Skip to content

nml is a "simple" matrix/numerical analysis library written in pure C. The scope of the library is to highlight various algorithm implementations related to matrices. Code readability was a major concern.

License

Notifications You must be signed in to change notification settings

santtyan/neat-matrix-library

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

 .----------------.  .----------------.  .----------------. 
| .--------------. || .--------------. || .--------------. |
| | ____  _____  | || | ____    ____ | || |   _____      | |
| ||_   \|_   _| | || ||_   \  /   _|| || |  |_   _|     | |
| |  |   \ | |   | || |  |   \/   |  | || |    | |       | |
| |  | |\ \| |   | || |  | |\  /| |  | || |    | |   _   | |
| | _| |_\   |_  | || | _| |_\/_| |_ | || |   _| |__/ |  | |
| ||_____|\____| | || ||_____||_____|| || |  |________|  | |
| |              | || |              | || |              | |
| '--------------' || '--------------' || '--------------' |      Neat Matrix Library
 '----------------'  '----------------'  '----------------'       

nml is a simple matrix and linear algebra library written in standard C.

Code should be portable and there are no dependencies.

For a detailed explanation of the library code please check this blog post.

It currently supports:

The library is still under development, but a few thousands test cases are already implemented, covering the most complex algorithms (REF, RREF, LUP, QR, DET, INV, BACKWARD SUBSTITION, FORWARD SUBSTITION, etc.)

Table of Contents

Compile / Run Examples

The build file for the library it's called nml.sh. It's actually a bash script (not a makefile!).

Building the library

./nml.sh clean build

This will compile the library, create a dist folder where you will find *.a static library file and the header files.

gcc and ar should be available in $PATH.

If you want to use the clang compiler instead of gcc you need to manually edit the ./nml.sh file, changing the variable CC from gcc to clang. Nothing else should be changed.

# COMPILING RELATED
CC=clang #<----------------- here
CCFLAGS="-Wall -c"
CCFLAGS_EXAMPLES="-Wall"

Building the examples

Examples can be found in the ./examples folder.

To build the code examples:

./nml.sh clean examples
  1. This will create an examples/lib folder where the libnml.a and the header files will be copied;
  2. The examples/*.c will be compiled with the latest version of libnml;
  3. For each examples/*.c an executable (*.ex) will be created.

To run an example:

# ./nml.sh clean examples && ./examples/<example name>.ex
./nml.sh clean examples && ./examples/playground.ex

Running the tests

To run the tests

./nml.sh clean test
  1. This will create a test/lib folder where the libnml.a and the header files will be copied;
  2. Each test tests/*.c will be compiled with the latest version of libnml;
  3. For each test tests/*/c an executable (*.ex) will be created.

The test data was generated using sympy. In the tests/generators/ folder you can find the python3 (.py) scripts used to generate the data.

Cleaning

./nml.sh clean

This will clean everything (*.o,*.ex,*.a) and will leave the library folder in a clean state.

How to use the library

A few examples can be found in the ./examples folder folder.

Creating matrices

All the methods are interacting with the nml_mat struct:

typedef struct nml_mat_s {
  unsigned int num_rows;
  unsigned int num_cols;
  double **data;
  int is_square;
} nml_mat;

To interact the elements of the matrix:

nml_mat *m = ...
m->data[i][j] = ...

Creating a new Matrix

The methods for a creating a new matrix are:

  • nml_mat *nml_mat_new(unsigned int num_rows, unsigned int num_cols)
    • Creates a num_rows * num_cols matrix of zeroes.
  • nml_mat *nml_mat_sqr(unsigned int size)
    • Creates a square size * size matrix of zeroes.
  • nml_mat *nml_mat_eye(unsigned int size)
    • Creates an identity size * size matrix.
  • nml_mat *nml_mat_cp(nml_mat *m)
    • Returns a new identitcal copy of matrix m.

Everytime we create a matrix, we dynamically allocate memory. To free the memory please use: nml_mat_free(nml_mat *m).

