So, changing your colons to semicolons, we see that the solution set is a one-dimensional subset of C^3, but its intersection with R^3 is zero-dimensional.

So, changing your colons to semicolons, we see that the solution set is a one-dimensional subset of C^3, but its intersection with R^3 is zero-dimensional.

That's tricky as there are no parameters. I guess that the difference from regular solve is that it uses the SolveTools:-SemiAlgebraic algorithm, which in turn uses the massive RegularChains package.

It is also interesting that the solution is returned as a listlist.

That's tricky as there are no parameters. I guess that the difference from regular solve is that it uses the SolveTools:-SemiAlgebraic algorithm, which in turn uses the massive RegularChains package.

It is also interesting that the solution is returned as a listlist.

@GOODLUCK Why do your equations contain, repeatedly, expressions such as A/A, ctilde^s/ctilde^s?

@ysf Could you provide more details, please? Does it put the entire program in a single long line in the file? Could you upload your Maple code, please?

@ysf Could you provide more details, please? Does it put the entire program in a single long line in the file? Could you upload your Maple code, please?

@GOODLUCK What do you mean by calling mutilde^H, etc., "variables"? The variable is mutilde, and mutilde^H is an expression. It wouldn't matter if mutilde always appeared as mutilde^H, but it also appears as mutilde^B.

Because of your mixed used of capital letters, I can't tell whether you are using codegen[fortran] or CodeGeneration[Fortran]. So which one is it?

What version of Maple are you using? Your applyrule works for me in Maple 17. You need to correct the result to (X^(2*k) - Y^(2*k))^n. But even without the correction, the applyrule works.

@wolfman29 It is a bug, because it is supposed to accept functions that way in 2D input, with a popup asking whether you want to define a function or make a remember table entry.

@wolfman29 It is a bug, because it is supposed to accept functions that way in 2D input, with a popup asking whether you want to define a function or make a remember table entry.

@brian bovril I corrected the code in the Answer, changing 2 to 3. Note that ilog10(x) is one less than the number of digits in x, so the precision is now 2 digits more than are in x. Since the next fibonacci could have at most one more digit than x, this gives one guard digit to guard against round-off errors in the intermediate calculation.

Regarding the Font: So, were you able to get through the menu sequence? If so, did it seem to change anything at all?

@brian bovril I corrected the code in the Answer, changing 2 to 3. Note that ilog10(x) is one less than the number of digits in x, so the precision is now 2 digits more than are in x. Since the next fibonacci could have at most one more digit than x, this gives one guard digit to guard against round-off errors in the intermediate calculation.

Regarding the Font: So, were you able to get through the menu sequence? If so, did it seem to change anything at all?

@J4James Where did you get the "exact" solution? It seems wrong. You can check that it doesn't satisfy the original equations like this:

F1test:= unapply(Exact, eta):
eval(eq1, F1= F1test):
plot(lhs(%), eta= 0..n);

It should be identically 0, modulo some rounding errors. Note that I am not comparing Exact against the computed numeric solution; I am comparing against the original differential equation.

Please ask any further questions in a separate thread. You're lucky that I saw your Reply, because it is difficult to find Replies to old Answers.