The 2nd time period is due to your adjust in curiosity rate. $varepsilon$ is solely what You can not demonstrate. If everything is neat, your $varepsilon$ really should not be way too substantial. You can even see that this is quite close to a Taylor expansion when everything is linear, https://www.youtube.com/watch?v=qMmsQ4kKgY4