Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Taylors Series

  1. Sep 15, 2007 #1
    I was going through the derivation of the Taylors series in my book (Engineering Mathematics by Jaggi & Mathur), and there was one step that escaped me. They proved that the derivative of f(x+h) is the same wrt h and wrt (x+h). If someone could explain that, Id be really grateful.
  2. jcsd
  3. Sep 15, 2007 #2
    Do you mean df(x+h)/dh = df(x+h)/d(x+h)? If so that just does not make any sense, you cannot say that. However, engineering books and physics books on math sometimes use erroneous manipulation which leads to correct results. I am one who does not understand their manipulations so I cannot help with that, but I can see right now that is invalid.
  4. Sep 15, 2007 #3
    for the derivation of Taylor Series... just keep doing integration by part until you get bored with it.
  5. Sep 15, 2007 #4


    User Avatar
    Science Advisor

    Kummer, why does df(x+h)/dh = df(x+h)/d(x+h) make no sense?

    By the chain rule, df(x+h)/dh= f'(x+h) since the derivative of x+ h, wrt h, is 1. To do
    df(x+h)/d(x+h) more easily, change the variable. Let y= x+ h so the problem becomes
    df(y)/dy. Of course, that is f"(y)= f'(x+h).
  6. Sep 15, 2007 #5
    I see what you did.

    I am just mentioning I do not know, usually, differential usage. For instance, I have seen books write d(x+1) I am not sure what this means. And I have seen dx^2, this is even more mysterious.
  7. Sep 16, 2007 #6

    Gib Z

    User Avatar
    Homework Helper

    Its just like in a change of variables, say we have the integral [tex]\int \frac{log x}{x} dx[/tex] and we let u= log x, then integral becomes [tex]\int u du[/tex] which, we could just write differently as [tex]\int log x d(log x)[/tex].
  8. Sep 16, 2007 #7


    User Avatar
    Science Advisor

    You can differentiate a function of x with respect to any function of x.
    [tex]\frac{dg(x)}{df(x)}= \frac{dg(x)}{dx}\frac{dx}{df(x)}= \frac{dg(x)}{dx}\frac{1}{\frac{df}{dx}}[/tex]
    by the chain rule.

    In particular,
    [tex]\frac{dg(x)}{dx^2}= \frac{\frac{dg}{dx}}{2x}[/tex]
  9. Sep 16, 2007 #8
    It's a classic situation where a student of analysis (when he's only part way through his course) will cry foul. He will claim that differentiation is defined as an operation to be done on *one* function, by the taking of a particular limit. Unfortunately (or fortunately, I would say), physics and engineers follow an *algebraic* tradition, over pure analysis. We tend to just perform algebraic manipulations, without particular regard over the validity, until they fail, and we learn that step is not valid in some context. It says a lot that the algebraic route can work well enough that whole generations of physicists and engineers have used it to great effect without ever learning the distinction. In fact, only recently (by mathematical standards), has the algebraic method gained more legitimacy: search for non-standard analysis, or synthetic differential geometry.

    As maybe obvious, this happens to be one of my personal little nooks of fascination... :blushing:
  10. Sep 21, 2007 #9
    Thank you. My exams just finished, sorry for taking so long to reply.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook