Services & Resources / Wolfram Forums
-----
 /
MathGroup Archive
2007
*January
*February
*March
*April
*May
*June
*July
*August
*September
*October
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 2007

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: question about Protect

  • To: mathgroup at smc.vnet.net
  • Subject: [mg75149] Re: question about Protect
  • From: dimitris <dimmechan at yahoo.com>
  • Date: Wed, 18 Apr 2007 05:09:02 -0400 (EDT)
  • References: <f013jm$8f5$1@smc.vnet.net>

Hello.

What version do you use?
On mine I get

In[1]:=
Unprotect[Limit];

In[2]:=
Attributes[Limit]

Out[2]=
{Listable}

In[3]:=
Integrate[Exp[-x], {x, 1, Infinity}]

Out[3]=
1/E

In[4]:=
Attributes[Limit]

Out[4]=
{Listable, Protected}

Note that Limit MISSES the attribute ReadProtected from the attributes
list appeared
on your post.
(As long as I remember you use 5.2 for Mac; If so, does it seems
logical
this difference between the two versions?)

In[5]:=
Unprotect[Limit];

In[6]:=
Attributes[Limit]

Out[6]=
{Listable}

In[7]:=
Integrate[Exp[-x], {x, 1, 2}]

Out[7]=
(-1 + E)/E^2

In[8]:=
Attributes[Limit]

Out[8]=
{Listable}

>It seems that built-in functions, when they call on another built in
>function will sometimes Protect if it had previously been Unprotected.

Your statement seems to have a basement, but your examples are quite
contradictory. I mean, in the first example it seems that after
Integrate obtained the
antiderivative and search for convergence (where Limit is first
called) it
then (re)calls Limit in order to apply the Nweton-Leibniz formula.

That is, I think something like the following take place

In[24]:=
(Limit[Exp[-x], x -> #1] & ) /@ {1, Infinity}
ant = Integrate[Exp[-x], x]
(Limit[ant, x -> #1] & ) /@ {1, Infinity}
(Plus[#2 - #1] & ) @@ %

Out[24]=
{1/E, 0}
Out[25]=
-E^(-x)
Out[26]=
{-(1/E), 0}
Out[27]=
1/E

And in the end Limit was Protect(ed).

In the second example (since Limit was not protected and based on your
argument)
it seems that Limit was not called at all. But then how Integrate
checked for convergence
and moreover how the Newton-Leibniz formula was applied?

I mean by hand we can write down

In[32]:=
Simplify[(ant /. x -> 2) - (ant /. x -> 1)]

Out[32]=
(-1 + E)/E^2

but as far as know even for simple cases like these (that is no
indeterminate forms that
required limits or improper integrals and the staf) Mathematica
evaluates the antiderivative
at the end-points with something like

In[34]:=
Simplify[Limit[ant, x -> 2] - Limit[ant, x -> 1]]

Out[34]=
(-1 + E)/E^2

I was puzzled. But now I got more confused!

>I have not investigated this sufficiently to be able to describe the
>exact circumstancs under which this happens...

I hope you will find the time to do so. With your mathematical and
Mathematica
knowledge I am quite sure you will succeed in finding what is going on
and then...
share your conclusion with us if you don't mind!

>I speculate that a Mathematica function that calls on another Mathematica =
function
>sometimes (or often or always (?)) Unprotects it and when it finished
>its job Protects it again. If this is correct than a function that
>had been earlier Unprotected by a user would end up automatically
>Protected. Probably a better approach would be to simply save and
>restore the attributes the function had when it was called; so this
>might be a minor bug?

Bug or not a bug it deserves a thorough explanation by someone from
WRI.
Until then above explanation (or should I say guessing?) of you is
what we
just have on our hands!


Kind Regards
Dimitris








=CF/=C7 dimitris =DD=E3=F1=E1=F8=E5:
> Hello.
>
> The following code add a rule for the Limit command
>
> In[1]:=
> Off[General::spell1]
> Unprotect[Limit];
> Limit[a___] := Null /; (Print[InputForm[limit[a]]]; False)
>
> For example
>
> In[7]:=
> Integrate[1/Sqrt[Abs[x]], {x, -1, 2}]
>
> >From In[7]:=
> InputForm[limit[1 + (1 + x)/2 + (3*(1 + x)^2)/8, x -> -1, Direction ->
> -1, Assumptions -> True]]
> >From In[7]:=
> InputForm[limit[(-I)/Sqrt[x], x -> 0, Direction -> 1, Assumptions ->
> True]]
> >From In[7]:=
> InputForm[limit[I/Sqrt[x], x -> 0, Assumptions -> True]]
> >From In[7]:=
> InputForm[limit[2*Sqrt[x], x -> 1, Direction -> 1, Assumptions ->
> True]]
> >From In[7]:=
> InputForm[limit[2*Sqrt[x], x -> 0, Direction -> -1, Assumptions ->
> True]]
> >From In[7]:=
> InputForm[limit[1/Sqrt[x], x -> 0, Direction -> -1, Assumptions ->
> True]]
> >From In[7]:=
> InputForm[limit[1/Sqrt[x], x -> 0, Assumptions -> True]]
> >From In[7]:=
> InputForm[limit[1/Sqrt[2] - (-2 + x)/(4*Sqrt[2]) + (3*(-2 + x)^2)/
> (32*Sqrt[2]), x -> 2, Direction -> 1, Assumptions -> True]]
> >From In[7]:=
> InputForm[limit[2*Sqrt[x], x -> 2, Direction -> 1, Assumptions ->
> True]]
> >From In[7]:=
> InputForm[limit[2*Sqrt[x], x -> 0, Direction -> -1, Assumptions ->
> True]]
>
> Out[7]=
> 2*(1 + Sqrt[2])
>
> Note that I have NOT protect the Limit command. Nevertheless,
>
> In[8]:=
> Clear[Limit]
> Clear::wrsym: Symbol Limit is Protected.
>
>
> Why do we get this message? How Limit was protected WITHOUT telling
> so?
>
> In[12]:=
> Information["Limit", LongForm -> True]
>
> "Limit[expr, x->x0] finds the limiting value of expr when x approaches
> x0."*Button[More..., ButtonData :> "Limit",
>    Active -> True, ButtonStyle -> "RefGuideLink"]
> Attributes[Limit] = {Listable, Protected}
> Limit[a___] := Null /; (Print[InputForm[limit[a]]]; False)
> Options[Limit] = {Analytic -> False, Assumptions :> $Assumptions,
> Direction -> Automatic}
>
> Of course
>
> In[19]:=
> Unprotect[Limit];
> Clear[Limit];
> Protect[Limit];
>
> In[22]:=
> Information["Limit", LongForm -> True]
>
> "Limit[expr, x->x0] finds the limiting value of expr when x approaches
> x0."*Button[More..., ButtonData :> "Limit",
>    Active -> True, ButtonStyle -> "RefGuideLink"]
> Attributes[Limit] = {Listable, Protected}
> Options[Limit] = {Analytic -> False, Assumptions :> $Assumptions,
> Direction -> Automatic}
>
> but the question still remains!



  • Prev by Date: Re: LegendreP Evaluation Mystery
  • Next by Date: Re: functional programming
  • Previous by thread: Re: question about Protect
  • Next by thread: Re: Re: question about Protect