# limit of series 1 A

Show 40 post(s) from this thread on one page
Page 2 of 2 First 12
• Jan 18th 2011, 10:29 AM
HallsofIvy
Quote:

Originally Posted by transgalactic
i have dy/dx=x+y

y0=1
y1=1+x+(x^2)/2
y2=1+x+(x^2)/2+(x^3)/6
what is the limit of the series yn
?

The limit of that sequence is $e^x$ as you should be able to recognize the MacLaurin series for $e^x$
However, the derivative of $y= e^x$ is $e^x= y$, not x+ y.

If you are trying to use Picard's method to solve this equation, as Ackbeet suggested, then you are doing it wrong.

From dy/dx= x+y we get dy= (x+y)dx so that $y= \int (x+ y)dx$. Of course, since we don't know y but we can approximate it. If we are given y(0)= 1 (which you don't say but is implied by "y0= 1") as a first approximation, assume y(x)= 1 for all x. Then dy/dx= x+1 so that dy= (x+1)dx and [tex]y= (1/2)x^2+ x+ C[tex] and y(0)= C= 1 gives $y1= (1/2)x^2+ x+ 1$, as you have.

But then, $y2= \int x+ y1 dx= \int x+ (1/2)x^2+ x+ 1 dx= \int (1/2)x^2+ 2x+ 1 dx= (1/6)x^3+ x^2+ x+ C$ and, again, $y(0)= C= 1$ gives $y2= (1/6)x^2+ x^2+ x+ 1$ which is NOT what you have. I expect you forgot the added x.

Now, $y3= \int x+ y2 dx= \int x+ (1/6)x^3+ x^2+ x+ 1 dx$ $= \int (1/6)x^3+ x^2+ 2x+ 1 dx= (1/24)x^4+ (1/3)x^3+ x^2+ x+ C$ and $y(0)= C= 1$ gives $y3= (1/24)x^4+ (1/3)x^3+ x^2+ x+ 1$.

One more time: $y4= \int x+ y3 dy= \int x+(1/24)x^3+ (1/3)x^3+ x^2+ x+ 1 dx$ $= \int (1/24)x^4+ (1/3)x^3+ x^2+ 2x+ 1 dx= (1/5!)x^5+ (1/12)x^4+ (1/3)x^3+ x^2+ x+ C$ and $y(0)= C= 1$ gives
$y4= (1/5!)x^5+ (2/4!)x^4+ (2/3!)x^3+ (2/2!)x^2+ x+ 1$.

I think I would conjecture now that
$yn= (1/n!)x^n+ (2/(n-1)!)x^{n-1}+ \cdot\cdot\cdot+ (2/3!)x^3+ (2/2!)x^2+ x+ 1$
and I would attempt to prove that by "induction"

When n= 0, that says that y0= 1 which is true.

Now assume that, for some k, $yk= (1/k!)x^k+ (2/(k-1)!)x^{k-1}+\cdot\cdot\cdot+ (2/3!)x^3+ (2/2!)x^2+ x+ 1$ so that
$y(k+1)= \int x+ yk dx= \int (1/k!)x^k+ (2/(k-1)!)x^{k-1}+ \cdot\cdot\cdot+ (2/3!)x^3+ (2/2!)x^2+ 2x+ 1 dx$ $= (1/(k+1)!)x^{k+1}+ (2/k!)x^k+ \cdot\cdot\cdot+ (2/4!)x^4+ (2/3!)x^3+ x^2+ x+ C$ and $y(k+1)(0)= C= 1$ gives
$y(k+1)= (1/(k+1)!)x^{k+1}+ (2/k!)x^k+ \cdot\cdot\cdot+ (2/4!)x^4+ (2/3!)x^3+ (2/2!) x^2+ x+ 1$
just as we wish.
• Jan 21st 2011, 03:25 AM
transgalactic
Quote:

Originally Posted by Ackbeet
Ok. So y(0) = 1 implies that

$y_{0}=1,$

$y_{1}=\displaystyle\int(x+1)\,dx=1+x+\frac{x^{2}}{ 2},$

$y_{2}=\displaystyle\int\left(x+1+x+\frac{x^{2}}{2} \right)\,dx=1+x+x^{2}+\frac{x^{3}}{6},$

$y_{3}=\displaystyle\int\left(x+1+x+x^{2}+\frac{x^{ 3}}{6}\right)dx=1+x+\frac{x^{2}}{2}+\frac{x^{3}}{6 }+\frac{x^{4}}{24},\dots$

That's what you got, I believe. The sequence $\{y_{n}\}$ converges to $e^{x}.$ To convince yourself of that fact, just expand $e^{x}$ in a Taylor series about 0, and you'll see that it's the exact same series as you're getting here.

How rigorous a proof does this require?

i dont have y2 of a exponent
y2=1+ $\int t+1+t+\frac{t^3}{6}dt$

i get $t^2$
i should get $\frac{t^2}{2}$
Show 40 post(s) from this thread on one page
Page 2 of 2 First 12