On a low dimensional model for magnetostriction