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {


  nml_mat* m, *mx;

  printf("\nCreating an empty matrix with 2x3\n");
  m = nml_mat_new(2,3);
  nml_mat_print(m);
  nml_mat_free(m);

  printf("\nCreating a square matrix 5x5 \n");
  m = nml_mat_sqr(5);
  nml_mat_print(m);
  nml_mat_free(m);

  printf("\nCreating an ID 7x7 Matrix and copying it into another matrix:\n");
  m = nml_mat_eye(7);
  mx = nml_mat_cp(m);
  nml_mat_print(m);
  nml_mat_print(mx);
  nml_mat_free(m);
  nml_mat_free(mx);

  return 0;
}

To run the example:

./nml.sh clean examples && examples/creating_a_matrix.ex

Creating a marray from an array (double[N])

An array can be used as the "data source" for the Matrix by using:

  • nml_mat *nml_mat_from(unsigned int num_rows, unsigned int num_cols, unsigned int n_vals, double *vals)
    • num_rows and num_cols represent the dimensions of the matrix;
    • n_vals how many values to read from the vals source. If n_vals is smaller than the product num_cols * num_rows, 0.0 will be used as the default value;
    • vals the array containing double values.
#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) { 
    double array[6] = { 
        1.0, 0.2, 3.0, 4.0, 5.0, 3.1 
    };
    nml_mat* my;
    
    // 3 rows, 2 columns
    // read exactly 6 numbers from array[6]
    my = nml_mat_from(3, 2, 6, array);
    nml_mat_print(my);
    nml_mat_free(my);

    // 4 rows, 2 columns
    // read exactly 3 numbers from array[6]
    my = nml_mat_from(4, 2, 3, array);
    nml_mat_print(my);
    nml_mat_free(my);

    return 0;
}

To run the example:

./nml.sh clean examples && examples/creating_a_matrix_from_an_array.ex

Creating a Matrix from an external file

The two methods that can be used to create a matrix from a file on disk are:

  • nml_mat *nml_mat_fromfile(const char *file)
    • Create a matrix from the file path. If the file cannot be opened a NULL matrix will be returned.
  • nml_mat *nml_mat_fromfilef(FILE *f)
    • Creates a matrix from am already opened stream f. Does not automatically close the stream (FILE).

In the file, the matrix has the following format:

4 5
0.0     1.0     2.0     5.0     3.0
3.0     8.0     9.0     1.0     4.0
2.0     3.0     7.0     1.0     1.0
0.0     0.0     4.0     3.0     8.0

On the first line 4 represents the number of rows and 5 represents the number of columns of the Matrix Then next lines contain the matrix elements: 4 * 5 = 20 numbers.

Example code:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    const char *f = "examples/data/matrix1.data";
    nml_mat *from_file = nml_mat_fromfile(f);
    nml_mat_print(from_file);
    nml_mat_free(from_file);

    // Or if the file is already opened

    FILE *m_file = fopen("examples/data/matrix2.data", "r");
    nml_mat *from_file2 = nml_mat_fromfilef(m_file);
    nml_mat_print(from_file2);
    nml_mat_free(from_file2);
    fclose(m_file);

    return 0;
}

To run the example:

./nml.sh clean examples && ./examples/creating_a_matrix_from_file.ex

Creating a matrix from user input

The nml_mat *nml_mat_fromfilef(FILE *f) can be called, with f=stdin.

Code example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *from_file2 = nml_mat_fromfilef(stdin);
    nml_mat_print(from_file2);
    nml_mat_free(from_file2);
    return 0;
}

To run the example:

./nml.sh clean examples && examples/creating_a_matrix_from_user_input.ex

Creating randomized matrices

Creating a randomized matrix can be done with the following two methods:

  • nml_mat *nml_mat_rnd(unsigned int num_rows, unsigned int num_cols, double min, double max)
    • Creates a randomized matrix of size num_rows * num_cols;
    • The random values are between min and max;
  • nml_mat *nml_mat_sqr_rnd(unsigned int size, double min, double max)
    • Creates a randomized matrix of size size * size;
    • The random values are between min and max;
#include <stdlib.h>
#include <stdio.h>
#include <time.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
  srand(time(NULL)); // Should be called once per program
  nml_mat *m = nml_mat_rnd(5, 5, -10.0, 10.0);
  nml_mat_print(m);
  nml_mat_free(m);
  return 0;
}

To run the example:

./nml.sh clean examples && examples/create_randomized_matrix.ex

Check if two matrices are equal

There are two "equality" methods for matrices:

  • int nml_mat_eqdim(nml_mat *m1, nml_mat *m2)

    • Tests if two matrices have the same dimension.
  • int nml_mat_eq(nml_mat *m1, nml_mat *m2, double tolerance)

    • Test if two matrices are equal:
      • They have the same dimensions
      • The elements are equal or close of being equal.
    • If you want the elements to be "exactly" eqaul, tolerance=0.0
#include <stdlib.h>
#include <stdio.h>
#include <time.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {

    srand(time(NULL));
    nml_mat *m1 = nml_mat_rnd(2, 3, 1.0, 10.0);
    nml_mat *m2 = nml_mat_rnd(2, 3, 1.0, 10.0);

    if (nml_mat_eq(m1, m2, 0.001)) {
        printf("Wow, what were the oddss..\n");
    } else {
        printf("It's ok, nobody is that lucky!\n");
    }
    if (nml_mat_eqdim(m1, m2)) {
        printf("At least we know they both have the same number of rows and columns.\n");
    }

    nml_mat_free(m1);
    nml_mat_free(m2);
    return 0;
}

To run the example:

