source: CIVL/examples/accuracy/laplace2d.cvl@ 1aaefd4

/* 2-dimensional Laplace equation solver with constant boundaries.
 * This operation 2nd order accurate.  To verify with CIVL:
 *              civl verify laplace.cvl
 */

$input int rows; // the number of rows
$assume(rows >= 1);
$input int cols; // the number of columns
$assume(cols >= 1);
$input double h; // distance between two consecutive points in horizontal or vertical direction
$assume(0<h && h<1);
$input double u[rows][cols]; // initial state
double result[rows][cols]; // new state after update
// assume phi:R^2->R has 4 continuous derivatives in [-1,1]x[-1,1]:
$abstract $differentiable(4, [-1,1][-1,1]) $real phi($real x, $real y);

/* Performs one update step */
void laplace() {
  /*@ loop invariant 1<=i && i<=rows-1;
    @ loop invariant $forall (int i0 : 1..i-1)
    @   $forall (int j:1..cols-1)
    @      result[i0][j] = (u[i0-1][j]+u[i0][j-1]-4*u[i0][j]+u[i0+1][j]+u[i0][j+1])/(h*h);
    @ loop assigns i, result[1..rows-2][1..cols-2];
    @*/
  for (int i = 1; i < rows-1; i++) {
    /*@ loop invariant 1<=j && j<=cols-1;
      @ loop invariant $forall (int j0 : 1..j-1)
      @   result[i][j0] = (u[i-1][j0]+u[i][j0-1]-4*u[i][j0]+u[i+1][j0]+u[i][j0+1])/(h*h);
      @ loop assigns j, result[i][1..cols-2];
      @*/
    for (int j = 1; j < cols-1; j++) {
      result[i][j] = (u[i-1][j]+u[i][j-1]-4*u[i][j]+u[i+1][j]+u[i][j+1])/(h*h);
    }
  }
}

void main() {
  $assume($forall (int i,j:($domain(2)){0..rows-1, 0..cols-1}) u[i][j] == phi(i*h, j*h));
  $assume(rows*h<=1 && cols*h<=1);
  laplace();
  $assert($uniform (int i,j:($domain(2)){1..rows-2, 1..cols-2})
    result[i][j] - $D[phi,{x,2}](i*h,j*h) - $D[phi,{y,2}](i*h,j*h) == $O(h*h));
}