Group Abstract

Message Boards

WOLFRAM COMMUNITY

6.8K Views

3 Replies

4 Total Likes

View groups...

Follow this post

Share this post:

GROUPS:

Speed up evaluation of this integral?

tom ri

Posted 7 years ago

I am trying to evaluate the integral $$ \int_0^{2\pi}\frac{D_{11}^2+2D_{12}^2+D_{22}^2+D_{33}^2}{1+e\cos x(t)}dx $$ Where $D_{ij}$ are third derivatives wrt $t$ as defined below. a[psi_] := k1/(1 + eCos[psi]); d[psi_] := p1/(1 + eCos[psi]); D11 = D[a[x[t]](Cos[x[t]]^2 - 1/3), {t, 3}]; D12 = D[a[x[t]]Cos[x[t]]Sin[x[t]], {t, 3}]; D22 = D[a[x[t]](Sin[x[t]]^2 - 1/3), {t, 3}]; D33 = D[-a[x[t]]/3]; To do this I have written the input Assuming[Element[k1, Reals] && Element[e, Reals] && x'[t] == p2/(d[x[t]]^2) && Element[p2, Reals] && Element[p2, Reals], Integrate[(D11^2 + 2D12^2 + D22^2 + D33^2)/(1 + eCos[x[t]])^2, {x[ t], 0, 2Pi}]] This is taking a very long time to evaluate (has been running for hours), is it OK to specify to Mathematica* that I have the condition $\dot{x}=\frac{p_2}{d(x(t))^2}$ in this manner? How could this be sped up? This integral arose in the study of the loss of energy of an orbiting mass due to gravitational radiation.

POSTED BY: tom ri

3 Replies

Sort By:

Alexander Trounev

Alexander Trounev, A&E Trounev IT Consulting

Posted 7 years ago

It takes less than a minute. But first you need to simplify the expression a[psi_] := k1/(1 + eCos[psi]); d[psi_] := p1/(1 + eCos[psi]); D11 = D[a[x[t]](Cos[x[t]]^2 - 1/3), {t, 3}]; D12 = D[a[x[t]]Cos[x[t]]Sin[x[t]], {t, 3}]; D22 = D[a[x[t]](Sin[x[t]]^2 - 1/3), {t, 3}]; D33 = D[-a[x[t]]/3, {t,3}]; x'[t] = p2/d[x[t]]^2 x''[t] = D[x'[t], t] x'''[t] = D[x'[t], t, t] Et = (D11^2 + 2D12^2 + D22^2 + D33^2)/(1 + eCos[x[t]])^2 //FullSimplify Integrate[Et, {x[t], 0, 2*Pi}] //AbsoluteTiming {2.57266, ((1536 + 20320 e^2 + 35616 e^4 + 10988 e^6 + 315 e^8) k1^2 p2^6 \[Pi])/(24 p1^12)}

It takes less than a minute. But first you need to simplify the expression

a[psi_] := k1/(1 + e*Cos[psi]);
d[psi_] := p1/(1 + e*Cos[psi]);
D11 = D[a[x[t]]*(Cos[x[t]]^2 - 1/3), {t, 3}];
D12 = D[a[x[t]]*Cos[x[t]]*Sin[x[t]], {t, 3}];
D22 = D[a[x[t]]*(Sin[x[t]]^2 - 1/3), {t, 3}];
D33 = D[-a[x[t]]/3, {t,3}];


x'[t] = p2/d[x[t]]^2

x''[t] = D[x'[t], t]

x'''[t] = D[x'[t], t, t]

Et = (D11^2 + 2*D12^2 + D22^2 + D33^2)/(1 + e*Cos[x[t]])^2 //FullSimplify
Integrate[Et, {x[t], 0, 2*Pi}] //AbsoluteTiming
{2.57266, ((1536 + 20320 e^2 + 35616 e^4 + 10988 e^6 + 
        315 e^8) k1^2 p2^6 \[Pi])/(24 p1^12)}

POSTED BY: Alexander Trounev

Tim Laska

Tim Laska, A Priori, Ltd

Posted 7 years ago

Very insightful. There is a problem that the OP's numerator is not dimensionally consistent. The D33 term is the partial derivative with respect to nothing, which bothered me. Presuming that D33 should also be the third derivative with respect to t, your approach is even faster and contains no conditional expression. a[psi_] := k1/(1 + eCos[psi]); d[psi_] := p1/(1 + eCos[psi]); D11 = D[a[x[t]](Cos[x[t]]^2 - 1/3), {t, 3}]; D12 = D[a[x[t]]Cos[x[t]]Sin[x[t]], {t, 3}]; D22 = D[a[x[t]](Sin[x[t]]^2 - 1/3), {t, 3}]; (D33=D[-a[x[t]]/3])(* From the OP ) D33 = D[-a[x[t]]/3, {t, 3}]; x'[t] = p2/d[x[t]]^2; x''[t] = D[x'[t], t]; x'''[t] = D[x'[t], t, t]; Et = (D11^2 + 2D12^2 + D22^2 + D33^2)/(1 + eCos[x[t]])^2 // FullSimplify Integrate[Et, {x[t], 0, 2Pi}] // AbsoluteTiming (* {2.819966958116478`,((1536+20320 e^2+35616 e^4+10988 e^6+315 e^8) \ k1^2 p2^6 \[Pi])/(24 p1^12)} *)

Very insightful. There is a problem that the OP's numerator is not dimensionally consistent. The D33 term is the partial derivative with respect to nothing, which bothered me. Presuming that D33 should also be the third derivative with respect to t, your approach is even faster and contains no conditional expression.

a[psi_] := k1/(1 + e*Cos[psi]);
d[psi_] := p1/(1 + e*Cos[psi]);
D11 = D[a[x[t]]*(Cos[x[t]]^2 - 1/3), {t, 3}];
D12 = D[a[x[t]]*Cos[x[t]]*Sin[x[t]], {t, 3}];
D22 = D[a[x[t]]*(Sin[x[t]]^2 - 1/3), {t, 3}];
(*D33=D[-a[x[t]]/3]*)(* From the OP *)
D33 = D[-a[x[t]]/3, {t, 3}];

x'[t] = p2/d[x[t]]^2;

x''[t] = D[x'[t], t];

x'''[t] = D[x'[t], t, t];

Et = (D11^2 + 2*D12^2 + D22^2 + D33^2)/(1 + e*Cos[x[t]])^2 // 
  FullSimplify
Integrate[Et, {x[t], 0, 2*Pi}] // AbsoluteTiming

(* {2.819966958116478`,((1536+20320 e^2+35616 e^4+10988 e^6+315 e^8) \
k1^2 p2^6 \[Pi])/(24 p1^12)} *)

POSTED BY: Tim Laska

Alexander Trounev

Alexander Trounev, A&E Trounev IT Consulting

Posted 7 years ago

Thank you, I did not even notice the error in the definition of D3. After correcting the code, the result has changed. I also changed my post.

POSTED BY: Alexander Trounev

Reply to this discussion

Reply Preview

Attachments

Remove Add a file to this post

Follow this discussion

or Discard

Feedback