./nml.sh clean examples && ./examples/matrix_equality.ex

Accesing and modifying matrix elements

Select rows and columns

Two methods can be used to select rows and columns from a source matrix (nml_mat*):

  • nml_mat *nml_mat_col_get(nml_mat *m, unsigned int col)
  • nml_mat *nml_mat_row_get(nml_mat *m, unsigned int row)

The following code extracts every column of a given random matrix into a temporary column matrix (nml_mat*):

#include <stdlib.h>
#include <stdio.h>
#include <time.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
  printf("\nExtract all matrix columns from a Matrix as matrices\n");
  srand(time(NULL));
  nml_mat *m = nml_mat_rnd(5, 5, -10.0, 10.0);
  nml_mat *col;
  nml_mat_print(m);
  int i = 0;
  for(i = 0; i < m->num_cols; i++) {
    col = nml_mat_col_get(m, i);
    nml_mat_print(col);
    nml_mat_free(col);
  }
  nml_mat_free(m);
  return 0;
}

To run the example:

./nml.sh clean examples && examples/select_columns.ex

Set all elements to a value

Use: void nml_mat_all_set(nml_mat *matrix, double value)

#include <stdlib.h>
#include <stdio.h>
#include <math.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    // Creates a matrix of zeros of size = 5
    nml_mat *pi_mat = nml_mat_sqr(5);

    // Sets all elements to PI
    nml_mat_all_set(pi_mat, M_PI);

    nml_mat_print(pi_mat);
    nml_mat_free(pi_mat);
    return 0;
}

To run the example:

./nml.sh clean examples && ./examples/set_all_elements.ex

Set the first diagonal to a value

Use: int nml_mat_diag_set(nml_mat *matrix, double value)

#include <stdlib.h>
#include <stdio.h>
#include <math.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    // Creates a matrix of zeros of size = 5
    nml_mat *pi_mat = nml_mat_sqr(5);

    // Sets the first diagonal to PI
    nml_mat_diag_set(pi_mat, M_PI);

    nml_mat_print(pi_mat);
    nml_mat_free(pi_mat);
    return 0;
}

To run the example:

./nml.sh clean examples && examples/set_diagonal_elements.ex

Scalar multiply the matrix

Use:

  • nml_mat *nml_mat_smult(nml_mat *m, double num)
    • Multiplies all elements of matrix m with num. A new matrix is returned.
  • int nml_mat_smult_r(nml_mat *m, double num)
    • Multiplies all elements of matrix m with num. All changes are done on matrix m.
#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *m = nml_mat_eye(5);

    // Multiply all elements of m with 2.0
    // and return a new matrix

    nml_mat *new_m = nml_mat_smult(m, 2.0);

    if (!(nml_mat_eq(m, new_m, 0.0))) {
        printf("It's normal to see this message.\n");
    }

    // Multiply all elements of m with 2.0
    // m is modified, no new matrix is created
    nml_mat_smult_r(m, 2.0);

    if (nml_mat_eq(m, new_m, 0.0)) {
        printf("It's even more normal to see this message.\n");
    }

    nml_mat_free(m);
    nml_mat_free(new_m);
    
    return 0;
}

To run the example:

./nml.sh clean examples && examples/scalar_multiply.ex

Multiply rows

Use:

  • nml_mat *nml_mat_row_mult(nml_mat *m, unsigned int row, double num)
    • Multiplies all elements from row row in matrix m with scalar num. A new matrix is returned. m remains un-altered.
  • int nml_mat_row_mult_r(nml_mat *m, unsigned int row, double num)
    • Multiplies all elements from row row in matrix m with scalar num. The changes are done directly on matrix m.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *a = nml_mat_new(4,5);
    nml_mat_all_set(a, 1.0);
    int i = 0;
    for(; i < a->num_rows; ++i) {
        // Changes are doing on matrix a
        // row[i] is multiplied with (double) i
        nml_mat_row_mult_r(a, i, (double)i);
    }
    nml_mat_print(a);

    // Create a new matrix b by multiplying row[1] 
    // in matrix a with 5.0.
    // Matrix a remains unchanged
    nml_mat *b = nml_mat_row_mult(a, 1, 5.0);
    nml_mat_print(b);
    nml_mat_free(a);
    nml_mat_free(b);
    return 0;
}

To run the example:

./nml.sh clean examples && examples/multiply_rows.ex

To run the example:

./nml.sh clean examples && examples/multiply_rows.ex

Add rows

The following methods are used to add a row to another row (with a multiplicator). This method is usally used when implementing various forms of matrix reduction or decompositions.

