Fourier transform of “hyperbolically distorted” Gaussian / Bessel-type integrals
up vote
2
down vote
favorite
Dear Math enthusiasts,
I'm trying to see if I can find an analytical expression for the Fourier transform of a Gaussian pulse $p(tau) = {rm e}^{-Btau^2}$ that is distorted by a hyperbolic distortion of the form $tau(t) = sqrt{t^2+a^2}-a$. What I mean is I look at the function $p(tau(t))$ and I try to transform this time $t$ to frequency.
This gives a Fourier integral of the following form $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} {rm e}^{-jmath omega t} {rm d}t.$$
Substituting the square-root doesn't help much, since it appears back when resubstituting $t$. I therefore tried to substitute $t = a sinh(x)$ instead. Assuming I didn't make a stupid mistake, this gives $$G(omega) = aint_{-infty}^infty {rm e}^{-Bleft(a cosh(x)-aright)^2} {rm e}^{-jmath omega a sinh(x)} cosh(x) {rm d}x.$$
From [*], eqn (2.3) I know that $int_{0}^infty {rm e}^{a cosh(x)} cosh(x) {rm d}x = K_0(a)$ where $K_0(a)$ is the modified Bessel function of the second kind. As the integrand is symmetric, it should be no problem to extend this to $(-infty,infty)$. But still I'm not sure this is enough to solve the integral. Mathematica has refused to give me anything, but it's possible I used it in a wrong way. I tried manipulating the exponent to bring everything to one hyperbolic trig function, but I failed.
Any hints how I can proceed?
edit: Note that since $p(tau(t))$ is even symmetric, $G(omega)$ is real-valued and even and we can get rid of complex numbers altogether, i.e., $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t = 2 int_{0}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t.$$ Does this make it easier? I'm not sure.
integration fourier-transform bessel-functions
This question has an open bounty worth +50
reputation from Florian ending tomorrow.
This question has not received enough attention.
add a comment |
up vote
2
down vote
favorite
Dear Math enthusiasts,
I'm trying to see if I can find an analytical expression for the Fourier transform of a Gaussian pulse $p(tau) = {rm e}^{-Btau^2}$ that is distorted by a hyperbolic distortion of the form $tau(t) = sqrt{t^2+a^2}-a$. What I mean is I look at the function $p(tau(t))$ and I try to transform this time $t$ to frequency.
This gives a Fourier integral of the following form $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} {rm e}^{-jmath omega t} {rm d}t.$$
Substituting the square-root doesn't help much, since it appears back when resubstituting $t$. I therefore tried to substitute $t = a sinh(x)$ instead. Assuming I didn't make a stupid mistake, this gives $$G(omega) = aint_{-infty}^infty {rm e}^{-Bleft(a cosh(x)-aright)^2} {rm e}^{-jmath omega a sinh(x)} cosh(x) {rm d}x.$$
From [*], eqn (2.3) I know that $int_{0}^infty {rm e}^{a cosh(x)} cosh(x) {rm d}x = K_0(a)$ where $K_0(a)$ is the modified Bessel function of the second kind. As the integrand is symmetric, it should be no problem to extend this to $(-infty,infty)$. But still I'm not sure this is enough to solve the integral. Mathematica has refused to give me anything, but it's possible I used it in a wrong way. I tried manipulating the exponent to bring everything to one hyperbolic trig function, but I failed.
Any hints how I can proceed?
edit: Note that since $p(tau(t))$ is even symmetric, $G(omega)$ is real-valued and even and we can get rid of complex numbers altogether, i.e., $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t = 2 int_{0}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t.$$ Does this make it easier? I'm not sure.
integration fourier-transform bessel-functions
This question has an open bounty worth +50
reputation from Florian ending tomorrow.
This question has not received enough attention.
add a comment |
up vote
2
down vote
favorite
up vote
2
down vote
favorite
Dear Math enthusiasts,
I'm trying to see if I can find an analytical expression for the Fourier transform of a Gaussian pulse $p(tau) = {rm e}^{-Btau^2}$ that is distorted by a hyperbolic distortion of the form $tau(t) = sqrt{t^2+a^2}-a$. What I mean is I look at the function $p(tau(t))$ and I try to transform this time $t$ to frequency.
This gives a Fourier integral of the following form $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} {rm e}^{-jmath omega t} {rm d}t.$$
Substituting the square-root doesn't help much, since it appears back when resubstituting $t$. I therefore tried to substitute $t = a sinh(x)$ instead. Assuming I didn't make a stupid mistake, this gives $$G(omega) = aint_{-infty}^infty {rm e}^{-Bleft(a cosh(x)-aright)^2} {rm e}^{-jmath omega a sinh(x)} cosh(x) {rm d}x.$$
From [*], eqn (2.3) I know that $int_{0}^infty {rm e}^{a cosh(x)} cosh(x) {rm d}x = K_0(a)$ where $K_0(a)$ is the modified Bessel function of the second kind. As the integrand is symmetric, it should be no problem to extend this to $(-infty,infty)$. But still I'm not sure this is enough to solve the integral. Mathematica has refused to give me anything, but it's possible I used it in a wrong way. I tried manipulating the exponent to bring everything to one hyperbolic trig function, but I failed.
Any hints how I can proceed?
edit: Note that since $p(tau(t))$ is even symmetric, $G(omega)$ is real-valued and even and we can get rid of complex numbers altogether, i.e., $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t = 2 int_{0}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t.$$ Does this make it easier? I'm not sure.
integration fourier-transform bessel-functions
Dear Math enthusiasts,
I'm trying to see if I can find an analytical expression for the Fourier transform of a Gaussian pulse $p(tau) = {rm e}^{-Btau^2}$ that is distorted by a hyperbolic distortion of the form $tau(t) = sqrt{t^2+a^2}-a$. What I mean is I look at the function $p(tau(t))$ and I try to transform this time $t$ to frequency.
This gives a Fourier integral of the following form $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} {rm e}^{-jmath omega t} {rm d}t.$$
Substituting the square-root doesn't help much, since it appears back when resubstituting $t$. I therefore tried to substitute $t = a sinh(x)$ instead. Assuming I didn't make a stupid mistake, this gives $$G(omega) = aint_{-infty}^infty {rm e}^{-Bleft(a cosh(x)-aright)^2} {rm e}^{-jmath omega a sinh(x)} cosh(x) {rm d}x.$$
From [*], eqn (2.3) I know that $int_{0}^infty {rm e}^{a cosh(x)} cosh(x) {rm d}x = K_0(a)$ where $K_0(a)$ is the modified Bessel function of the second kind. As the integrand is symmetric, it should be no problem to extend this to $(-infty,infty)$. But still I'm not sure this is enough to solve the integral. Mathematica has refused to give me anything, but it's possible I used it in a wrong way. I tried manipulating the exponent to bring everything to one hyperbolic trig function, but I failed.
Any hints how I can proceed?
edit: Note that since $p(tau(t))$ is even symmetric, $G(omega)$ is real-valued and even and we can get rid of complex numbers altogether, i.e., $$G(omega) = int_{-infty}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t = 2 int_{0}^infty {rm e}^{-Bleft(sqrt{t^2+a^2}-aright)^2} cos( omega t) {rm d}t.$$ Does this make it easier? I'm not sure.
integration fourier-transform bessel-functions
integration fourier-transform bessel-functions
edited Nov 22 at 9:58
asked Nov 16 at 12:30
Florian
1,2981720
1,2981720
This question has an open bounty worth +50
reputation from Florian ending tomorrow.
This question has not received enough attention.
This question has an open bounty worth +50
reputation from Florian ending tomorrow.
