Consider the Fredholm Equation of the second kind,

$$\phi(x) = 3 + \lambda \int_{0}^{\pi} \text{cos}(x-s) \, \phi(s) \,ds$$

Where the analytical solution is found as,

$$\phi(x) = 3 + \frac{6\lambda}{1 - \lambda \frac{\pi}{2}}\,\text{sin}(x)$$

How could one use Mathematica to find a numerical solution to the same integral equation by using the method of successive approximations (i.e. the Neumann series approach)?


Use DSolve:

PHI = 
  DSolveValue[ϕ[x] == 3 + λ Integrate[ Cos[x - s] ϕ[s], {s, 0, Pi}], ϕ, x]
(*Function[{x}, (3 (-2 + π λ - 4 λ Sin[x]))/(-2 + π λ)]*)

The solution can be further used in the form PHI[x].

| improve this answer | |
  • $\begingroup$ Thank you, but how can I use the function Mathematica returns to, say, investigate the convergence of the new $\phi (x)$ function? $\endgroup$ – LightningStrike Mar 1 '19 at 19:40
  • 1
    $\begingroup$ @ user57401 I modified my answer! $\endgroup$ – Ulrich Neumann Mar 1 '19 at 19:57

Following Weisstein, Eric W. "Integral Equation Neumann Series." From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/IntegralEquationNeumannSeries.html, the Neumann series approximation is:

n = 10;  (* for example *)
ϕ[x_, 0] = 3;
Do[ϕ[x_, j_] = 3 + λ Integrate[Cos[x - p] ϕ[p, j - 1], {p, 0, π}], {j, n}]

The last term in the series ϕ[x,n] is the approximation to ϕ[x].

Here is what Mathematica returns for ϕ[x,10].


To investigate convergence, I guess we could look at the difference ϕ[x,n] - ϕ[x] as n gets large, since you know ϕ[x].

| improve this answer | |
  • $\begingroup$ Thank you! When I try to run this, my output is returning the value of 3? How did you get Mathematica to return the series above for [Phi][10]? $\endgroup$ – LightningStrike Mar 1 '19 at 20:27
  • $\begingroup$ Please clear out your variables, perhaps with Evaluation: Quit Kernel: Local. To print the final (nth) value: \[Phi][x, n]. $\endgroup$ – mjw Mar 1 '19 at 20:52
  • $\begingroup$ Made some edits to my answer. Had a couple of typos. Within a function definition it is x_, otherwise x. Also, \[Phi][x,j] needs two arguments, one for x and one for the jth approximation. Hope its clear. $\endgroup$ – mjw Mar 1 '19 at 20:55
  • $\begingroup$ @m_goldberg, how do you post symbols rather than the clutzy [Phi] type of notation here? $\endgroup$ – mjw Mar 3 '19 at 1:57
  • 1
    $\begingroup$ I use halirutan's plug-in. You can learn more about it here $\endgroup$ – m_goldberg Mar 3 '19 at 2:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.