Use:

  • nml_mat *nml_mat_row_addrow(nml_mat *m, unsigned int where, unsigned int row, double multiplier)
    • This will do the following: m->data[where][...] *= m->data[row][...] * multiplier. The results will be kept in a new matrix. Matrix m remains unchanged.
  • int nml_mat_row_addrow_r(nml_mat *m, unsigned int where, unsigned int row, double multiplier)
    • This will do the following: m->data[where][...] *= m->data[row][...] * multiplier. The changes are done directly on m.
#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {

    nml_mat *m = nml_mat_rnd(5, 4, 1.0, 2.0);
    nml_mat_print(m);

    // Add row[1] elements to row[2] elements

    nml_mat_row_addrow_r(m, 2, 1, 1.0);

    // Add row[1] to row[0] with a multiplier of 2.0

    nml_mat_row_addrow_r(m, 0, 1, 2.0);

    nml_mat_print(m);
    nml_mat_free(m);

    return 0;
}

To run the example:

./nml.sh clean examples && ./examples/row_plus_row.ex

Modifying the matrix structure

Remove rows and columns

To remove columns:

  • nml_mat *nml_mat_col_rem(nml_mat *m, unsigned int column)
    • A new matrix is being created, m remains the same.

To remove rows:

  • nml_mat *nml_mat_row_rem(nml_mat *m, unsigned int row)
    • A new matrix is being created, m remains the same.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {

    nml_mat *m = nml_mat_sqr_rnd(4, 1.0, 2.0);
    nml_mat_print(m);

    // Remove column[1] from m
    // m remains the same
    // less_columns is another matrix 
    nml_mat *less_columns = nml_mat_col_rem(m, 1);
    nml_mat_print(less_columns);

    // Remove row[0] from less_columns
    // less_columns remains the same
    // less_rows is another matrix
    nml_mat *less_rows = nml_mat_row_rem(less_columns, 0);
    nml_mat_print(less_rows);

    nml_mat_free(m);
    nml_mat_free(less_columns);
    nml_mat_free(less_rows);

    return 0;
}

To run the example:

./nml.sh examples && ./examples/remove_columns_rows.ex

Swap rows and columns

Use:

  • nml_mat *nml_mat_row_swap(nml_mat *m, unsigned int row1, unsigned int row2)
  • int nml_mat_row_swap_r(nml_mat *m, unsigned int row1, unsigned int row2)
  • nml_mat *nml_mat_col_swap(nml_mat *m, unsigned int col1, unsigned int col2)
  • int nml_mat_col_swap_r(nml_mat *m, unsigned int col1, unsigned int col2)
#include <stdlib.h>
#include <stdio.h>
#include <time.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
  srand(time(NULL));

  nml_mat *m = nml_mat_sqr_rnd(8, 0.0, 10.0);

  printf("m=");
  nml_mat_print(m);

  printf("m= (...after swapping col1=%d with col2=%d):\n", 1, 2);
  nml_mat_col_swap_r(m, 1, 2);
  nml_mat_print(m);

  printf("newm= (...after swapping col1=%d with col2=%d and creating a new matrix):\n", 0, 1);
  nml_mat *newm = nml_mat_col_swap(m, 0, 1);
  nml_mat_print(newm);

  printf("m= (...after swapping row1=%d with row2=%d)\n", 0, 2);
  nml_mat_row_swap_r(m, 0, 2);
  nml_mat_print(m);

  nml_mat_free(m);
  nml_mat_free(newm);

  return 0;
}

Output:

m=
3.467541		8.965948		0.685298		7.802247		2.363902		0.097955		6.325767		7.162469
9.613042		6.399923		3.512556		5.528620		9.520015		2.886378		1.363232		1.838682
2.722997		5.410983		2.397803		9.870863		9.601451		1.591262		4.335792		1.660161
2.318266		3.094436		8.189400		9.242210		3.818718		1.198454		2.423213		6.939399
0.471069		7.252268		8.869578		0.994544		5.301898		9.007567		0.172999		7.586261
2.295128		4.212713		3.072580		0.849523		7.941120		6.407214		6.049472		3.469933
9.157250		5.896395		0.705688		0.491878		6.985944		2.762460		8.673238		1.114106
4.783565		7.369584		0.592299		4.774641		7.395844		1.942471		7.115440		9.204845

m= (...after swapping col1=1 with col2=2):

