f



Signed square ss(x) = (x^2)^(1/2)*x = sgn(x)*x^2

Hi

Since the power is more relevant to analyze than the amplitude I have come =
up with the idea of using signed square ss(x) =3D x^2 for positive x and -x=
^2 for negative x or ss(x) =3D (x^2)^(1/2)*x =3D sgn(x)*x^2

Instead of analyzing the power as the square of the amplitude the signed po=
wer could be used instead. This would preserve more information of the orig=
inal signal. The original amplitude signal can be recreated from the signed=
 square power signal with signed square root.

I can imagine signed square to be applicable to codec evaluation and signal=
 compression since the power of the signal is more relevant to compress tha=
n the amplitude.

It would be easy to make a simple test with Scilab and some sound or image =
file.

Would it be possible to analyze this theoretically with Karhunen-Lo=E8ve th=
eory?

David


0
10/29/2013 2:49:11 PM
comp.dsp 20333 articles. 1 followers. allnor (8509) is leader. Post Follow

2 Replies
1026 Views

Similar Articles

[PageSpeed] 38

On Tuesday, October 29, 2013 10:49:11 AM UTC-4, davidjons...@gmail.com wrot=
e:
> Hi
>=20
>=20
>=20
> Since the power is more relevant to analyze than the amplitude I have com=
e up with the idea of using signed square ss(x) =3D x^2 for positive x and =
-x^2 for negative x or ss(x) =3D (x^2)^(1/2)*x =3D sgn(x)*x^2
>=20
>=20
>=20
> Instead of analyzing the power as the square of the amplitude the signed =
power could be used instead. This would preserve more information of the or=
iginal signal. The original amplitude signal can be recreated from the sign=
ed square power signal with signed square root.
>=20
>=20
>=20
> I can imagine signed square to be applicable to codec evaluation and sign=
al compression since the power of the signal is more relevant to compress t=
han the amplitude.
>=20
>=20
>=20
> It would be easy to make a simple test with Scilab and some sound or imag=
e file.
>=20
>=20
>=20
> Would it be possible to analyze this theoretically with Karhunen-Lo=E8ve =
theory?
>=20
>=20
>=20
> David

I don't know off hand where a cubic representation has an advantage over th=
e classical cases, but one may exist. While you are in this vein, you may w=
ish to look up the Teager-Kaiser energy operator.

Clay
0
clay (793)
10/29/2013 5:31:21 PM
On Tuesday, October 29, 2013 6:31:21 PM UTC+1, cl...@claysturner.com wrote:
> 
> I don't know off hand where a cubic representation has an advantage over the classical cases, but one may exist. While you are in this vein, you may wish to look up the Teager-Kaiser energy operator.
> 

Signed square is not cubic. There are only squared terms involved. Maybe you prefer signed square written like this

         x^2 for x >= 0 
ss(x) = 
        -x^2 for x < 0

I could not find any application of it on Internet. The inverse to ss(x) is signed square root and it has applications in geology and exists in some software.

Teager-Kaiser energy operator TKEO = x^2(n) - x(n+1)*x(n-1) seems to find energy differences which might not be what I am looking for.

There must be a problem with TKEO since it gives the same value for -x(n) as for x(n) (for the same x(n+1) and x(n-1)).

That problem could be avoided by defining a signed square TKEO = SSTKEO(x(n)) = ss(x(n)) - sgn2(x(n+1), x(n-1))*x(n+1)*x(n-1) 
where sgn2(a, b) is defined as being positive only when a and b are positive and negative if either or both a and b are negative.

The only eventual problem with ss(x) is that it has a discontinuous second derivative. And sgn() and sgn2() are discontinuous.

David
0
10/30/2013 9:18:19 AM
Reply:

Similar Artilces:

how to "(1-3*x^2)/(1-3*x+x^2+2*x^3) => 1/(1-2*x) + x/(1-x-x^2)"?
hi normal(1/(1-2*x) + x/ (1-x-x^2),expanded); 1/(1-2*x) + x/(1-x-x^2) => (1-3*x^2)/(1-3*x+x^2+2*x^3) how to do the reverse? dillogimp@gmail.com writes: > normal(1/(1-2*x) + x/ (1-x-x^2),expanded); > 1/(1-2*x) + x/(1-x-x^2) => (1-3*x^2)/(1-3*x+x^2+2*x^3) > how to do the reverse? convert(%,parfrac,x); -- Joe Riel ...