This question has not received enough attention.
add a comment |
add a comment |
1 Answer
1
active
oldest
votes
up vote
1
down vote
HINT
Firstly,
$$begin{aligned}
&G(omega,a,B)=aintlimits_{-infty}^infty e^{-a^2Bleft(sqrt{left(frac taright)^2+1}-1right)^2}e^{-jaomegafrac ta},mathrm ddfrac ta=aintlimits_{-infty}^infty e^{-a^2B(sqrt{t^2+1}-1)^2}e^{-jaomega t},mathrm dt,\
&G(omega,a,B)=aH(aomega,a^2B),\
end{aligned}$$
$$H(Omega,c)=intlimits_{-infty}^infty e^{-c(sqrt{t^2+1}-1)^2}e^{-jOmega t},mathrm dt.tag1$$
Let
$$tau(t) = sqrt{t^2+1}-1,quad tauin[0,infty),$$
so
$$t(tau)=pmsqrt{(tau+1)^2-1}=pmsqrt{tau^2+2tau},quad tin(-infty,infty),$$
$$dt=pmdfrac{tau+1}{sqrt{tau^2+2tau}},$$
$$begin{aligned}
&H(Omega,c)=intlimits_{-infty}^{-1}e^{-ctau^2(t)}e^{-jOmega t},mathrm dt
+intlimits_{-1}^infty e^{-ctau^2(t)}e^{-jOmega t},mathrm dt\
&=-intlimits_{-infty}^0e^{-ctau^2}e^{jOmegasqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau
+intlimits_{-1}^infty e^{-ctau^2}e^{-jOmega sqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau,
end{aligned}$$
$$H(Omega,c)=2intlimits_0^{infty} e^{-ctau^2} cos(Omegasqrt{tau^2+2tau})dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau.tag2$$
By parts:
$$begin{aligned}
&H(omega,c) = dfrac2Omegaintlimits_0^{infty} e^{-ctau^2},dsin(Omegasqrt{tau^2+2tau})\
&=-dfrac2Omega e^{-ctau^2} sin(Omegasqrt{tau^2+2tau})bigg|_0^infty
+dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}
sin(Omegasqrt{tau^2+2tau}),dtau,end{aligned}$$
$$H(Omega,c)=dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}sin(Omegasqrt{tau^2+2tau}),dtau.tag3$$
I don't see how to obtain the closed form of the integral $(3).$
At the same time, for $c<<Omega$ the exponent in the formula $(1)$ allows polynomial approximation.
If $cgtrsimOmega,$ then the integral $(3)$ can be presented as $1D$ series, using the series expansion for sine expression in the form of
$$sin(Omegasqrt{x^2+2x}) = sqrt{dfrac x2}Omegaleft(2 + dfrac{3-4Omega^2}6 x + dfrac{16 Omega^4 - 120 Omega^2 - 15}{240} x^2 + dfrac{-64Omega^6 + 1680Omega^4 - 1260Omega^2 + 315}{20160} x^3 + dfrac{256Omega^8 - 16128Omega^6 + 90720Omega^4 + 15120Omega^2 - 14175}{2903040} x^4 + dfrac{-1024Omega^{10} + 126720Omega^8 - 2217600Omega^6 + 1663200Omega^4 - 623700Omega^2 + 1091475}{638668800}x^5 + dotsright).
$$
(the asymptotic form of
$$sin(Omegasqrt{x^2+2x}) = Omega{cos((x+1)Omega)left(-dfrac1{2x} + dfrac 1{2x^2} + dfrac{Omega^2-30}{48x^3} - dfrac{Omega^2 - 14}{16x^4} - dfrac{Omega^4 - 540 Omega^2 + 5040}{3840x^5} + dfrac{Omega^4 - 220 Omega^2 + 1584}{768x^6} + dotsright)} + {sin((x+1)Omega)left(1 - dfrac{Omega^2}{8x^2} + dfrac{Omega^2}{4x^3} + dfrac{Omega^2(Omega^2-168)}{384 x^4} - dfrac{Omega^2(Omega^2-72)}{96 x^5} - dfrac{Omega^2(Omega^4 - 1320 Omega^2 + 59400)}{46080 x^6} + dfrac{Omega^2(Omega^4 - 520 Omega^2 + 17160)}{7680 x^7} + dotsright)}
$$ is not so usable)
and the integral
$$intlimits_0^infty x^d e^{-cx^2},mathrm dx = dfrac{Gammaleft(frac{d+1}2right)}{2c^{frac{d+1}2}}.$$
Length of the series can be selected, using the numeric experiments.