3.467541		0.685298		8.965948		7.802247		2.363902		0.097955		6.325767		7.162469
9.613042		3.512556		6.399923		5.528620		9.520015		2.886378		1.363232		1.838682
2.722997		2.397803		5.410983		9.870863		9.601451		1.591262		4.335792		1.660161
2.318266		8.189400		3.094436		9.242210		3.818718		1.198454		2.423213		6.939399
0.471069		8.869578		7.252268		0.994544		5.301898		9.007567		0.172999		7.586261
2.295128		3.072580		4.212713		0.849523		7.941120		6.407214		6.049472		3.469933
9.157250		0.705688		5.896395		0.491878		6.985944		2.762460		8.673238		1.114106
4.783565		0.592299		7.369584		4.774641		7.395844		1.942471		7.115440		9.204845

newm= (...after swapping col1=0 with col2=1 and creating a new matrix):

0.685298		3.467541		8.965948		7.802247		2.363902		0.097955		6.325767		7.162469
3.512556		9.613042		6.399923		5.528620		9.520015		2.886378		1.363232		1.838682
2.397803		2.722997		5.410983		9.870863		9.601451		1.591262		4.335792		1.660161
8.189400		2.318266		3.094436		9.242210		3.818718		1.198454		2.423213		6.939399
8.869578		0.471069		7.252268		0.994544		5.301898		9.007567		0.172999		7.586261
3.072580		2.295128		4.212713		0.849523		7.941120		6.407214		6.049472		3.469933
0.705688		9.157250		5.896395		0.491878		6.985944		2.762460		8.673238		1.114106
0.592299		4.783565		7.369584		4.774641		7.395844		1.942471		7.115440		9.204845

m= (...after swapping row1=0 with row2=2)

2.722997		2.397803		5.410983		9.870863		9.601451		1.591262		4.335792		1.660161
9.613042		3.512556		6.399923		5.528620		9.520015		2.886378		1.363232		1.838682
3.467541		0.685298		8.965948		7.802247		2.363902		0.097955		6.325767		7.162469
2.318266		8.189400		3.094436		9.242210		3.818718		1.198454		2.423213		6.939399
0.471069		8.869578		7.252268		0.994544		5.301898		9.007567		0.172999		7.586261
2.295128		3.072580		4.212713		0.849523		7.941120		6.407214		6.049472		3.469933
9.157250		0.705688		5.896395		0.491878		6.985944		2.762460		8.673238		1.114106
4.783565		0.592299		7.369584		4.774641		7.395844		1.942471		7.115440		9.204845

To run the example:

./nml.sh examples && ./examples/swap_rows_and_columns.ex

Concatenate matrices

Two or more matrices can be concatenated (horizontally) or (vertically) into one matrix.

To achieve this, please use:

  • nml_mat *nml_mat_cath(unsigned int mnun, nml_mat **matrices)
    • For horizontal concatenation. A new matrix is returned.
    • num represents the number of matrices to concatenate.
    • matrices the matrices to be concatenated.
  • nml_mat *nml_mat_catv(unsigned int mnum, nml_mat **matrices)
    • For vertical concatenation. A new matrix is returned.
    • num represents the number of matrices to concatenate.
    • matrices the matrices to be concatenated.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
  nml_mat *I = nml_mat_eye(3);
  nml_mat *Ix2 = nml_mat_smult(I, 2.0);
  nml_mat *rndm = nml_mat_rnd(3, 4, 1.0, 5.0);

  nml_mat **ms = malloc(sizeof(*ms) * 2);
  ms[0] = I;
  ms[1] = Ix2;
  
  nml_mat *concats1 = nml_mat_cath(2, ms);

  ms[0] = concats1;
  ms[1] = rndm;

  nml_mat *concats2 = nml_mat_cath(2, ms);

  printf("\nConcatenate horizontally\n");
  printf("I=\n");
  nml_mat_print(I);
  printf("Ix2=\n");
  nml_mat_print(Ix2);
  printf("rndm=\n");
  nml_mat_print(rndm);
  printf("concats1=\n");
  nml_mat_print(concats1);
  printf("concats2=\n");
  nml_mat_print(concats2);

  free(ms);
  nml_mat_free(I);
  nml_mat_free(Ix2);
  nml_mat_free(concats1);
  nml_mat_free(concats2);
  nml_mat_free(rndm);

