set_row, set_rowex

set a constraint in the lp.

unsigned char set_row(lprec *lp, int row_no, REAL *row);

unsigned char set_rowex(lprec *lp, int row_no, int count, REAL *row, int *colno);

Return Value

set_row, set_rowex return TRUE (1) if the operation was successful. A return value of FALSE (0) indicates an error.



Pointer to previously created lp model. See return value of make_lp, copy_lp, read_lp, read_LP, read_mps, read_freemps, read_MPS, read_freeMPS, read_XLI


The row number that must be changed.


Number of elements in row and colno.


An array with 1+get_Ncolumns (count for set_rowex, if colno is different from NULL) elements that contains the values of the row.


A zero-based array with count elements that contains the column numbers of the row. However this variable can also be NULL. In that case element i in the variable row is column i and values start at element 1.


The set_row, set_rowex functions change the values of an existing row in the model at once.

Note that add_constraint, add_constraintex, str_add_constraint add a row to the model, making the number of rows one larger. These functions change an existing row.

Note that for set_row (and set_rowex when colno is NULL) element 1 of the array is column 1, element 2 is column 2, ... element 0 is not used.

set_rowex has the possibility to specify only the non-zero elements. And in contrary to set_row, set_rowex reads the arrays starting from element 0. However when colno is NULL then set_rowex acts as set_row and then values start at element 1. When colno is provided, then it specifies the column numbers of the non-zero elements. This will speed up building the model considerably if there are a lot of zero values.

In most cases the matrix is sparse and has many zero value. Thus it is almost always better to use set_rowex instead of set_row. set_rowex is always at least as performant as set_row.

It is more performant to call these functions than multiple times set_mat.

Note that unspecified values by set_rowex are set to zero.

Note that these routines will perform much better when set_add_rowmode is called before adding constraints.


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

int main(void)
  lprec *lp;
  REAL row[1+3];     /* must be 1 more than number of columns ! */
  REAL sparserow[2]; /* must be the number of non-zero values */
  int colno[2];

  /* Create a new LP model */
  lp = make_lp(2, 3);
  if(lp == NULL) {
    fprintf(stderr, "Unable to create new LP model\n");

  row[1] = 1.0;
  row[2] = 0.0; /* also zero elements must be provided */
  row[3] = 2.0;
  set_row(lp, 1, row); /* changes the values of existing row 1 */
  colno[0] = 1; sparserow[0] = 1.0; /* column 1 */
  colno[1] = 3; sparserow[1] = 2.0; /* column 3 */
  set_rowex(lp, 2, 2, sparserow, colno);


lp_solve API reference

See Also make_lp, copy_lp, copy_lp, read_lp, read_LP, read_mps, read_freemps, read_MPS, read_freeMPS, read_XLI, get_row, get_rowex, add_constraint, add_constraintex, str_add_constraint, set_obj_fn, set_obj_fnex, str_set_obj_fn, set_obj, set_add_rowmode, is_add_rowmode, get_constr_type, is_constr_type, del_constraint, add_column, add_columnex, str_add_column, set_column, set_columnex, get_column, get_columnex, get_mat