Derivative of an interpol function, what to do with charfun2

El 23/11/13 17:16, ???????? ?????? escribi?:
> |f:log(%e^(-x)+sin(y))$
> y0:1$
> x1:0$
> x2:10$
> n:40$
> s:float((x2-x1)/n);
> results: rk(f,y,y0,[x,x1,x2,s])$
> load(interpol)$
> g:cspline(results)$|

After claculating g, write

define(
   e(x),
   block([terms, cubics, charac],
     /* 1 */ terms: args(g),
     /* 2 */ cubics: map(first, terms),
     /* 3 */ charac: map(second, terms),
     /* 4 */ (diff(cubics,x) - f).charac) )$



1: list of the terms of g
2: list with the cubic polynomials
3: list of the corresponding charfun2's
4: differentiate, subtract f and multiply each residual by the 
corresponding charfun2 call.

Now, you can calculate:

e(5.6);

I think this is what you are asking for. Perhaps you have to make some 
changes to fit your specific needs.


> |
> d:diff(g,x);
> e:d-f;|
> I want to get some values of e between x1 and x2 and output a descrete 
> plot of this values. The problem is that when cspline returns 
> function, the function includes charfun2() functions, and when I'm 
> trying to get for example e(1) diff says that he expects a variable as 
> a second argument, but finds a number. After taking diff there 
> shouldn't be any diffs left, but they are written before charfun2.
> e(x,y) is a <a 
> href='https://en.wikipedia.org/wiki/Residual_%28numerical_analysis%29'>residual</a>. 
>
> Is it at least possible in maxima?

--
Mario