  // -------------------------------------
  // Vertical concatenation
  // -------------------------------------

  nml_mat *A = nml_mat_rnd(3, 4, 1.0, 4.0);
  nml_mat *B = nml_mat_rnd(5, 4, 10.0, 20.0);
  nml_mat *C = nml_mat_eye(4);

  nml_mat **ABarr = malloc(sizeof(*ABarr) * 2);
  ABarr[0] = A;
  ABarr[1] = B;
  nml_mat *ABCat = nml_mat_catv(2, ABarr);

  printf("\nA=\n");
  nml_mat_print(A);
  printf("\nB=\n");
  nml_mat_print(B);
  printf("\nC=\n");
  nml_mat_print(C);
  printf("\nA concat B =\n");
  nml_mat_print(ABCat);

  free(ABarr);
  nml_mat_free(A);
  nml_mat_free(B);
  nml_mat_free(C);

  return 0;
}

To run the example:

./nml.sh clean examples && ./examples/concatenate_matrices.ex

Matrices operations

Add and subtract matrices

To add or subtract two matrices, the following methods can be used:

  • nml_mat *nml_mat_add(nml_mat *m1, nml_mat *m2)
    • Adds two matrices, the results are kept in a new nml_mat*. m1 and m2 remain unchanged.
  • int nml_mat_add_r(nml_mat *m1, nml_mat *m2)
    • Add two matrices, the results are kept in m1. m2 remains unchanged.
  • nml_mat *nml_mat_sub(nml_mat *m1, nml_mat *m2)
    • Subtracts two matrices, the results are kept in a new nml_mat*. m1 and m2 remain unchanged.
  • int nml_mat_sub_r(nml_mat *m1, nml_mat *m2)
    • Subtracts two matrices, the results are kept in m1. m2 remains unchanged.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *m1 = nml_mat_sqr_rnd(4, 0.0, 10.0);
    nml_mat *m2 = nml_mat_sqr_rnd(4, 0.0, 10.0);
    printf("m1=\n");
    nml_mat_print(m1);
    
    printf("m2=\n");
    nml_mat_print(m2);


    // Add the matrices to, result is kept in m3
    // m1 and m2 remain unchanged
    nml_mat *m3 = nml_mat_add(m1, m2);
    printf("m3=\n");
    nml_mat_print(m3);

    // Add the matrices, the result is kept in m1
    // m1 is modified, m2 remains unchanged
    nml_mat_add_r(m1, m2);
    printf("m1=\n");
    nml_mat_print(m1);
    
    nml_mat_free(m1);
    nml_mat_free(m2);
    nml_mat_free(m3);

    return 0;
}

To run the example:

./nml.sh examples && examples/add_matrices.ex

Multiply matrices (dot)

To multiply two matrices, the following method can be used:

  • nml_mat *nml_mat_dot(nml_mat *m1, nml_mat *m2)
    • Multiplies two matrices, the result is kept in a new nml_mat*. m1 and m2 remain unchanged.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *m1 = nml_mat_sqr_rnd(4, 0.0, 10.0);
    nml_mat *m2 = nml_mat_sqr_rnd(4, 0.0, 10.0);

    printf("m1=\n");
    nml_mat_print(m1);
    
    printf("m2=\n");
    nml_mat_print(m2);

    // Multiply matrices
    nml_mat *m3 = nml_mat_dot(m1, m2);
    printf("m3=\n");
    nml_mat_print(m3);

    nml_mat_free(m1);
    nml_mat_free(m2);
    nml_mat_free(m3);

    return 0;
}

To run the example:

./nml.sh examples && examples/dot_matrices.ex

Transpose matrices

To transpose a matrix, the following method can be used:

  • nml_mat *nml_mat_transp(nml_mat *m)
    • A new nml_mat* will be created, representing the transpose matrix of m. m remains unchanged.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *m1 = nml_mat_rnd(1, 5, 1.0, 10.0);
    nml_mat_print(m1);

    nml_mat *m2 = nml_mat_transp(m1);
    nml_mat_print(m2);

    nml_mat_free(m1);
    nml_mat_free(m2);

    return 0;
}

To run the example:

./nml.sh clean examples && examples/transpose.ex

Calculate trace

To calculate the trace of the matrix the following method can be used: double nml_mat_trace(nml_mat* m).

Row Echelon

Calculate Row Echelon Form using Gaussian Elimination

To bring the matrix in Row Echelon Form the following method can be used: nml_mat *nml_mat_ref(nml_mat *m).

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {

  double v1[9] = {
    0.0, 1.0, 2.0,
    1.0, 2.0, 1.0,
    2.0, 7.0, 8.0
  };

  nml_mat *m1 = nml_mat_from(3, 3, 9, v1);
  printf("\nm1=\n");
  nml_mat_print(m1);
  nml_mat *refm1 = nml_mat_ref(m1);
  printf("\nrefm1=\n");
  nml_mat_print(refm1);

  nml_mat_free(m1);
  nml_mat_free(refm1);
  return 0;
}

Output:

m1=

0.000000		1.000000		2.000000
1.000000		2.000000		1.000000
2.000000		7.000000		8.000000


refm1=

1.000000		2.000000		1.000000
0.000000		1.000000		2.000000
0.000000		0.000000		0.000000

To run the example:

./nml.sh examples && ./examples/row_echelon.ex

Calculate Reduced Row Echelon Form using Gauss-Jordan

To bring the matrix in Reduced Row Echelon Form the following method can be used: nml_mat *nml_mat_rref(nml_mat *m).

