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

Delta vs differential

  1. Nov 20, 2009 #1
    Hi guys

    Can anybody help me? What is the difference between a delta [tex]\delta W[/tex] and a differential [tex]dW[/tex]? ([tex]W[/tex] a scalar function, for example.) In other words, when shold be used a delta and when a differential? Thanks.
  2. jcsd
  3. Nov 20, 2009 #2


    User Avatar
    Science Advisor
    Homework Helper
    Gold Member

    Suppose W is a differentiable function of x. Consider some value of x, say x = a and W(a). Now suppose we have a "nearby" point b = a + h, so h would be small. Then:

    [itex]\Delta W = W(b) - W(a)[/itex] represents the [exact] change in W from a to b.

    The differential of W is defined to be the change on the tangent line at a:

    [itex]dW = W'(a)h[/itex].

    For small h we have [itex]\Delta W \approx dW[/itex]

    I'm assuming that your use of [itex]\delta[/itex] has the same meaning as the common usage of [itex]\Delta[/itex]. If I'm wrong about that, feel free to ignore this reply :rolleyes:
  4. Nov 20, 2009 #3


    User Avatar
    Gold Member

    The only time I've ever seen delta used like that is with inexact differentials, but I've never done any work with them. Here's a wikipedia article about them though: http://en.wikipedia.org/wiki/Inexact_differential
  5. Nov 20, 2009 #4
    I've also seen [itex]\delta[/itex] used as the variation of a function (calculus of variations or differential geometry). Ie., [itex]W(x) + \delta W(x)[/itex] where [itex]\delta W(x)[/itex] is a function that is "small" in the neighborhood of interest.
  6. Nov 21, 2009 #5


    User Avatar
    Science Advisor

    dX is in mathematical terms something which is called a one-form. You can integrate it to obtain X. Physically, this X has to be well-defined then. In thermodynamics for instance the quantity X has to be a function of state. A counterexample would be the heat Q or the mechanical work W. You can't define a state with defnit heat or mechanical work; these quantities only have meaning if you go from one thermodynamical state to another. So if I would write dQ or dW for the changes, this would imply that I could obtain Q and W for a state by integrating, which is not well-defined. That's why people often choose to write [itex]\delta[/itex] instead of d for these quantities.

    If you want to know the exact mathematical difference, the answer lies in differential geometry I think; like I said, a quantity dX is in diff.geometry a one-form which lives in a dual vector space called the dual tangent space, while [itex]\delta X[/itex] indicates either an arbitrary change (like in the variational principle; here the [itex]\delta[/itex] gets you from one field solution to another which can't be accomplished by a mere coordinate transformation), a coordinate change (if you want for instance to know the behaviour of X under a spacetime transformation; here X is in a representation of some group which describes coordinate transformations like the Lorentz group, the Poincare group or the Galilei group) or a change in some internal space (where X is then a gauge field in some representation of some gauge group and where you perform in infinitesimal gauge transformation).

    I hope this helps a little :)
  7. Nov 22, 2009 #6
    Hi all.
    My original question had to do with a problem of electrodynamics. I have not yet a clear answer but the support received has given me a broader perspective about the problem. Thank you all for your help.
  8. Nov 25, 2009 #7
    In my experience it was used as a precursor for differentiation, for example,

    The gradient of the line connecting the points (f(x), x) and (f(x+δx), x+δx) is [f(x+δx)-f(x)]/δx, in the limit δx -> dx, we get the gradient to be df/dx.

    Ie. δf = f(x+dx) - f(x), and df = f(x+dx) - f(x)

    This seems to be how a lot of physics lecturers used calculus, although I can't say I'd ever seen this in my maths introduction.

    Just seems to me to be a bit of a formalism to make the differentiation clear, without resorting to limits as something goes to zero.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook