Please post the entire matrix as a Maple worksheet.

I don't see any exponentials in the example entry that you posted.

@nm I think that Ctrl-JT is short enough to be considered a single command.

@Kitonum solve[RealDomain] is exactly the same as solve. The part in square brackets is ignored. I think that you mean RealDomain[solve].

@toandhsp 

RealDomain:-solve({eq1,eq2}, {x,y});

@nm Or you can get exact solutions with option explicit:

eq1:=(x+y)*(x^2+y^2) = 5500;
eq2:=(x-y)*(x^2-y^2) = 352;
solve({eq1,eq2}, {x,y}, explicit);

@Ratch I see the sort for a fraction of a second, then it disappears.

@Preben Alsholm 

Your bcsy has boundary conditions specified at x=0, x=1, and x=inf.

You asked the same question five days ago, here, and I answered it. Did you have some problem using that answer?

@nm There is no way in Maple to change an operator's precedence (see ?precedence ). &= has a rather high precedence, and := has the absolute lowest. I can change it so that the grouping symbol is unevaluation quotes:

x &= 'r1 + r2';

or any other grouping symbol, but there must be some grouping symbol.

@Ratch And where did you put the sort command?

@Markiyan Hirnyk 

I agree that int and Int should give infinity and Float(infinity) as answers.

While the overall series is not clear, it is clear that the coefficient of the (x-Pi)^(-1) term is not zero and has modulus 1.

@Markiyan Hirnyk Please clarify your statement about the general solution. And what do you mean by "exception"? It seems that the general solution cannot be produced by dsolve.

@BobWild Do you mean that you want to solve part b, i.e., get the analytical solution? That is just the same answer that I gave before with different specific values.

Or do you mean that you want to solve part c, i.e., get the matrix/vector solution using a stepsize of h=Pi/8? That's more complicated. To help you with that, I would need to see the page that precedes equation 11.18 in your book. Specifically, I need to see how I the mesh points are defined. I don't know if they are using a midpoint or endpoint method. You will need to set up the matrix and vector for that, because the mesh options to dsolve(..., numeric) will not let you make a mesh that small.

@BobWild Correct: You do not need to define the mesh points. That is done automatically. Often it helps to adjust the number of mesh points. This is done with the options initmesh and maxmesh. See ?dsolve,bvp for precise details, or post a specific BVP problem here. But it is best to not add any options unless you initially get an error message without them. An error message usually gives a suggestion for which option to adjust and which direction to adjust it.

There is no need to create a matrix and no option for it either. That is all handled internally. You can pass in a vector for an approximate solution. And you can receive the output in matrix/vector form, but the procedure form of output is usually more convenient, with interpolations handled automatically.

Please read the help page ?dsolve,bvp .

@BobWild That is correct; only the boundary conditions are needed.