Account Options

  1. Sign in
The old Google Groups will be going away soon, but your browser is incompatible with the new version.
Google Groups Home
« Groups Home
Wie man leicht nachweist ...
There are currently too many topics in this group that display first. To make this topic appear first, remove this option from another topic.
There was an error processing your request. Please try again.
flag
  8 messages - Collapse all  -  Translate all to Translated (View all originals)
The group you are posting to is a Usenet group. Messages posted to this group will make your email address visible to anyone on the Internet.
Your reply message has not been sent.
Your post was successful
 
From:
To:
Cc:
Followup To:
Add Cc | Add Followup-to | Edit Subject
Subject:
Validation:
For verification purposes please type the characters you see in the picture below or the numbers you hear by clicking the accessibility icon. Listen and type the numbers you hear
 
Alexander Huber  
View profile   Translate to Translated (View Original)
 More options Sep 8 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: Alexander Huber <ahu...@dogge.cosy.sbg.ac.at>
Date: 1998/09/08
Subject: Wie man leicht nachweist ...
Hallo zusammen!

Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)

Fuer p=0 ergibt diese Funktion das "geometrische Mittel", fuer p=-1
ergibt sie das "harmonische Mittel". Anscheinend ist es nun "leicht",
nachzuweisen, dass die Funktion fuer p -> +oo max(x,y) und fuer p ->
-oo min(x,y) ergibt. Tja, ich hab eine Weile rumprobiert, doch ohne
nennenswerten Erfolg.

Kann mir bitte jemand zeigen, wie total "leicht" das ist.

Danke,
Alexander.


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Axel Reichert  
View profile   Translate to Translated (View Original)
 More options Sep 8 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: Axel Reichert <axel.reich...@gmx.de>
Date: 1998/09/08
Subject: Re: Wie man leicht nachweist ...

Alexander Huber <ahu...@dogge.cosy.sbg.ac.at> writes:
> Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)

> Fuer p=0 ergibt diese Funktion das "geometrische Mittel"

Falsch. Fuer p = 0 ist 1/p nicht definiert.

Tschoe!

--
Axel Reichert -- http://mt.mpie-duesseldorf.mpg.de/people/reich/


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
David Kastrup  
View profile   Translate to Translated (View Original)
 More options Sep 8 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: David Kastrup <d...@mailhost.neuroinformatik.ruhr-uni-bochum.de>
Date: 1998/09/08
Subject: Re: Wie man leicht nachweist ...

Alexander Huber <ahu...@dogge.cosy.sbg.ac.at> writes:
> Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)

> Fuer p=0 ergibt diese Funktion das "geometrische Mittel", fuer p=-1
> ergibt sie das "harmonische Mittel". Anscheinend ist es nun "leicht",
> nachzuweisen, dass die Funktion fuer p -> +oo max(x,y) und fuer p ->
> -oo min(x,y) ergibt. Tja, ich hab eine Weile rumprobiert, doch ohne
> nennenswerten Erfolg.

> Kann mir bitte jemand zeigen, wie total "leicht" das ist.

Für p=1 gibt sie das arithmetische Mittel, nicht für p=0.

Mach Fallunterscheidung:

x=y -> m_p = x

|x|<|y|:

y ((1+(x/y)^p)/2)^(1/p)

Für  p->+oo geht das gegen y (rechte Seite wird ((1+0)/2)^0 = 1
Ebenso für |x|>|y| und p->-oo

Die beiden anderen Fälle ergeben sich entsprechend anders rum.

In die Hose geht das Verfahren bei |x|=|y| und x /= y, etwa x=-1 und
y=1.  Aber falls x und y nicht beide positiv reell sind, gibt es eh
Ärger mit mehrdeutigen Potenzen.

--
David Kastrup                                     Phone: +49-234-700-5570
Email: d...@neuroinformatik.ruhr-uni-bochum.de       Fax: +49-234-709-4209
Institut für Neuroinformatik, Universitätsstr. 150, 44780 Bochum, Germany


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Alexander Huber  
View profile   Translate to Translated (View Original)
 More options Sep 9 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: Alexander Huber <ahu...@dogge.cosy.sbg.ac.at>
Date: 1998/09/09
Subject: Re: Wie man leicht nachweist ...

