New version, old bugs
- To: mathgroup at smc.vnet.net
- Subject: [mg44093] New version, old bugs
- From: Maxim <dontsendhere@.>
- Date: Wed, 22 Oct 2003 03:24:52 -0400 (EDT)
- Sender: owner-wri-mathgroup at wolfram.com
Last week I got a message from WRI technical support saying that the
issues in my March 2002 bug report were resolved in version 5. I thought
at first this must have been a mistake, because I sent my bug report on
version 5.0 to WRI in July 2003. But it turned out they were referring
to my Mathematica 4.0 bug report from the last year...so I suppose I
should expect the answer to my last report in February 2005. So far I
only received a recommendation to send each problematic example as a
separate message, so now I'm contemplating creating and mailing about
200 separate notebooks (also consider that some examples are related).
But technical support is supposed to be helpful, so that's okay; what
bothers me here is that some issues from the 4.0 report definitely
haven't been resolved! For example:
In[1]:=
2*x*y /. x*y -> z
2*x*y /. s:x*y -> z
Out[1]=
2*z
Out[2]=
2*x*y
This example has been in my 4.0 bug report (although in a more
complicated form), and it still works the same way, so why saying that
the problem was solved when it wasn't?
Also there are some issues that I suppose are open to discussion:
In[1]:=
Sum[If[EvenQ[k], 1, 0], {k, 1, n}]
Out[1]=
0
In[2]:=
Sum[k[1 - 1], {k[0], 1, n}]
Out[2]=
n*k[0]
The summand is evaluated outside of the Sum, but at the same time
Mathematica checks if it's free of the iterator variable only before
doing this evaluation.
Another issue is the series expansions at singularities or points on
branch cuts:
In[1]:=
HypergeometricPFQ[{1/3, 2/3, 1}, {1/4, 1/2}, z] + O[z, 1]^2
Out[1]=
HypergeometricPFQ[{1/3, 2/3, 1}, {1/4, 1/2}, 1] +
(16/9)*HypergeometricPFQ[{4/3, 5/3, 2}, {5/4, 3/2}, 1]*(z - 1)+
O[z-1]^2
Essentially Mathematica just returns ComplexInfinity here.
Maxim Rytin
m.r at prontomail.com