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
add a comment |
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
up vote
1
down vote
HINT
Firstly,
$$begin{aligned}
&G(omega,a,B)=aintlimits_{-infty}^infty e^{-a^2Bleft(sqrt{left(frac taright)^2+1}-1right)^2}e^{-jaomegafrac ta},mathrm ddfrac ta=aintlimits_{-infty}^infty e^{-a^2B(sqrt{t^2+1}-1)^2}e^{-jaomega t},mathrm dt,\
&G(omega,a,B)=aH(aomega,a^2B),\
end{aligned}$$
$$H(Omega,c)=intlimits_{-infty}^infty e^{-c(sqrt{t^2+1}-1)^2}e^{-jOmega t},mathrm dt.tag1$$
Let
$$tau(t) = sqrt{t^2+1}-1,quad tauin[0,infty),$$
so
$$t(tau)=pmsqrt{(tau+1)^2-1}=pmsqrt{tau^2+2tau},quad tin(-infty,infty),$$
$$dt=pmdfrac{tau+1}{sqrt{tau^2+2tau}},$$
$$begin{aligned}
&H(Omega,c)=intlimits_{-infty}^{-1}e^{-ctau^2(t)}e^{-jOmega t},mathrm dt
+intlimits_{-1}^infty e^{-ctau^2(t)}e^{-jOmega t},mathrm dt\
&=-intlimits_{-infty}^0e^{-ctau^2}e^{jOmegasqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau
+intlimits_{-1}^infty e^{-ctau^2}e^{-jOmega sqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau,
end{aligned}$$
$$H(Omega,c)=2intlimits_0^{infty} e^{-ctau^2} cos(Omegasqrt{tau^2+2tau})dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau.tag2$$
By parts:
$$begin{aligned}
&H(omega,c) = dfrac2Omegaintlimits_0^{infty} e^{-ctau^2},dsin(Omegasqrt{tau^2+2tau})\
&=-dfrac2Omega e^{-ctau^2} sin(Omegasqrt{tau^2+2tau})bigg|_0^infty
+dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}
sin(Omegasqrt{tau^2+2tau}),dtau,end{aligned}$$
$$H(Omega,c)=dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}sin(Omegasqrt{tau^2+2tau}),dtau.tag3$$
I don't see how to obtain the closed form of the integral $(3).$
At the same time, for $c<<Omega$ the exponent in the formula $(1)$ allows polynomial approximation.
If $cgtrsimOmega,$ then the integral $(3)$ can be presented as $1D$ series, using the series expansion for sine expression in the form of
$$sin(Omegasqrt{x^2+2x}) = sqrt{dfrac x2}Omegaleft(2 + dfrac{3-4Omega^2}6 x + dfrac{16 Omega^4 - 120 Omega^2 - 15}{240} x^2 + dfrac{-64Omega^6 + 1680Omega^4 - 1260Omega^2 + 315}{20160} x^3 + dfrac{256Omega^8 - 16128Omega^6 + 90720Omega^4 + 15120Omega^2 - 14175}{2903040} x^4 + dfrac{-1024Omega^{10} + 126720Omega^8 - 2217600Omega^6 + 1663200Omega^4 - 623700Omega^2 + 1091475}{638668800}x^5 + dotsright).
