(n);
// Call "timing", using a lambda function for F
// Remember: we cannot pass arrow\_matrix\_2\_times\_x directly to timing
// the timing function expects a n object with operator()(void)
duration_t elapsed = timing([&a, &d, &x, &y] () {
arrow_matrix_2_times_x(d, a, x, y);
}, 10);
// Call "timing", using a lambda function for F
duration_t elapsed_efficient = timing([&a, &d, &x, &y] () {
efficient_arrow_matrix_2_times_x(d, a, x, y);
}, 10);
// Output timings
std::cout << std::setw(8)<< n
<< std::scientific << std::setprecision(3)
<< std::setw(15) << elapsed.count() * 1e-9 // ns to s
<< std::setw(15) << elapsed_efficient.count() * 1e-9 // ns to s
<< std::endl;
}
}
}
int main(void) {
// Test vectors
VectorXd a(5);
a << 1., 2., 3., 4., 5.;
VectorXd d(5);
d <<1., 3., 4., 5., 6.;
VectorXd x(5);
x << -5., 4., 6., -8., 5.;
VectorXd yi;
// Run both functions
arrow_matrix_2_times_x(a,d,x,yi);
VectorXd ye(yi.size());
efficient_arrow_matrix_2_times_x(a,d,x,ye);
// Compute error
double err = (yi - ye).norm();
// Output error
std::cout << "--> Correctness test." << std::endl;
std::cout << "Error: " << err << std::endl;
// Print out runtime
std::cout << "--> Runtime test." << std::endl;
runtime_arrow_matrix();
// Print out runtime with chrono
std::cout << "--> Runtime test. with chrono." << std::endl;
runtime_arrow_matrix_with_chrono();
// Final test: exit with error if error is too big
double eps = std::numeric_limits<double>::denorm_min();
exit(err < eps);
// Test vectors
VectorXd a(5);
a << 1., 2., 3., 4., 5.;
VectorXd d(5);
d << 1., 3., 4., 5., 6.;
VectorXd x(5);
x << -5., 4., 6., -8., 5.;
VectorXd yi;
// Run both functions
arrow_matrix_2_times_x(a, d, x, yi);
VectorXd ye(yi.size());
efficient_arrow_matrix_2_times_x(a, d, x, ye);
// Compute error
double err = (yi - ye).norm();
// Output error
std::cout << "--> Correctness test." << std::endl;
std::cout << "Error: " << err << std::endl;
// Print out runtime
std::cout << "--> Runtime test." << std::endl;
runtime_arrow_matrix();
// Print out runtime with chrono
std::cout << "--> Runtime test. with chrono." << std::endl;
runtime_arrow_matrix_with_chrono();
// Final test: exit with error if error is too big
double eps = std::numeric_limits<double>::denorm_min();
exit(err < eps);
}
Assignments/Codes/MatVec/ArrowMatrix/templates/arrowmatvec.cpp
View file @ b9817f75
#include <iostream>
#include <iomanip>
#include <iostream>
#include <Eigen/Dense>
......@@ -17,20 +17,15 @@ using namespace Eigen;
/* SAM_LISTING_BEGIN_0 */
void arrow_matrix_2_times_x(const VectorXd &d, const VectorXd &a,
const VectorXd &x, VectorXd &y) {
assert(d.size() == a.size() && a.size() == x.size() &&
"Vector size must be the same!");
int n = d.size();
VectorXd d_head = d.head(n-1);
VectorXd a_head = a.head(n-1);
MatrixXd d_diag = d_head.asDiagonal();
MatrixXd A(n,n);
A << d_diag, a_head,
a_head.transpose(), d(n-1);
y = A*A*x;
assert(d.size() == a.size() && a.size() == x.size() &&
"Vector size must be the same!");
int n = d.size();
VectorXd d_head = d.head(n - 1);
VectorXd a_head = a.head(n - 1);
MatrixXd d_diag = d_head.asDiagonal();
MatrixXd A(n, n);
A << d_diag, a_head, a_head.transpose(), d(n - 1);
y = A * A * x;
}
/* SAM_LISTING_END_0 */
......@@ -42,15 +37,13 @@ void arrow_matrix_2_times_x(const VectorXd &d, const VectorXd &a,
* @param[out] y The vector y = A*A*x
*/
/* SAM_LISTING_BEGIN_1 */
void efficient_arrow_matrix_2_times_x(const VectorXd &d,
const VectorXd &a,
const VectorXd &x,
VectorXd &y) {
assert(d.size() == a.size() && a.size() == x.size() &&
void efficient_arrow_matrix_2_times_x(const VectorXd &d, const VectorXd &a,
const VectorXd &x, VectorXd &y) {
assert(d.size() == a.size() && a.size() == x.size() &&
"Vector size must be the same!");
int n = d.size();
int n = d.size();
// TODO: Implement an efficient version of arrow\_matrix\_2\_times\_x
// TODO: Implement an efficient version of arrow\_matrix\_2\_times\_x
}
/* SAM_LISTING_END_1 */
......@@ -58,40 +51,38 @@ void efficient_arrow_matrix_2_times_x(const VectorXd &d,
* Repeat tests 10 times, and output the minimal runtime
* amongst all times. Test both the inefficient and the efficient
* versions.