x^(3/2)+x^(5/2) = (x+x^2)*x^(1/2)
Is there a simple way to get Maple to convert x^(3/2) + x^(5/2) to (x+x^2)*x^(1/2). I am willing to assume(x>0). Maple does recognize these as being equal: > p:=x^(3/2)+x^(5/2): > q:=(x+x^2)*sqrt(x): > simplify(p-q); 0 Here's a clumsy way to do it: > p:=x^(3/2)+x^(5/2): > assume(t>0): > subs(x=t^2,p); > simplify(%,{t^2=x}); > collect(%,t); Is there something simpler? (This has undoubtedly come up before, but I cannot find it anywhere.) --Edwin If you do not insist upon your incomplete factoriz...

x^2-x-x^3 -> -(x^3-x^2+x) #2
The CAS "simplifies" x^2-x-x^3 to -(x^3-x^2+x) (flag -114 clear). Is there a way to set it to simplify this to -x^3+x^2-x ? John Pressman wrote: > The CAS "simplifies" x^2-x-x^3 to -(x^3-x^2+x) (flag -114 clear). Is > there a way to set it to simplify this to -x^3+x^2-x ? FDISTRIB Regards, -- Beto Reply: Erase between the dot (inclusive) and the @. Responder: Borra la frase obvia y el punto previo. ...

x^2-x-x^3 -> -(x^3-x^2+x)
The CAS "simplifies" x^2-x-x^3 to -(x^3-x^2+x) (114 clear). Is there a way to set it to simplify this to -x^3+x^2-x ? In article <2cdd7b99.0312120050.4f899916@posting.google.com>, voldermort@hotmail.com (John Pressman) wrote: > The CAS "simplifies" x^2-x-x^3 to -(x^3-x^2+x) (114 clear). Is there a > way to set it to simplify this to -x^3+x^2-x ? On the hp49/HP49+ use FDISTRIB. ...

x**2 or x**2.0 or x*x
makes this any difference ? x**2 or x**2.0 or x*x x is real Cheers -Ralf "Ralf Schaa" <schaa@geo.uni-koeln.de> wrote in message news:1133832989.230157.34130@g49g2000cwa.googlegroups.com... > makes this any difference ? > > x**2 or x**2.0 or x*x > x is real > > Cheers > -Ralf > It might be. Some would implement the first as x*x. Some would also implement the second as X*x as a special case of the more general expansion of exp(ln(x)*2)) (providing my math is correct anyway).. Jim Ralf Schaa wrote: > makes this any difference ? > > x...

x = x
x = x - (x+1) + x + 2: Numbers: 101 (TM) Code. -100=-101+1==-99-1===-100 -099=-101+2==-98-1===-99 -098=-101+3==-97-1===-98 -097=-101+4==-96-1===-97 -096=-101+5==-95-1===-96 -095=-101+6==-94-1===-95 -094=-101+7==-93-1===-94 -093=-101+8==-92-1===-93 -092=-101+9==-91-1===-92 -091=-101+10==-90-1===-91 -090=-101+11==-89-1===-90 -089=-101+12==-88-1===-89 -088=-101+13==-87-1===-88 -087=-101+14==-86-1===-87 -086=-101+15==-85-1===-86 -085=-101+16==-84-1===-85 -084=-101+17==-83-1===-84 -083=-101+18==-82-1===-83 -082=-101+19==-81-1===-82 -081=-101+20==-80-1===-81 -080=-101+21==-79-1===-80 -079=-101+22==-...

Integral: Exp[-(x-m)^2/(2 s^2)] x^3 (1+x^2)^-1
Hi, any suggestion to make the integral of: Exp[-(x-m)^2/(2 s^2)] x (1+x^2)^-1 Exp[-(x-m)^2/(2 s^2)] x^2 (1+x^2)^-1 Exp[-(x-m)^2/(2 s^2)] x^3 (1+x^2)^-1 between -inf and +inf (or indefinite)? Look like it is not possible, but it is too long time I do not make integrals with more advanced techinques (as going to the complex plane)... so if you have suggestions (wonderful a solution :) ).... THANKS Ale ...

HoldForm[1*2*3] should give 1 x 2 x 3 not 2 x 3
HoldForm[] is loosing the 1* when it apparently should not: In[1]:= HoldForm[1*2*3] Out[1]= 2 x 3 In[2]:= HoldForm[1*1*1] Out[2]= 1 x 1 x 1 In[3]:= HoldForm[3*2*1] Out[3]= 3 x 2 In[4]:= HoldForm[2*2*2] Out[4]= 2 x 2 x 2 In[5]:= HoldForm[2*1*3] Out[5]= 2 x 3 In[6]:= HoldForm[1*2] Out[6]= 1 x 2 In[7]:= HoldForm[1*2*1] Out[7]= 1 x 2 x 1 Q.E.D. Indeed, the same happens with Hold and HoldComplete. I'd say this is a bug. Cheers -- Sjoerd On Feb 25, 11:07 am, "Q.E.D." <a...@netzero.net> wrote: > HoldForm[] is loosing the 1* when it...