David Kastrup <d...@mailhost.neuroinformatik.ruhr-uni-bochum.de> writes:
> > Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)

> > Fuer p=0 ergibt diese Funktion das "geometrische Mittel", fuer p=-1
> > ergibt sie das "harmonische Mittel". Anscheinend ist es nun "leicht",
> > nachzuweisen, dass die Funktion fuer p -> +oo max(x,y) und fuer p ->
> > -oo min(x,y) ergibt. Tja, ich hab eine Weile rumprobiert, doch ohne
> > nennenswerten Erfolg.

> > Kann mir bitte jemand zeigen, wie total "leicht" das ist.

> Für p=1 gibt sie das arithmetische Mittel, nicht für p=0.

Scheinbar handelt es sich bei p=0 um einen Grenzfall, der als
(x*y)^1/2 identifiziert werden kann, was dem geometrischen Mittel
entspricht.

Danke jedenfalls fuer deinen Hinweis, wie obige Behauptung gezeigt
werden kann ...

Alexander


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Torsten Roensch  
View profile   Translate to Translated (View Original)
 More options Sep 9 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: Torsten Roensch <roen...@rcs.urz.tu-dresden.de>
Date: 1998/09/09
Subject: Re: Wie man leicht nachweist ...
Alexander Huber schrieb:

> Scheinbar handelt es sich bei p=0 um einen Grenzfall, der als
> (x*y)^1/2 identifiziert werden kann, was dem geometrischen Mittel
> entspricht.

Fuer sehr kleines positives p (z.B. p=0.001) liegen x^p und y^p
beide sehr nahe an 1 und ihre Differenz deshalb sehr nahe an 0.
Das _Quadrat_ ihrer Differenz ist noch einmal um eine Groessenordnung
kleiner und kann deshalb vernachlaessigt werden:

x^(2p) - 2(xy)^p + y^(2p) = (x^p - y^p)^2 =:= 0,

also x^(2p) + y^(2p) =:= 2(xy)^p.                      (*)

("=:=" steht hier fuer das gewellte "=", also "ungefaehr gleich".)

Nun ist (m_p(x,y))^2 = ( ((x^p + y^p)/2)^(1/p) )^2
                     = ( ((x^p + y^p)/2)^2 )^(1/p)
                     = ( (x^(2p) + 2(xy)^p + y^(2p))/4 )^(1/p)
                    =:=( (2(xy)^p + 2(xy)^p)/4 )^(1/p)    wegen (*)
                     = ( (4(xy)^p)/4 )^(1/p)
                     = ( (xy)^p )^(1/p)
                     = xy
und folglich m_p(x,y)=(xy)^(1/2).

Gruss
Torsten


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Uwe van Treek  
View profile   Translate to Translated (View Original)
 More options Sep 9 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: "Uwe van Treek" <uwe.van-tr...@primus-online.de>
Date: 1998/09/09
Subject: Re: Wie man leicht nachweist ...

>Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)

>Fuer p=0 ergibt diese Funktion das "geometrische Mittel", fuer p=-1
>ergibt sie das "harmonische Mittel".

p=-1 -> harmonisches Mittel (einfach)
p=1  -> arithmetisches Mittel (einfach)
p=0 -> geometrisches Mittel Beweis:
Es ist in der Form 0^oo -> also L'Hospital vorbereiten, dafür braucht man
einen Ausdruck 0/0:

m_p = exp( 1/p ln( 1/2(x^p +y^p ) ) ) = exp(  ln( 1/2(x^p + y^p))  / p )  =
exp( 0 / 0 )
Auf 0/0 L'Hospital => d/dp( ln( 1/2 (x^p + y^p) )) = 1/( 1/2(x^p+y^p) ) *
1/2 (x^p ln(x) + y^p ln(y)) )
Für p=0 -> 1/( 1/2( x^0+y^0)) * 1/2( x^0 ln(x) +y^0 ln(y) ) = 1/1 *
/2( ln(x) + ln(y) ) = 1/2 ( ln(x) + ln(y) )
Das ist der 0/0 - Ausdruck von oben
Einsetzen in exp(0/0):  exp( 1/2 (ln(x) + ln(y) ) = sqrt( exp(ln(x)) *
exp(ln(y) ) = sqrt( x * y )
Also m_0(x,y) = sqrt( x y )   q.e.d.

