Chapter H Problem Sheet 8 (Lectures 16-17)

H.1 Dilation formula

For \(c>0\) and \(u\in L^1(\mR )\) show that

\[ \mathcal {F}\left (x\mapsto u(cx)\right )= \omega \mapsto \frac {1}{c}\left (\mathcal {F}(u)\right )\left (\frac {\omega }{c}\right ). \]

  • Solution. We have (using the substitution \(y=cx\))

    \[ \mathcal {F}\left (x\mapsto u(cx)\right ) =\frac {1}{\sqrt {2\pi }}\int _{-\infty }^\infty \e ^{-i\omega x}u(cx)\,dx =\frac {1}{\sqrt {2\pi }}\int _{-\infty }^\infty \e ^{-i\omega y/c}u(y)\,\frac {1}{c}\,dy =\frac {1}{c}\left (\mathcal {F}(u)\right )\left (\frac {\omega }{c}\right ). \]

H.2 Fourier transform of general Gaussian

Use Example 14.10 and Problem H.1 to obtain the formula in Remark 14.11.

  • Solution. From Example 14.10 we have that \(\mathcal {F}\left (x\mapsto \e ^{-x^2/2}\right )=\omega \mapsto \e ^{-\omega ^2/2}\). Applying Problem H.1 with \(c=\sqrt {a}\) gives

    \[ \mathcal {F}\left (x\mapsto \e ^{-(\sqrt {a} x)^2/2}\right )=\omega \mapsto \frac {1}{\sqrt {a}}\mapsto \e ^{-(\omega /\sqrt {a})^2/2}, \]

    which is

    \[ \mathcal {F}\left (x\mapsto \e ^{-ax^2/2}\right )=\omega \mapsto \frac {1}{\sqrt {a}}\mapsto \e ^{-\omega /(2a)}, \]

    as claimed in Remark 14.11.

H.3 Fourier transform of sine

Let \(a\in \mR \). Show that the Fourier (-Schwartz) transform of \(\sin (ax)\) equals \(\sqrt {2\pi }~\dfrac {\delta _a-\delta _{-a}}{2i}\).

  • Solution. We have

    \[ \sin (ax)=\frac {\e ^{iax}-\e ^{-iax}}{2i}, \]

    and using Example 17.17 (applied with \(a\) and \(-a\)) then gives

    \[ \mathcal {F}(x\mapsto \sin (ax))=\sqrt {2\pi }\frac {\delta _a-\delta _{-a}}{2i}. \]

H.4 Distributional dilation formula

For \(b>0\) and a function \(f\) define

\[ (D_bf)(x):=f(bx), \]

and for a distribution \(u\) define the distribution \(\tilde {D}_au\) by

\[ (\tilde {D}_au)(\varphi ):=\frac {1}{a}u\left (D_{1/a}\varphi \right ). \]

(a) Show that if \(u\) is the regular distribution corresponding to the locally integrable function \(f\), then \(\tilde {D}_au\) is the regular distribution corresponding the function \(D_af\).

(b) Show that we have

\[ \mathcal {F}\tilde {D}_au=\frac {1}{a}\tilde {D}_{1/a}\mathcal {F}u. \]

  • Solution. (a) We have for the regular distribution corresponding to the function \(D_af\)

    \[ T_{D_af}(\varphi ) =\int _\mR (D_af)(x)\varphi (x)\,dx =\int _\mR f(ax)\varphi (x)\,dx. \]

    Using the change of variables \(y:=ax\) gives

    \[ \int _\mR f(ax)\varphi (x)\,dx =\int _\mR f(y)\varphi (a^{-1}y)\,a^{-1}dy =\frac {1}{a}T_f\left (D_{1/a}\varphi \right ), \]

    where \(T_f\) is the regular distribution corresponding to the function \(f\). Therefore

    \[ T_{D_af}(\varphi )=\frac {1}{a}T_f\left (D_{1/a}\varphi \right ), \]

    and since this holds for all \(\varphi \in \mathcal {D}\) we have

    \[ T_{D_af}=\frac {1}{a}T_fD_{1/a}. \]

    From the definition of \(\tilde {D}_a\) we then have \(T_{D_af}=\tilde {D}_aT_f\) as desired.

    (b) We have (using the definition of Fourier transform and dilation for distributions):

    \[ \left (\mathcal {F}\tilde {D}_au\right )(\varphi ) =\tilde {D}_au\left (\hat {\varphi }\right ) =\frac {1}{a}u\left (D_{1/a}\hat {\varphi }\right ). \]

    Using the dilation formula for functions we have

    \[ \frac {1}{a}D_{1/a}\hat {\varphi }=\mathcal {F}D_a\varphi . \]

    Therefore

    \[ \left (\mathcal {F}\tilde {D}_au\right )(\varphi ) =u\left (\mathcal {F}D_a\varphi \right ). \]

    Using the definition of Fourier transform and dilation (with \(a\) replaced by \(1/a\)) for distributions we have

    \[ u\left (\mathcal {F}D_a\varphi \right ) =\left (\mathcal {F}u\right )\left (D_a\varphi \right ) =\frac {1}{a}\left (\tilde {D}_{1/a}\mathcal {F}u\right )(\varphi ), \]

    so that

    \[ \left (\mathcal {F}\tilde {D}_au\right )(\varphi ) =\frac {1}{a}\left (\tilde {D}_{1/a}\mathcal {F}u\right )(\varphi ), \]

    and since this holds for all \(\varphi \in \mathcal {S}\) we have

    \[ \mathcal {F}\tilde {D}_au=\frac {1}{a}\tilde {D}_{1/a}\mathcal {F}. \]