$$
(the asymptotic form of
$$sin(Omegasqrt{x^2+2x}) = Omega{cos((x+1)Omega)left(-dfrac1{2x} + dfrac 1{2x^2} + dfrac{Omega^2-30}{48x^3} - dfrac{Omega^2 - 14}{16x^4} - dfrac{Omega^4 - 540 Omega^2 + 5040}{3840x^5} + dfrac{Omega^4 - 220 Omega^2 + 1584}{768x^6} + dotsright)} + {sin((x+1)Omega)left(1 - dfrac{Omega^2}{8x^2} + dfrac{Omega^2}{4x^3} + dfrac{Omega^2(Omega^2-168)}{384 x^4} - dfrac{Omega^2(Omega^2-72)}{96 x^5} - dfrac{Omega^2(Omega^4 - 1320 Omega^2 + 59400)}{46080 x^6} + dfrac{Omega^2(Omega^4 - 520 Omega^2 + 17160)}{7680 x^7} + dotsright)}
$$ is not so usable)
and the integral
$$intlimits_0^infty x^d e^{-cx^2},mathrm dx = dfrac{Gammaleft(frac{d+1}2right)}{2c^{frac{d+1}2}}.$$
Length of the series can be selected, using the numeric experiments.
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
add a comment |
up vote
1
down vote
HINT
Firstly,
$$begin{aligned}
&G(omega,a,B)=aintlimits_{-infty}^infty e^{-a^2Bleft(sqrt{left(frac taright)^2+1}-1right)^2}e^{-jaomegafrac ta},mathrm ddfrac ta=aintlimits_{-infty}^infty e^{-a^2B(sqrt{t^2+1}-1)^2}e^{-jaomega t},mathrm dt,\
&G(omega,a,B)=aH(aomega,a^2B),\
end{aligned}$$
$$H(Omega,c)=intlimits_{-infty}^infty e^{-c(sqrt{t^2+1}-1)^2}e^{-jOmega t},mathrm dt.tag1$$
Let
$$tau(t) = sqrt{t^2+1}-1,quad tauin[0,infty),$$
so
$$t(tau)=pmsqrt{(tau+1)^2-1}=pmsqrt{tau^2+2tau},quad tin(-infty,infty),$$
$$dt=pmdfrac{tau+1}{sqrt{tau^2+2tau}},$$
$$begin{aligned}
&H(Omega,c)=intlimits_{-infty}^{-1}e^{-ctau^2(t)}e^{-jOmega t},mathrm dt
+intlimits_{-1}^infty e^{-ctau^2(t)}e^{-jOmega t},mathrm dt\
&=-intlimits_{-infty}^0e^{-ctau^2}e^{jOmegasqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau
+intlimits_{-1}^infty e^{-ctau^2}e^{-jOmega sqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau,
end{aligned}$$
$$H(Omega,c)=2intlimits_0^{infty} e^{-ctau^2} cos(Omegasqrt{tau^2+2tau})dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau.tag2$$
By parts:
$$begin{aligned}
&H(omega,c) = dfrac2Omegaintlimits_0^{infty} e^{-ctau^2},dsin(Omegasqrt{tau^2+2tau})\
&=-dfrac2Omega e^{-ctau^2} sin(Omegasqrt{tau^2+2tau})bigg|_0^infty
+dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}
sin(Omegasqrt{tau^2+2tau}),dtau,end{aligned}$$
$$H(Omega,c)=dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}sin(Omegasqrt{tau^2+2tau}),dtau.tag3$$
I don't see how to obtain the closed form of the integral $(3).$
At the same time, for $c<<Omega$ the exponent in the formula $(1)$ allows polynomial approximation.
If $cgtrsimOmega,$ then the integral $(3)$ can be presented as $1D$ series, using the series expansion for sine expression in the form of
$$sin(Omegasqrt{x^2+2x}) = sqrt{dfrac x2}Omegaleft(2 + dfrac{3-4Omega^2}6 x + dfrac{16 Omega^4 - 120 Omega^2 - 15}{240} x^2 + dfrac{-64Omega^6 + 1680Omega^4 - 1260Omega^2 + 315}{20160} x^3 + dfrac{256Omega^8 - 16128Omega^6 + 90720Omega^4 + 15120Omega^2 - 14175}{2903040} x^4 + dfrac{-1024Omega^{10} + 126720Omega^8 - 2217600Omega^6 + 1663200Omega^4 - 623700Omega^2 + 1091475}{638668800}x^5 + dotsright).
$$
(the asymptotic form of
$$sin(Omegasqrt{x^2+2x}) = Omega{cos((x+1)Omega)left(-dfrac1{2x} + dfrac 1{2x^2} + dfrac{Omega^2-30}{48x^3} - dfrac{Omega^2 - 14}{16x^4} - dfrac{Omega^4 - 540 Omega^2 + 5040}{3840x^5} + dfrac{Omega^4 - 220 Omega^2 + 1584}{768x^6} + dotsright)} + {sin((x+1)Omega)left(1 - dfrac{Omega^2}{8x^2} + dfrac{Omega^2}{4x^3} + dfrac{Omega^2(Omega^2-168)}{384 x^4} - dfrac{Omega^2(Omega^2-72)}{96 x^5} - dfrac{Omega^2(Omega^4 - 1320 Omega^2 + 59400)}{46080 x^6} + dfrac{Omega^2(Omega^4 - 520 Omega^2 + 17160)}{7680 x^7} + dotsright)}
$$ is not so usable)
and the integral
$$intlimits_0^infty x^d e^{-cx^2},mathrm dx = dfrac{Gammaleft(frac{d+1}2right)}{2c^{frac{d+1}2}}.$$
Length of the series can be selected, using the numeric experiments.
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
add a comment |
up vote
1
down vote
up vote
1
down vote
HINT
Firstly,
$$begin{aligned}
&G(omega,a,B)=aintlimits_{-infty}^infty e^{-a^2Bleft(sqrt{left(frac taright)^2+1}-1right)^2}e^{-jaomegafrac ta},mathrm ddfrac ta=aintlimits_{-infty}^infty e^{-a^2B(sqrt{t^2+1}-1)^2}e^{-jaomega t},mathrm dt,\
&G(omega,a,B)=aH(aomega,a^2B),\
end{aligned}$$
$$H(Omega,c)=intlimits_{-infty}^infty e^{-c(sqrt{t^2+1}-1)^2}e^{-jOmega t},mathrm dt.tag1$$
Let
$$tau(t) = sqrt{t^2+1}-1,quad tauin[0,infty),$$
so
$$t(tau)=pmsqrt{(tau+1)^2-1}=pmsqrt{tau^2+2tau},quad tin(-infty,infty),$$
$$dt=pmdfrac{tau+1}{sqrt{tau^2+2tau}},$$
$$begin{aligned}
&H(Omega,c)=intlimits_{-infty}^{-1}e^{-ctau^2(t)}e^{-jOmega t},mathrm dt
+intlimits_{-1}^infty e^{-ctau^2(t)}e^{-jOmega t},mathrm dt\
&=-intlimits_{-infty}^0e^{-ctau^2}e^{jOmegasqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau
+intlimits_{-1}^infty e^{-ctau^2}e^{-jOmega sqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau,
end{aligned}$$
$$H(Omega,c)=2intlimits_0^{infty} e^{-ctau^2} cos(Omegasqrt{tau^2+2tau})dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau.tag2$$
By parts:
$$begin{aligned}
&H(omega,c) = dfrac2Omegaintlimits_0^{infty} e^{-ctau^2},dsin(Omegasqrt{tau^2+2tau})\
&=-dfrac2Omega e^{-ctau^2} sin(Omegasqrt{tau^2+2tau})bigg|_0^infty
+dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}
sin(Omegasqrt{tau^2+2tau}),dtau,end{aligned}$$
$$H(Omega,c)=dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}sin(Omegasqrt{tau^2+2tau}),dtau.tag3$$
I don't see how to obtain the closed form of the integral $(3).$
At the same time, for $c<<Omega$ the exponent in the formula $(1)$ allows polynomial approximation.
If $cgtrsimOmega,$ then the integral $(3)$ can be presented as $1D$ series, using the series expansion for sine expression in the form of
$$sin(Omegasqrt{x^2+2x}) = sqrt{dfrac x2}Omegaleft(2 + dfrac{3-4Omega^2}6 x + dfrac{16 Omega^4 - 120 Omega^2 - 15}{240} x^2 + dfrac{-64Omega^6 + 1680Omega^4 - 1260Omega^2 + 315}{20160} x^3 + dfrac{256Omega^8 - 16128Omega^6 + 90720Omega^4 + 15120Omega^2 - 14175}{2903040} x^4 + dfrac{-1024Omega^{10} + 126720Omega^8 - 2217600Omega^6 + 1663200Omega^4 - 623700Omega^2 + 1091475}{638668800}x^5 + dotsright).
$$
(the asymptotic form of
$$sin(Omegasqrt{x^2+2x}) = Omega{cos((x+1)Omega)left(-dfrac1{2x} + dfrac 1{2x^2} + dfrac{Omega^2-30}{48x^3} - dfrac{Omega^2 - 14}{16x^4} - dfrac{Omega^4 - 540 Omega^2 + 5040}{3840x^5} + dfrac{Omega^4 - 220 Omega^2 + 1584}{768x^6} + dotsright)} + {sin((x+1)Omega)left(1 - dfrac{Omega^2}{8x^2} + dfrac{Omega^2}{4x^3} + dfrac{Omega^2(Omega^2-168)}{384 x^4} - dfrac{Omega^2(Omega^2-72)}{96 x^5} - dfrac{Omega^2(Omega^4 - 1320 Omega^2 + 59400)}{46080 x^6} + dfrac{Omega^2(Omega^4 - 520 Omega^2 + 17160)}{7680 x^7} + dotsright)}
$$ is not so usable)
and the integral
$$intlimits_0^infty x^d e^{-cx^2},mathrm dx = dfrac{Gammaleft(frac{d+1}2right)}{2c^{frac{d+1}2}}.$$
Length of the series can be selected, using the numeric experiments.
HINT
Firstly,
$$begin{aligned}
&G(omega,a,B)=aintlimits_{-infty}^infty e^{-a^2Bleft(sqrt{left(frac taright)^2+1}-1right)^2}e^{-jaomegafrac ta},mathrm ddfrac ta=aintlimits_{-infty}^infty e^{-a^2B(sqrt{t^2+1}-1)^2}e^{-jaomega t},mathrm dt,\
&G(omega,a,B)=aH(aomega,a^2B),\
end{aligned}$$
$$H(Omega,c)=intlimits_{-infty}^infty e^{-c(sqrt{t^2+1}-1)^2}e^{-jOmega t},mathrm dt.tag1$$
Let
$$tau(t) = sqrt{t^2+1}-1,quad tauin[0,infty),$$
so
$$t(tau)=pmsqrt{(tau+1)^2-1}=pmsqrt{tau^2+2tau},quad tin(-infty,infty),$$
$$dt=pmdfrac{tau+1}{sqrt{tau^2+2tau}},$$
$$begin{aligned}
&H(Omega,c)=intlimits_{-infty}^{-1}e^{-ctau^2(t)}e^{-jOmega t},mathrm dt
+intlimits_{-1}^infty e^{-ctau^2(t)}e^{-jOmega t},mathrm dt\
&=-intlimits_{-infty}^0e^{-ctau^2}e^{jOmegasqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau
+intlimits_{-1}^infty e^{-ctau^2}e^{-jOmega sqrt{tau^2+2tau}}dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau,
end{aligned}$$
$$H(Omega,c)=2intlimits_0^{infty} e^{-ctau^2} cos(Omegasqrt{tau^2+2tau})dfrac{tau+1}{sqrt{tau^2+2tau}},mathrm dtau.tag2$$
By parts:
$$begin{aligned}
&H(omega,c) = dfrac2Omegaintlimits_0^{infty} e^{-ctau^2},dsin(Omegasqrt{tau^2+2tau})\
&=-dfrac2Omega e^{-ctau^2} sin(Omegasqrt{tau^2+2tau})bigg|_0^infty
+dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}
sin(Omegasqrt{tau^2+2tau}),dtau,end{aligned}$$
$$H(Omega,c)=dfrac{4c}Omegaintlimits_0^{infty} tau e^{-ctau^2}sin(Omegasqrt{tau^2+2tau}),dtau.tag3$$
I don't see how to obtain the closed form of the integral $(3).$
At the same time, for $c<<Omega$ the exponent in the formula $(1)$ allows polynomial approximation.
