🚀 go-pugleaf

	The example of integration that did not work in REDUCE 3.4
	submitted by Geoff Prince seems to be general.  The following
	gives proper results:
	    int(x^n,x) with variable n (even though n might be 0).
	However as near as I can tell
	    int(x^(a/b),x) only works for constants a and b if b is 2,
	    3, or 4.  It may work for larger b is the fraction a/b is
	reducible to a'/b' where b' is 2, 3, or 4.
	    The following statement lets the general case be
	    evaluated:  for all n let int(x^n,x)=x^(n+1)/(n+1); Even
	    with that statement, the original system correctly
	    evaluates the expression when n equals -1.

Even that does not work in all cases, but intpatch, which is available
in the MISC section of the network library (or from me), is intended to
fix this problem (not very elegantly) as one of its functions.  (The
other is to allow DEPENDent integrands.)  Here is an (edited) example.
[On my own system I have intpatch precompiled and auto-load it via the
PSL-REDUCE initialization file.]

4: for all n let int(x^n,x)=x^(n+1)/(n+1);

7: int(x^(-3/5),x);

      1
INT(------,X)
      3/5
     X

8: in "REDUCE/intpatch.red"$

10: input 7;

    2/5
 5*X
--------
   2


Dr. Francis J. Wright
Visiting from August 1992 to July 1993:	|  EMail: fjw@cbpfsu1.cat.cbpf.br
Centro Brasileiro de Pesquisas Fisicas	|  Phone: +55 (21) 541-0337 R 176
Rua Dr. Xavier Sigaud 150,  Urca	|  FAX:   +55 (21) 541-2047
22290 Rio de Janeiro - RJ,  BRAZIL	|  Telex: 21-22563 CBPQ BR