The following discussion assumes that the walls are either horizontal or vertical i.e. aligned to the x and y axis. For angled walls the maths becomes significantly more complex.


There are 2 common ways to define movement in 2D space.

1) The direction the object is moving measured as an angle measured clockwise from the positive x axis and a speed (a simple scalar value) i.e. angle and speed

2) The objects movement is defined by 2 velocities measured along the x and y axis i.e. vx and vy

In the real world we think of (1) but in computer programs the maths is less complex when using (2) for both movement and rebounding off walls.

With option (2) then if there is a collision with a horizontal wall we can negate the x velocity and for a vertical wall negate the y velocity i.e.

1. // Horizontal wall
2. vx = -vx;
3. // Vertical wall
4. vy = -vy;
   

With (1) the maths becomes more challenging

1. // Rebound using speed and angle
2. float vx = speed * cos(angle);
3. float vy = speed * sin(angle);
4. // If horizontal wall then
   
5. vx = -vx;
6. // If vertical wall then
   
7. vy = -vy;
8. angle = atan2(vy,vx);
   

Updating the objects position also becomes easier with option (2). If the object is in position [x,y] then the new position [nx,ny] after time t has elaspsed is

1. // Option 2
2. nx = x + vx * t;
3. ny = y * vy * t;

and for (1)

1. // Option1
2. nx = x + speed * cos(angle) * t;
3. ny = y + speed * sin(angle) * t;
   

What you might of noticed is when using option (1) apart from the extensive use of the trignometric functions the calculations involve resolving the speed along the x and y axis anyway.

Note: angle is measured in radians.

HTH