|
|
|
last edited 15 years ago by gdr |
Edit detail for #312 Bug in sqrt() for UnivariateTaylorSeries revision 1 of 2
| 1 2 | ||
|
Editor:
Time: 2007/11/17 22:24:55 GMT-8 |
||
| Note: documentation of fix | ||
changed: - \begin{axiom} UTSx := UTS(FRAC INT,x,0) monx2 := monomial(1,2)$UTSx sqrt(monx2) \end{axiom} The last line should of course be $x$ and not $1$. Clearly, sqrt() shifts the coefficients of the result such that the leading term is always O(1). From kratt6 Fri Oct 27 03:02:19 -0500 2006 From: kratt6 Date: Fri, 27 Oct 2006 03:02:19 -0500 Subject: Fix by Igor Khavkine Message-ID: <20061027030219-0500@wiki.axiom-developer.org> I'm forwarding this fix proposed by Igor himself, lest it is lost in the archives. The relevant message in axiom-math is http://lists.gnu.org/archive/html/axiom-math/2006-08/msg00025.html For simplicity, here is the fix:: diff -ur axiom-20050901/src/algebra/sttaylor.spad.pamphlet axiom-20050901.new/src/algebra/sttaylor.spad.pamphlet --- axiom-20050901/src/algebra/sttaylor.spad.pamphlet 2005-01-30 07:03:10.000000000 -0500 +++ axiom-20050901.new/src/algebra/sttaylor.spad.pamphlet 2006-08-21 15:49:25.000000000 -0400 @@ -448,6 +448,9 @@ RATPOWERS => co**rn * YS(powerrn(rn,(invCo :: A) * x,#1)) error "** rational power of coefficient undefined" + zeropref : ST A := new(ord::I::NNI,0::A)$LA :: ST A + concat(zeropref, power) + if A has Field then mapdiv(x,y) == delay If somebody could complete the documentation as outlined in http://lists.gnu.org/archive/html/axiom-developer/2006-08/msg00486.html that would be great. Finally, I have to confess that it is very likely myself who introduced this bug, while fixing #63 Martin From kratt6 Fri Oct 27 03:03:15 -0500 2006 From: kratt6 Date: Fri, 27 Oct 2006 03:03:15 -0500 Subject: fix by Igor Khavkine Message-ID: <20061027030315-0500@wiki.axiom-developer.org> Status: open => fix proposed Here is the documented patch. (Documentation is partially due to Waldek Hebisch) <a href="powern.patch">powern.patch</a> Martin
axiomUTSx := UTS(FRAC INT,x,0)
 | (1) |
Type: Domain
axiommonx2 := monomial(1,2)$UTSx
 | (2) |
Type: UnivariateTaylorSeries?(Fraction Integer,x,0)
axiomsqrt(monx2)
 | (3) |
Type: UnivariateTaylorSeries?(Fraction Integer,x,0)
The last line should of course be 
and not 
. Clearly, sqrt() shifts the coefficients of the result such that the leading term is always O(1).
Fix by Igor Khavkine --kratt6, Fri, 27 Oct 2006 03:02:19 -0500 reply
I'm forwarding this fix proposed by Igor himself, lest it is lost in the archives.
The relevant message in axiom-math is
http://lists.gnu.org/archive/html/axiom-math/2006-08/msg00025.html
For simplicity, here is the fix:
diff -ur axiom-20050901/src/algebra/sttaylor.spad.pamphlet axiom-20050901.new/src/algebra/sttaylor.spad.pamphlet
--- axiom-20050901/src/algebra/sttaylor.spad.pamphlet 2005-01-30 07:03:10.000000000 -0500
+++ axiom-20050901.new/src/algebra/sttaylor.spad.pamphlet 2006-08-21 15:49:25.000000000 -0400
@@ -448,6 +448,9 @@
RATPOWERS => co**rn * YS(powerrn(rn,(invCo :: A) * x,#1))
error "** rational power of coefficient undefined"
+ zeropref : ST A := new(ord::I::NNI,0::A)$LA :: ST A
+ concat(zeropref, power)
+
if A has Field then
mapdiv(x,y) == delay
If somebody could complete the documentation as outlined in
http://lists.gnu.org/archive/html/axiom-developer/2006-08/msg00486.html
that would be great.
Finally, I have to confess that it is very likely myself who introduced this bug, while fixing #63
Martin
Status: open => fix proposedHere is the documented patch. (Documentation is partially due to Waldek Hebisch)
Martin