If $cgtrsimOmega,$ then the integral $(3)$ can be presented as $1D$ series, using the series expansion for sine expression in the form of
$$sin(Omegasqrt{x^2+2x}) = sqrt{dfrac x2}Omegaleft(2 + dfrac{3-4Omega^2}6 x + dfrac{16 Omega^4 - 120 Omega^2 - 15}{240} x^2 + dfrac{-64Omega^6 + 1680Omega^4 - 1260Omega^2 + 315}{20160} x^3 + dfrac{256Omega^8 - 16128Omega^6 + 90720Omega^4 + 15120Omega^2 - 14175}{2903040} x^4 + dfrac{-1024Omega^{10} + 126720Omega^8 - 2217600Omega^6 + 1663200Omega^4 - 623700Omega^2 + 1091475}{638668800}x^5 + dotsright).
$$
(the asymptotic form of
$$sin(Omegasqrt{x^2+2x}) = Omega{cos((x+1)Omega)left(-dfrac1{2x} + dfrac 1{2x^2} + dfrac{Omega^2-30}{48x^3} - dfrac{Omega^2 - 14}{16x^4} - dfrac{Omega^4 - 540 Omega^2 + 5040}{3840x^5} + dfrac{Omega^4 - 220 Omega^2 + 1584}{768x^6} + dotsright)} + {sin((x+1)Omega)left(1 - dfrac{Omega^2}{8x^2} + dfrac{Omega^2}{4x^3} + dfrac{Omega^2(Omega^2-168)}{384 x^4} - dfrac{Omega^2(Omega^2-72)}{96 x^5} - dfrac{Omega^2(Omega^4 - 1320 Omega^2 + 59400)}{46080 x^6} + dfrac{Omega^2(Omega^4 - 520 Omega^2 + 17160)}{7680 x^7} + dotsright)}
$$ is not so usable)
and the integral
$$intlimits_0^infty x^d e^{-cx^2},mathrm dx = dfrac{Gammaleft(frac{d+1}2right)}{2c^{frac{d+1}2}}.$$
Length of the series can be selected, using the numeric experiments.
edited 8 hours ago
answered 15 hours ago
Yuri Negometyanov
9,3511524
9,3511524
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
add a comment |
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
Wow, thanks! Interesting approach. I would need to consider an infinite number of terms for the series since I'm integrating $tau$ to $infty$, right? If I understand well, I could express $H(Omega, c)$ via an infinite sum over $Gamma((d+1)/2)/c^{(d+1)/2}$ with $d$ going from 3/2, 5/2, 7/2, ... and coefficients worked out accordingly. Would I be able to truncate this series somewhere safely? Can we say something on how rapidly its terms would be decaying for $drightarrow infty$?
– Florian
13 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian No doubt, you all understand well. BTW, Wolfram Alpha allows to get more terms series for sine expression and to get any estimation of the coeffitients. Partially, for $Omega x >> 2$ the approximation must use the series for $Omega xsqrt{1+2/x}.$
– Yuri Negometyanov
11 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
@Florian And thank you for comments, asymptotic series added.
– Yuri Negometyanov
10 hours ago
add a comment |
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3001082%2ffourier-transform-of-hyperbolically-distorted-gaussian-bessel-type-integrals%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown