0

In Calculus of variations, I. M. Gelfand, S. V. Fomin, Section 13 they are showing how to variate functional with variable endpoints. How do they get from (4) to the next formula of the differential of the functional? How do they simplify the integrals which boundaries have pluses-the last two integrals in (4)? Edit: I have uploaded picture of that section, sorry for not typing it but it will take quite some time. The problem

Blake
  • 355
  • 1
    Not everyone would have access to this book - it would be worth writing the formulae here so everyone can see what the problem is. – John Doe May 30 '17 at 16:53
  • Done. Thank you for reminding me, I was good idea. – Blake May 30 '17 at 16:58
  • Typing that up takes maybe 5-10 minutes, and if you're asking others to answer I don't think it's unreasonable to expect you to put that effort in. (And by the way, a screenshot would be better than taking a photo of the screen.) – epimorphic May 30 '17 at 17:51
  • I am not that guy that will start to defend himself but why did you write something negative? First of all I am not so good at Latex and it seems to take me more time to type it which is my problem of course. Second, I am asking this question from phone and it is even harder to write Latex from the phone for me. Third, I took picture because I am reading from TV that has PDF reader and from the smartphone I did not managed to take the snapshot correctly such that the two formulations are on one screenshot. I uploaded this because I was in a hurry to improve my question and to get good answer. – Blake May 30 '17 at 18:46

1 Answers1

1

$$\small\int_{x_0}^{x_1}F(x,y+h,y'+h')-F(x,y,y')\,dx+\int_{x_1}^{x_1+\delta x_1}F(x,y+h,y'+h')\,dx-\int_{x_0}^{x_0+\delta x_0}F(x,y+h,y'+h')\,dx$$

The first integral is simplified with Taylor's theorem.
The second becomes $$\int_{x_1}^{x_1+\delta x_1}\color{red}{F(x,y,y')}+\color{blue}{hF_y(x,y,y')}+\color{green}{h'F_{y'}(x,y,y')}\,dx\\ \small=\color{red}{[F(x,y,y')]_{x_1}\delta x_1}+\color{blue}{\int_{x_1}^{x_1+\delta x_1}hF_y(x,y,y')\,dx}+\color{green}{\left[hF_{y'}(x,y,y')\right]_{x_1}^{x_1+\delta x_1}-\int_{x_1}^{x_1+\delta x_1}h\frac d{dx}\big(F_{y'}(x,y,y')\big)\,dx}$$ Explanation:
$\color{red}{\text{red}}$ - For $\delta x_1$ small, this is just a small strip of the function from $x_1$ to $x_1+\delta x_1$, and $F$ hardly changes as a result. So this integral is just the area of a very thing strip (a rectangle), so it becomes base $\times$ height, where base is $\delta x_1$ and height is $F(x,y,y')$ evaluated at $x_1$.
$\color{blue}{\text{blue}}$ - Nothing changes.
$\color{green}{\text{green}}$ - integration by parts, differentiate the function with $F$ and integrate the $h'$.

The third integral follows a similar process as this, but just has negative signs.


Edit:

So the total integral becomes

$$\int_{x_0}^{x_1}hF_y(x,y,y')-h\frac d{dx}\big(F_{y'}(x,y,y')\big)\,dx\\ \small+{[F(x,y,y')]_{x_1}\delta x_1}+{\int_{x_1}^{x_1+\delta x_1}hF_y(x,y,y')\,dx}+{\left[hF_{y'}(x,y,y')\right]_{x_1}-\int_{x_1}^{x_1+\delta x_1}h\frac d{dx}\big(F_{y'}(x,y,y')\big)\,dx}\\\small-{[F(x,y,y')]_{x_0}\delta x_0}-{\int_{x_0}^{x_0+\delta x_0}hF_y(x,y,y')\,dx}-{\left[hF_{y'}(x,y,y')\right]_{x_0}+\int_{x_0}^{x_0+\delta x_0}h\frac d{dx}\big(F_{y'}(x,y,y')\big)\,dx}\\ =\int_{x_0+\delta x_0}^{x_1+\delta x_1}h\left(F_y-\frac d{dx}\big(F_{y'}\big)\right)\,dx+F\big|_{x_1}\delta x_1-F\big|_{x_0}\delta x_0+F_{y'}\cdot h\big|_{x_1}-F_{y'}\cdot h\big|_{x_0}$$

John Doe
  • 14,545
  • Thank you so much for the answer, really appreciate your work. I have one last question, it is really stupid but I am stuck with this all day and cannot concentrate. In the second form in the picture there aren't any integrals left, maybe they cancel each other but cannot see how. Please show me how. – Blake May 30 '17 at 17:39
  • @Blake I have added an edit. Continuing from the final result, I am not 100% sure how to remove the $\delta$'s. I thought of this, but I am not sure if it is really valid, or if there is a better way of writing it: $$$$ "Then you can remove the $\delta$'s from the integral by assuming they are small in comparison to the $x_i$'s and so give a value approximately equal to the value at the $x_i$'s." – John Doe May 30 '17 at 20:02
  • Thank you very very much. – Blake May 30 '17 at 20:18