Integrate[ x^2 * Erfc[a x] *(BesselJ[1, b x])^2 , {x, 0, Inf}]
Hi group, The integral Integrate[ x^2 * Erfc[a x] *(BesselJ[1, b x])^2 , {x, 0, Inf}] has a solution in Mathematica (in terms of HypergeometricPFQ) My question: how did Mathematica know? I tried the usual suspects: i) for example Gradshteyn and Ryzhik (1965), p.769, 6.784 is very close, but not close enough ii) Abramowitz & Stegun hasn't got it on the menu either Then how about expressing Erfc[a*x] in terms of... Erfc(z)=-1/(sqrt(pi)) * IncompleteGamma[1/2,x^2]+1 (Grads.+Ryzh. p. 942) ....and hoping to get lucky? ... no alas! sorry, this is a cross-post (als...

Risch(sqrt(x+sqrt(x^2+a^2))/x, x) hangs
Risch(sqrt(x+sqrt(x^2+a^2))/x, x) seems to hang on my 49G+. Does anyone know what it's trying to do, since (AFAIK) the algebraic case of the Risch algorithm is not implemented? Thanks, Bhuvanesh. The Risch algorithm is only a partial implementation. It does handle sqrt very easily AFAIK. I think it is documented in the latest Erable 3.024 distribution for the 48. Regards, Terry "Bhuvanesh" <lalu_bhatt@yahoo.com> wrote in message news:662e00ed.0408171050.4fabb8db@posting.google.com... > Risch(sqrt(x+sqrt(x^2+a^2))/x, x) seems to hang on my 49G+. Does > anyone know...

how to convert union(2*x+y=3,3*x+2*y=5) to {2*x+y=3,3*x+2*y=5}
Thanks! In article <1140494713.980133.168190@f14g2000cwb.googlegroups.com>, loric <dr.huiliu@gmail.com> wrote: >Thanks! It isn't clear exactly what your question is, but: > `union`({2*x+y=3},{3*x+2*y=5}); {2 x + y = 3, 3 x + 2 y = 5} > Union := proc() { seq( `if`(s::'set',op(s),s), s=args ) } end proc: > Union(2*x+y=3,3*x+2*y=5); {2 x + y = 3, 3 x + 2 y = 5} > Union(2*x+y=3,{3*x+2*y=5}); {2 x + y = 3, 3 x + 2 y = 5} ...

Can we recover f0(x) or f1(x), given g(x)=f0(x)*exp(j k0 x^2)+f1(x)*exp(j k1 x^2)?
where g(x), k0 and k1 are known and * denotes the convolution operation. From the chirp point view, exp(j k x^2) is a linear FM chirp. Is there a method that we can recover f0(x) from the composite? ...

Can we recover f0(x) or f1(x), given g(x)=f0(x)*exp(j k0 x^2)+f1(x)*exp(j k1 x^2)?
where g(x), k0 and k1 are known and * denotes the convolution operation. From the chirp point view, exp(j k x^2) is a linear FM chirp. Is there a method that we can recover f0(x) from the composite? Tristan Imp wrote: > Can we recover f0(x) or f1(x), given g(x)=f0(x)*exp(j k0 > x^2)+f1(x)*exp(j k1 x^2)? where g(x), k0 and k1 are known and * > denotes the convolution> operation. > > From the chirp point view, exp(j k x^2) is a linear FM chirp. > > Is there a method thatwe can recover f0(x) from the composite? As far as I can tell, you have one equation and two unk...

x - 1 = x
What languages might use a scheme similar to this? Could I build my own? x = x - (x+1) + x + 2: Numbers: 101 (TM) Code. -100=-101+1==-99-1===-100 -099=-101+2==-98-1===-99 -098=-101+3==-97-1===-98 -097=-101+4==-96-1===-97 -096=-101+5==-95-1===-96 -095=-101+6==-94-1===-95 -094=-101+7==-93-1===-94 -093=-101+8==-92-1===-93 -092=-101+9==-91-1===-92 -091=-101+10==-90-1===-91 -090=-101+11==-89-1===-90 -089=-101+12==-88-1===-89 -088=-101+13==-87-1===-88 -087=-101+14==-86-1===-87 -086=-101+15==-85-1===-86 -085=-101+16==-84-1===-85 -084=-101+17==-83-1===-84 -083=-101+18==-82-1===-83 -082=-101+19==-81-1...

x*x versus x^2
So I've been looking at execution time for various algorithms, and I found this interesting result: bigarr = fltarr(1000,1000) t1 = systime(/seconds) t = bigarr^2.0 t2 = systime(/seconds) t = bigarr*bigarr t3 = systime(/seconds) print,t2-t1 print,t3-t2 IDL prints: 0.024163008 0.010262012 Apparently multiplying an array by itself is twice as fast as using the carat operator! Anyone know why this is? Is it a memory issue or something? On Jul 9, 12:32 pm, Conor <cmanc...@gmail.com> wrote: > So I've been looking at execution time for various algorithms, and I &...