*/
*/
void runtime_arrow_matrix() {
// TODO: your code here, time the codes
// TODO: your code here, time the codes
}
int main(void) {
// Test vectors
VectorXd a(5);
a << 1., 2., 3., 4., 5.;
VectorXd d(5);
d <<1., 3., 4., 5., 6.;
VectorXd x(5);
x << -5., 4., 6., -8., 5.;
VectorXd yi;
// Run both functions
arrow_matrix_2_times_x(a,d,x,yi);
VectorXd ye(yi.size());
efficient_arrow_matrix_2_times_x(a,d,x,ye);
// Compute error
double err = (yi - ye).norm();
// Output error
std::cout << "--> Correctness test." << std::endl;
std::cout << "Error: " << err << std::endl;
// Print out runtime
std::cout << "--> Runtime test." << std::endl;
runtime_arrow_matrix();
// Final test: exit with error if error is too big
double eps = std::numeric_limits<double>::denorm_min();
exit(err < eps);
// Test vectors
VectorXd a(5);
a << 1., 2., 3., 4., 5.;
VectorXd d(5);
d << 1., 3., 4., 5., 6.;
VectorXd x(5);
x << -5., 4., 6., -8., 5.;
VectorXd yi;
// Run both functions
arrow_matrix_2_times_x(a, d, x, yi);
VectorXd ye(yi.size());
efficient_arrow_matrix_2_times_x(a, d, x, ye);
// Compute error
double err = (yi - ye).norm();
// Output error
std::cout << "--> Correctness test." << std::endl;
std::cout << "Error: " << err << std::endl;
// Print out runtime
std::cout << "--> Runtime test." << std::endl;
runtime_arrow_matrix();
// Final test: exit with error if error is too big
double eps = std::numeric_limits<double>::denorm_min();
exit(err < eps);
}
Assignments/Codes/MatVec/GramSchmidt/solutions/gramschmidt.cpp
View file @ b9817f75
......@@ -11,55 +11,46 @@ using namespace Eigen;
* \return Matrix with ONB of $span(a_1, \cdots, a_n)$ as columns
*/
/* SAM_LISTING_BEGIN_1 */
MatrixXd gram_schmidt(const MatrixXd & A) {
// We create a matrix Q with the same size and data of A
MatrixXd Q(A);
// The first vector just gets normalized
Q.col(0).normalize();
// Iterate over all other columns of A
for(unsigned int j = 1; j < A.cols(); ++j) {
// See eigen documentation for usage of col and leftCols
Q.col(j) -= Q.leftCols(j) * (Q.leftCols(j).transpose() * A.col(j));
// Normalize vector, if possible
// (otherwise it means columns of $\mathbf{A}$ are
// almost linear dependant)
double eps = std::numeric_limits<double>::denorm_min();
if( Q.col(j).norm() <= eps * A.col(j).norm() ) {
std::cerr << "Gram-Schmidt failed because "
<< "A has (almost) linear dependant "
<< "columns. Bye." << std::endl;
break;
} else {
Q.col(j).normalize();
}
MatrixXd gram_schmidt(const MatrixXd &A) {
// We create a matrix Q with the same size and data of A
MatrixXd Q(A);
// The first vector just gets normalized
Q.col(0).normalize();
// Iterate over all other columns of A
for (unsigned int j = 1; j < A.cols(); ++j) {
// See eigen documentation for usage of col and leftCols
Q.col(j) -= Q.leftCols(j) * (Q.leftCols(j).transpose() * A.col(j));
// Normalize vector, if possible
// (otherwise it means columns of $\mathbf{A}$ are
// almost linearly dependant)
double eps = std::numeric_limits<double>::denorm_min();