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {

  double v1[9] = {
    0.0, 1.0, 2.0,
    1.0, 2.0, 1.0,
    2.0, 7.0, 8.0
  };

  nml_mat *m1 = nml_mat_from(3, 3, 9, v1);
  printf("\nm1=\n");
  nml_mat_print(m1);
  nml_mat *rrefm1 = nml_mat_rref(m1);
  printf("\nrrefm1=\n");
  nml_mat_print(rrefm1);

  nml_mat_free(m1);
  nml_mat_free(rrefm1);
  return 0;
}

Output:

m1=

0.000000		1.000000		2.000000
1.000000		2.000000		1.000000
2.000000		7.000000		8.000000


rrefm1=

1.000000		0.000000		-3.000000
-0.000000		1.000000		2.000000
0.000000		0.000000		0.000000

To run the example:

./nml.sh clean examples && examples/reduced_row_echelon.ex

LU(P) Decomposition

To decompose a matrix using LU you can use: nml_mat_lup *nml_mat_lup_solve(nml_mat *m).

The result is a pointer nml_mat_lup*:

typedef struct nml_mat_lup_s {
  nml_mat *L;
  nml_mat *U;
  nml_mat *P;
  unsigned int num_permutations;
} nml_mat_lup;

To free the nml_mat_lup*. Please use void nml_mat_lup_free(nml_mat_lup* lu). This will also deallocate the memory for the three internal nml_mat* pointers.

LU decomposition is used for solving linear systems of equations, computing the determinant and the inverse of a matrix.

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *m1 = nml_mat_sqr_rnd(4, 0.0, 10.0);
    printf("m1=\n");
    nml_mat_print(m1);

    nml_mat_lup *m1_lup = nml_mat_lup_solve(m1);
    printf("L, U, P:\n");
    nml_mat_lup_print(m1_lup);

    nml_mat_free(m1);
    nml_mat_lup_free(m1_lup);

    return 0;
}

Output:

m1=

0.000078		1.315378		7.556053		4.586501
5.327672		2.189592		0.470446		6.788647
6.792964		9.346929		3.835021		5.194164
8.309653		0.345721		0.534616		5.297002

L, U, P:

1.000000		0.000000		0.000000		0.000000
0.817479		1.000000		0.000000		0.000000
0.000009		0.145116		1.000000		0.000000
0.641143		0.217108		-0.086373		1.000000


8.309653		0.345721		0.534616		5.297002
0.000000		9.064309		3.397983		0.863978
0.000000		0.000000		7.062947		4.461075
0.000000		0.000000		0.000000		3.590254


0.000000		0.000000		0.000000		1.000000
0.000000		0.000000		1.000000		0.000000
1.000000		0.000000		0.000000		0.000000
0.000000		1.000000		0.000000		0.000000

Running the example:

./nml.sh clean examples && examples/lup.ex

Matrix inverse

Calculating the inverse requires to decompose the matrix LU(P) decomposition first.

Afterwards obtaining the inverse is straightforward: nml_mat *nml_mat_inv(nml_mat_lup *m).

Example:

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
  printf("\nInverse of a matrix:\n");
  double m_v[16] = {
    2.0, 7.0, 6.0, 1.0,
    9.0, 5.0, 0.0, 2.0,
    4.0, 3.0, 8.0, 3.0,
    3.0, 5.0, 1.0, 9.0
  };
  nml_mat *m = nml_mat_from(4,4,16, m_v);
  nml_mat_lup *lup = nml_mat_lup_solve(m);
  nml_mat* minv = nml_mat_inv(lup);
  nml_mat *mdotminv = nml_mat_dot(m, minv);

  printf("m=");
  nml_mat_print(m);
  printf("minv=");
  nml_mat_print(minv);
  printf("(%%e) m * minv=");
  nml_mat_printf(mdotminv, "%e\t");
  printf("(%%f) m * minv=");
  nml_mat_printf(mdotminv, "%f\t");

  nml_mat_free(m);
  nml_mat_free(minv);
  nml_mat_free(mdotminv);
  return 0;
}

Output:

m=
2.000000		7.000000		6.000000		1.000000
9.000000		5.000000		0.000000		2.000000
4.000000		3.000000		8.000000		3.000000
3.000000		5.000000		1.000000		9.000000

minv=
-0.081577		0.112583		0.065924		-0.037929
0.174895		0.013245		-0.133955		0.022276
0.001505		-0.046358		0.127935		-0.032511
-0.070138		-0.039735		0.038230		0.114991