Slove 0 = a(1) + a(2)*x + a(3)*x^2 + a(4)*x^3 +...?
Hi all, My problem is very easy. But I don't know quite MatLab to know if there is a function to solve that: 0 = a(1) + a(2)*x + a(3)*x^2 + a(4)*x^3 +...? "a" is vector known. Thank you for your help Fabien. Look at the "roots" function. "Fabien Blarel" <eezfb@gwmail.nottingham.ac.uk> wrote in message news:eec6364.-1@WebX.raydaftYaTP... > Hi all, > > > My problem is very easy. But I don't know quite MatLab to know if > there is a function to solve that: > > > 0 = a(1) + a(2)*x + a(3)*x^2 + a(4)*x^3...

Re: How to simplify "Integrate[2 f[x], {x, 0, 1}]/2" to "Integrate[f[x], #2
On 4/16/10 at 5:53 AM, klaus.engel@tiscali.it (Klaus Engel) wrote: >I tried to simplify an awkward looking integral with "Mathematica 7" >using its "(Full)Simplify[...]" function. Unfortunately it failed to >do so, even though I know that this would be possible. I boiled down >the problem to the following very simple example ("f" is just a >generic, undefined function): The input >Integrate[2 f[x], {x, 0, 1}]/2 // FullSimplify >returns just the input >Integrate[2 f[x], {x, 0, 1}]/2 The problem is f is undefined. Consequen...

int(1/x, x): why ln(x) and not ln(abs(x))? #2
Hi! Why Maple developers decided to evaluate int(1/x, x) as ln(x) and not ln(abs(x))? Thanks, Humberto. Humberto Bortolossi wrote: > Hi! > > Why Maple developers decided to evaluate > > int(1/x, x) > > as > > ln(x) > > and not > > ln(abs(x))? > > Thanks, Humberto. Because ln(abs(x)) is wrong if x is complex. If x is not dimensionless then both ln(x) and ln(abs(x)) are both wrong, whether x is complex or not. Instead int(1/x,x) is ln(x/c) where c is an arbitrary constant with the same dimensions...

if d = x^2, and x=[0,0;1,0] then why is eval(d) not equal to x^2
Hello my name is Don Robison. I am having a problem with the way matlab evaluates expressions. It seems if I put a polynomial on the command prompt and I have declared the variables to be matrices the evaluation will be fine, but if I declare the polynomial to be a symbol then try to evaluate the symbolic expression instead of a literal they are not the same thing. Is there something other than eval I should be using to plug the matrices into the symbolic expression? eval seems to not be evaluating the exponent of the matrix. Don Robison <joe_blow_281@hotmail.com> wrote in messag...

How to simplify ArcCos[x/Sqrt[x^2+y^2]] to Pi/2-ArcTan[x/Abs[y]]?
Hi, ArcCos[x/Sqrt[x^2+y^2]] and Pi/2-ArcTan[x/Abs[y]] are the same. But I can not get the first expression be simplified to the second one. And the following command can not be simplified to zero in mathematica as well. FullSimplify[ArcCos[x/Sqrt[x^2 + y^2]] - ( Pi/2 - ArcTan[x/Abs[y]]), Element[x, Reals] && Element[y, Reals]] I'm wondering if there is any walkaround to do this simplification. Thanks, Peng Peng Yu schrieb: > Hi, > > ArcCos[x/Sqrt[x^2+y^2]] > > and > > Pi/2-ArcTan[x/Abs[y]] > > are the same. &...

