Difference between revisions of "MediaWiki:Sitenotice id"
(alternate solution for 1d) |
|||
Line 24: | Line 24: | ||
''@Un vieil homme:'' I think the Fubini idea doesn't work. We've proven Fubini at Kowalski's for Riemann-integrable functions so I guess that's the theorem we can use. Fubini only works for compact intervals which is not the case above. - A. | ''@Un vieil homme:'' I think the Fubini idea doesn't work. We've proven Fubini at Kowalski's for Riemann-integrable functions so I guess that's the theorem we can use. Fubini only works for compact intervals which is not the case above. - A. | ||
+ | |||
+ | ---- | ||
+ | |||
+ | '''Proposal for alternative solution in (1d)''' | ||
+ | |||
+ | In (1b), we showed that for \(f, g \in L^1(\mathbb{R}^n)\), the Fourier transform of the convolution is: | ||
+ | $$\widehat{f*g}(k) = \sqrt{2\pi}^n \hat f(k) \hat g(k)$$ | ||
+ | |||
+ | Here, we have \(n=1\) and can thus write: | ||
+ | $$\widehat{F_a*F_b}(k) = \sqrt{2\pi} \ \hat F_a(k) \hat F_b(k)$$ | ||
+ | |||
+ | We insert the Fourier transform of \(F_a\) and \(F_b\), which we found in (1c): | ||
+ | $$\begin{align} | ||
+ | \widehat{F_a*F_b}(k) &= \sqrt{2\pi} \ \left(\frac{1}{\sqrt{2a}} e^{-\frac{k^2}{4a}}\; \frac{1}{\sqrt{2b}} e^{-\frac{k^2}{4b}}\right) \\ | ||
+ | &= \sqrt{\frac{\pi}{2ab}}\exp\left(-\frac{k^2}{4}\left(\frac{1}{a}+\frac{1}{b}\right)\right) \\ | ||
+ | &= \sqrt{\frac{\pi}{a+b}} \sqrt{\frac{1}{2\frac{ab}{a+b}}} \exp\left(-\frac{k^2}{4}\frac{a+b}{ab}\right) \\ | ||
+ | &= \alpha \hat F_c(k) | ||
+ | \end{align}$$ | ||
+ | for \(c=\frac{ab}{a+b}\) and \(\alpha = \sqrt{\frac{\pi}{a+b}}\) | ||
+ | |||
+ | Because of the linearity of the Fourier transform, we can conclude: | ||
+ | $$(F_a*F_b)(x) = \alpha F_c(x)$$ | ||
+ | |||
+ | [[User:Nik|Nik]] ([[User talk:Nik|talk]]) 14:38, 14 January 2015 (CET) |
Revision as of 13:38, 14 January 2015
Instead of doing the substitution it should also be possible to write:
$$ \frac{1}{(2\pi)^{n/2}} \int_{\mathbb{R}^n } \int_{\mathbb{R}^n } f(x-y) g(y) e^{-i<k,x>} dy \: dx $$
as
$$ \frac{1}{(2\pi)^{n/2}} \int_{\mathbb{R}^n } \int_{\mathbb{R}^n } f(x-y) g(y) e^{-i<k,x-y+y>} dy \: dx $$
$$ = \frac{1}{(2\pi)^{n/2}} \int_{\mathbb{R}^n } \int_{\mathbb{R}^n } f(x-y) e^{-i<k,x-y>} g(y) e^{-i<k,y>} dy \: dx $$
and then use Fubini
Un vieil homme (talk) 20:43, 31 December 2014 (CET)
In 1c)the derivation got exchanged with the integral without proofing that this is allowed:
$$ \frac{d}{dk} \widehat{F_c}(k) = \frac{d}{dk} (\frac{1}{\sqrt{2\pi}} \int_{-\infty}^{\infty} F_c(x)e^{-ikx} dx) = \frac{1}{\sqrt{2\pi}} \int_{-\infty}^{\infty} F_c(x) \frac{d}{dk} e^{-ikx} dx = \frac{-i}{\sqrt{2\pi}} \int_{-\infty}^{\infty} e^{-cx^{2}}xe^{-ikx} dx $$
Alex (talk) 14:55, 4 January 2015 (CET)
In part d) of the problem used \( e^{-ax^{2}}\sqrt{2\pi} \widehat{F_l}(s) \) with \( s = 2axi \). But a Fourier-transform can't take complex arguments - I didn't know that either so I checked here: [1] So you might rather use the trick we've used in problem 3 with completing the square. Actually you can copy that from the guys of group Y on the Dropbox, they got a pretty solid solution. - A.
@Un vieil homme: I think the Fubini idea doesn't work. We've proven Fubini at Kowalski's for Riemann-integrable functions so I guess that's the theorem we can use. Fubini only works for compact intervals which is not the case above. - A.
Proposal for alternative solution in (1d)
In (1b), we showed that for \(f, g \in L^1(\mathbb{R}^n)\), the Fourier transform of the convolution is: $$\widehat{f*g}(k) = \sqrt{2\pi}^n \hat f(k) \hat g(k)$$
Here, we have \(n=1\) and can thus write: $$\widehat{F_a*F_b}(k) = \sqrt{2\pi} \ \hat F_a(k) \hat F_b(k)$$
We insert the Fourier transform of \(F_a\) and \(F_b\), which we found in (1c): $$\begin{align} \widehat{F_a*F_b}(k) &= \sqrt{2\pi} \ \left(\frac{1}{\sqrt{2a}} e^{-\frac{k^2}{4a}}\; \frac{1}{\sqrt{2b}} e^{-\frac{k^2}{4b}}\right) \\ &= \sqrt{\frac{\pi}{2ab}}\exp\left(-\frac{k^2}{4}\left(\frac{1}{a}+\frac{1}{b}\right)\right) \\ &= \sqrt{\frac{\pi}{a+b}} \sqrt{\frac{1}{2\frac{ab}{a+b}}} \exp\left(-\frac{k^2}{4}\frac{a+b}{ab}\right) \\ &= \alpha \hat F_c(k) \end{align}$$ for \(c=\frac{ab}{a+b}\) and \(\alpha = \sqrt{\frac{\pi}{a+b}}\)
Because of the linearity of the Fourier transform, we can conclude: $$(F_a*F_b)(x) = \alpha F_c(x)$$