(%e) m * minv=
1.000000e+00	4.163336e-17	4.163336e-17	-1.387779e-17
0.000000e+00	1.000000e+00	-2.775558e-17	2.775558e-17
5.551115e-17	6.938894e-17	1.000000e+00	5.551115e-17
0.000000e+00	5.551115e-17	-5.551115e-17	1.000000e+00

(%f) m * minv=
1.000000	0.000000	0.000000	-0.000000
0.000000	1.000000	-0.000000	0.000000
0.000000	0.000000	1.000000	0.000000
0.000000	0.000000	-0.000000	1.000000

To run the example:

./nml.sh clean examples && ./examples/inverse.ex

Matrix determinant

Calculating the determinant requires to decompose the matrix LU(P) decomposition first.

Afterwards obtaining the determinant is straightforward: double nml_mat_det(nml_mat_lup* lup).

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *m1 = nml_mat_sqr_rnd(4, 0.0, 10.0);
    nml_mat_lup *m1_lup = nml_mat_lup_solve(m1);
    
    printf("m1=\n");
    nml_mat_print(m1);
    printf("determinant=%lf\n", nml_mat_det(m1_lup));

    nml_mat_free(m1);
    nml_mat_lup_free(m1_lup);

    return 0;
}

Output:

m1=

0.000078		1.315378		7.556053		4.586501
5.327672		2.189592		0.470446		6.788647
6.792964		9.346929		3.835021		5.194164
8.309653		0.345721		0.534616		5.297002

determinant=-1909.979877

Runnning the example:

./nml.sh clean examples && examples/determinant.ex

Solve linear systems of equations

Solving A * x = B where A is lower triangular (Forward Substitution)

Use: nml_mat *nml_ls_solvefwd(nml_mat *low_triang, nml_mat *b).

Note: no validation will be performed to check is low_triang is a lower triangular matrix

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    
    FILE *input = fopen("./examples/data/matrix9_lower_triangular.data", "r");

    nml_mat *A = nml_mat_fromfilef(input);
    nml_mat *B = nml_mat_fromfilef(input);
    nml_mat *x = nml_ls_solvefwd(A, B);

    nml_mat_print(A);
    nml_mat_print(B);
    nml_mat_print(x);

    nml_mat_free(A);
    nml_mat_free(B);
    nml_mat_free(x);

    fclose(input);

    return 0;
}

To run the example:

./nml.sh clean examples && examples/forward_substition.ex

Solving A * x = B where A is upper traingular (Backward Substition)

Use: nml_mat *nml_ls_solvebck(nml_mat *upper_triang, nml_mat *b).

Note: no validation will be performed to check is upper_triang is an upper triangular matrix

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    
    FILE *input = fopen("./examples/data/matrix10_upper_triangular.data", "r");

    nml_mat *A = nml_mat_fromfilef(input);
    nml_mat *B = nml_mat_fromfilef(input);
    nml_mat *x = nml_ls_solvebck(A, B);

    nml_mat_print(A);
    nml_mat_print(B);
    nml_mat_print(x);

    nml_mat_free(A);
    nml_mat_free(B);
    nml_mat_free(x);

    fclose(input);

    return 0;
}

To run the example:

./nml.sh clean examples && examples/backward_substitution.ex

Solving A * x = B using LU(P) decomposition

Use: nml_mat *nml_ls_solve(nml_mat_lup *lup, nml_mat* b).

#include <stdlib.h>
#include <stdio.h>

#include "lib/nml.h"

int main(int argc, char *argv[]) {
    nml_mat *A = nml_mat_sqr_rnd(4, 1.0, 10.0);
    nml_mat *B = nml_mat_rnd(4, 1, 1.0, 10.0);
    nml_mat_lup *LUP = nml_mat_lup_solve(A);

    nml_mat *x = nml_ls_solve(LUP, B);
    nml_mat_print(x);

    nml_mat_free(A);
    nml_mat_free(B);
    nml_mat_free(x);
    nml_mat_lup_free(LUP);
}

To run the example:

./nml.sh clean examples && ./examples/ls_solve.ex

About

nml is a "simple" matrix/numerical analysis library written in pure C. The scope of the library is to highlight various algorithm implementations related to matrices. Code readability was a major concern.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C 81.9%
  • Python 11.3%
  • Shell 6.8%