Is this a normal problem of double-precision floating-point
Try using the nearly equal node with a tolerance setting.