Re: Plot of (2 x^2
Please check Table[{x, 2*x^2 - x^3, (2*x^2 - x^3)^ (1/3), Re[(2*x^2 - x^3)^(1/3)]}, {x, -4, 4}] Best regards, MATTHIAS BODE S 17.35775=B0, W 066.14577=B0 > Date: Sat, 28 Aug 2010 07:01:34 -0400 > From: bernard.vuilleumier@edu.ge.ch > Subject: Plot of (2 x^2 - x^3)^(1/3) > To: mathgroup@smc.vnet.net > > I see in Calcul Diff==E9rentiel et int==E9gral, N. Piskounov, Editions > MIR, Moscou 1970, on p. 210 the graph of (2 x^2 - x^3)^(1/3) with > negative values for x > 2, something like : > > Plot[Piecewise[{{(2 x^2 - x^3)^(1/3...

porting network driver from 2.2.x to 2.4.x #2
Hello, All! There are some sources of pseudo network interface driver develkoped for 2.2.x kernel. I have to port it onto 2.4.x kernel. I have explored the structure 'net_device' from linux/net/netdevice.h, there are few major changes. The main problem for now is - how to check that interface is ready to transmit data (in 2.2.x 'tbusy' field was used fot this and function 'test_and_set_busy' checking the 'tbusy' state). But thus field was removed from 2.4.x kernel, and now you should use the function 'netif_start_queue' or 'netif_stop_queu...

What is the difference between (x^2+y^2) and (x.^2+y.^2) in matlab?
z=sqrt(x^2+y^2) c=(z<15) imshow(c) gives a square cut diagonally half with lower right side being white and other being black z=sqrt(x.^2+y.^2) c=(z<15) imshow(c) gives a square with circle in centre What is the difference between (x^2+y^2) and (x.^2+y.^2) in matlab? "Sagar Chand" <sagarchand_9@rediffmail.com> wrote in message <n8efgg$100$1@newscl01ah.mathworks.com>... > z=sqrt(x^2+y^2) > c=(z<15) > imshow(c) > gives a square cut diagonally half with lower right side being white and other being black > > z=sqrt(x.^2...

x += ... is not the same than x = x + ... if x is mutable
Hi, I thought that x +=3D ... was the same than x =3D x + ..., but today I have= realized it is not true when operating with mutable objects. In Python 3.3 or 2.7 IDLE (Windows) compare: >>> a =3D [3] >>> b =3D a >>> a =3D a + [1] >>> b [3] and >>> a =3D [3] >>> b =3D a >>> a +=3D [1] >>> b [3, 1] Is this behaviour explained in the Python documentation?=20 Thanking you in advance, Bartolom=E9 Sintes bartolome.sintes@gmail.com writes: > Hi, > > I thought that x += ... was t...

Web resources about - Signed square ss(x) = (x^2)^(1/2)*x = sgn(x)*x^2 - comp.dsp

Signed number representations - Wikipedia, the free encyclopedia
In computing , signed number representations are required to encode negative numbers in binary number systems. In mathematics , negative numbers ...

Mitch Albom - JUST ADDED! GET "FIRST PHONE CALL" SIGNED... - Facebook
JUST ADDED! GET "FIRST PHONE CALL" SIGNED FOR CHRISTMAS FROM COMFORT OF YOUR HOME! As it is snowing this weekend in Michigan, the nice folks ...

Moving to a More Secure Standard: Please Update your Apps To Support Certificates Signed with SHA-2
We’re updating our encryption requirements for Facebook-connected apps to reflect a new and more secure industry standard. As a result, apps ...

Official document to be signed. - Flickr - Photo Sharing!
Explore Andy on Flickr's photos on Flickr. Andy on Flickr has uploaded 1958 photos to Flickr.

Stevie Wonder - Signed, Sealed, Delivered, I'm Yours (HD) - YouTube
UPDATE: I HAVE HAD TO DELETE ON AVERAGE 50 POLITICAL COMMENTS PER DAY, AND I AM GETTING SICK OF IT. THIS IS ABOUT MUSIC, NOT ANYBODY'S POLITICAL ...

Floyd Mayweather says fight with Manny Pacquiao not set yet and no deal has been signed
Unbeaten fighter Floyd Mayweather said Sunday that neither he nor Manny Pacquiao have signed a deal for a May mega-fight, but he still hopes ...

ALP leader John Robertson signed Man Monis letter
The state's Opposition Leader, John Robertson, has admitted he signed a letter conveying a request on behalf of Man Haron Monis, three years ...

Chris Hemsworth has signed on for an iconic role in the 'Ghostbusters' reboot
Janine Melnitz. It’s a big role for the Aussie star who just can’t seem to lose right now. Fans might remember Janine had an infatuation with ...

National Partnership funding agreements signed, funding fight continues
The state's community legal centres will find out within days how their funding has been impacted by a new national partnership agreement that ...

Myer confirms Jodi Anasta, Kris Smith and Rachael Finch signed on for another year - DailyTelegraph Search ...
MYER’S celebrity stable is safe for another year with the department store renewing the contracts of three of its ambassadors.

Resources last updated: 2/17/2016 1:18:20 PM