Why add a scale factor to the boundaries in piecewise_jerk_problem? Wouldn't this undermine the original physical constraints?

[dajiba6]

  // set x, x', x'' bounds
  for (int i = 0; i < num_of_variables; ++i) {
    if (i < n) {
      variables[i].emplace_back(constraint_index, 1.0);
      lower_bounds->at(constraint_index) =
          x_bounds_[i].first * scale_factor_[0];
      upper_bounds->at(constraint_index) =
          x_bounds_[i].second * scale_factor_[0];
    } else if (i < 2 * n) {
      variables[i].emplace_back(constraint_index, 1.0);

      lower_bounds->at(constraint_index) =
          dx_bounds_[i - n].first * scale_factor_[1];
      upper_bounds->at(constraint_index) =
          dx_bounds_[i - n].second * scale_factor_[1];
    } else {
      variables[i].emplace_back(constraint_index, 1.0);
      lower_bounds->at(constraint_index) =
          ddx_bounds_[i - 2 * n].first * scale_factor_[2];
      upper_bounds->at(constraint_index) =
          ddx_bounds_[i - 2 * n].second * scale_factor_[2];
    }
    ++constraint_index;
    ADEBUG<<"scale factor:"<<scale_factor_[0]<<" "<<scale_factor_[1]<<" "<<scale_factor_[2];
    ADEBUG << "ddx_bounds: lower, upper";
    for (int i = 0; i < n; ++i) {
      ADEBUG << ddx_bounds_[i].first << " " << ddx_bounds_[i].second;
    }
  }