(wollte ich auch nur mal für mich so beweisen... )

>  Anscheinend ist es nun "leicht",
>nachzuweisen, dass die Funktion fuer p -> +oo max(x,y) und fuer p ->
>-oo min(x,y) ergibt. Tja, ich hab eine Weile rumprobiert, doch ohne
>nennenswerten Erfolg.

Umformung: 1/2 vor die Wurzel ziehen, Wurzel durch x teilen
m_p =  1 / 2^p * x *  (  1 + (y/x)^p  ) ^ (1/p)

Fallunterscheidungen:  y  < oder > x  -> Wurzelargument -> 1+0 = 1 oder 1+oo
= (y/x) ^p
Ergibt dann als Wurzelergebnis  y/x oder 1, also mal x:   y, bzw. x
DAS stimmt schon
ABER: was ist mit dem 1/2^p? das stimmt doch nicht!
AUSSERDEM:  bei negativen x,y geht das mit dem max/min auch nicht richtig,
von den Schwierigkeiten beim Potenzieren von negativen Zahlen mal ganz
abgesehen

Dasselbe gilt für p -> -oo

Korrektur: m_p(x,y) = (x^p+y^p)^(1/p) für x,y >=0 (ausserdem reel), ergibt
für p->oo max(x,y) und für p->-oo min(x,y)


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Dr. Peter Kittel  
View profile   Translate to Translated (View Original)
 More options Sep 10 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: pet...@combo.ganesha.com (Dr. Peter Kittel)
Date: 1998/09/10
Subject: Re: Wie man leicht nachweist ...
In article <ydwsoi2y9m2....@dogge.cosy.sbg.ac.at> Alexander Huber <ahu...@dogge.cosy.sbg.ac.at> writes:

>Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)

>Fuer p=0 ergibt diese Funktion das "geometrische Mittel", fuer p=-1
>ergibt sie das "harmonische Mittel". Anscheinend ist es nun "leicht",
>nachzuweisen, dass die Funktion fuer p -> +oo max(x,y) und fuer p ->
>-oo min(x,y) ergibt. Tja, ich hab eine Weile rumprobiert, doch ohne
>nennenswerten Erfolg.

>Kann mir bitte jemand zeigen, wie total "leicht" das ist.

Ersetze bei x>y (x^p+y^p) durch (x^p*(1+(y/x)^p) bzw. im anderen
Fall andersherum ausklammern. Dann wird das zweite Glied der inneren
Klammer bei steigendem p zu einer Nullfolge. Bei negativem p
andersrum verfahren.

--
Best Regards, Dr. Peter Kittel       // E-Mail:
Private Site in Frankfurt, Germany \X/  peterk @ combo.ganesha.com


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Torsten Roensch  
View profile   Translate to Translated (View Original)
 More options Sep 10 1998, 3:00 am
Newsgroups: de.sci.mathematik
From: Torsten Roensch <roen...@rcs.urz.tu-dresden.de>
Date: 1998/09/10
Subject: Re: Wie man leicht nachweist ...
Uwe van Treek schrieb:

> >Gegeben ist folgende Funktion: m_p(x,y) = ((x^p+y^p)/2)^(1/p)
> >Anscheinend ist es nun "leicht",
> >nachzuweisen, dass die Funktion fuer p -> +oo max(x,y) und fuer p ->
> >-oo min(x,y) ergibt. Tja, ich hab eine Weile rumprobiert, doch ohne
> >nennenswerten Erfolg.

> Umformung: 1/2 vor die Wurzel ziehen, Wurzel durch x teilen
> m_p =  1 / 2^p * x *  (  1 + (y/x)^p  ) ^ (1/p)

> Fallunterscheidungen:  y  < oder > x  -> Wurzelargument -> 1+0 = 1 oder 1+oo
> = (y/x) ^p
> Ergibt dann als Wurzelergebnis  y/x oder 1, also mal x:   y, bzw. x
> DAS stimmt schon
> ABER: was ist mit dem 1/2^p? das stimmt doch nicht!

_Das_ stimmt wirklich nicht; es muss natuerlich (1/2)^(1/p) heissen.
Und (1/2)^(1/p) = 1/(2^(1/p)) geht ja wohl gegen 1 fuer p -> oo.

Gruss
Torsten


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
End of messages
« Back to Discussions « Newer topic     Older topic »