quote:
-----------------------------------------------------------
Original post by CobraA1
Oh, and by the way, "negative energy" (defined as matter with negative mass) can be used to bring matter to light speed, but nobody''s come up with a way to create this type of matter, not even theoretically. Even the anti-particles still have positive mass. As far as anybody can tell, matter that "blows" gravitationally has never been observed or created. So much for light-speed jets.
-----------------------------------------------------------

Negative energy hasn''t got anything to do with negative mass (or hasn''t got more to do with negative than with positive mass). Anti-particles have negative mass, but negative mass is positive energy.

WRONG. Anti-particles have negative CHARGE, not negative MASS. And sorry if I confused anybody with my re-definition of negative energy. I''ll try to be more clear on my definitions in the future. I meant to say that negative mass has not been measured or created. So, don''t expect any light-speed spacecraft anytime soon. Wait a second . . . Aren''t energy and mass related? E=M*c^2, right? I think ga is confusing energy with charge here. My definition stands, I guess .

Cobra: No, it wasn''t _my_ definition. Someone who added to ga''s, Vet I think it was said almost that exactly (before page 14 somewhere I think.) And why would it occur any less frequently than any other pattern?


Lack

Christianity, Creation, metric, Dvorak, and BeOS for all!

Well, let''s see what type of file would have all zeros. Definatly not an executable. It would be an ugly picture. Zip files would never be all zeros. Not text files. No audio format would have all zeros, or 3d format. Let me reiterate my definition of a ''random file'':

By ''random data is not compressible'', we mean that every file of length n is not compressible by the same algorithm. This also means that on average, over all files of the same length, you would see no gain. End of story.


Mike

Hey people, I just figured out a great compression scheme.

The compression scheme involves a little binary shifting and BS and can compress a 10 meg file into a -32K file. I wont be releasing a demo though because I want to first start one of the longest (if not longest) thread in gamdev.net history!

Thanks kieren_j, I can't stop laughing.

-Icarus

Edited by - Icarus on 4/25/00 7:25:22 PM

quote:Original post by Anonymous Poster
--If all files were just zeroes, wouldn''t that be wonderful? And how many files do you see having all zeroes just by chance? Intelligence, please, not excuses. We can''t depend on chance patterns to compress random files.

How many files with any other combination do you see happening by chance? If it''s random, each file would occur just as often as any other file.

Also, in someone else''s definition of a purely random file, every bit pattern of any size occurs just as frequently as any other. Is this even possible?

quote:Original post by Vetinari

Well, let''s see what type of file would have all zeros. Definatly not an executable. It would be an ugly picture. Zip files would never be all zeros. Not text files. No audio format would have all zeros, or 3d format. Let me reiterate my definition of a ''random file'':

See above.

quote:

By ''random data is not compressible'', we mean that every file of length n is not compressible by the same algorithm. This also means that on average, over all files of the same length, you would see no gain. End of story.

How can you infer your second statement from your first? By your first, a situation where only one file is non-compressible and all others compress very well fits the statement. And that would definitely average out to a gain.

Furthermore, if your algorithm can see that for the particular file there is no gain, and then just doesn''t compress it, then all of the files put together still average out to a gain.



Lack

Christianity, Creation, metric, Dvorak, and BeOS for all!

quote:

Original post by LackOfKnack

---------------------------------------------------------
Original post by Anonymous Poster
--If all files were just zeroes, wouldn't that be wonderful? And how many files do you see having all zeroes just by chance? Intelligence, please, not excuses. We can't depend on chance patterns to compress random files.
---------------------------------------------------------

How many files with any other combination do you see happening by chance? If it's random, each file would occur just as often as any other file.


answer to question: extremely rarely, even with randomly generated files.

answer to statement: Huh? The chances of identical files (especially large ones) generated randomly are close to zero.

quote:
Also, in someone else's definition of a purely random file, every bit pattern of any size occurs just as frequently as any other. Is this even possible?

I think what the person is trying to say is that no particular pattern appears more than others, so compression algorithms that depend on frequency would fail.

i.e. if you use 2 bits, '00' '01' '10' and '11' should appear with the same frequency, with none of them appearing to have a statistical advantage over the others.

quote:

------------------------------------------------------
Original post by Vetinari

Well, let's see what type of file would have all zeros. Definatly not an executable. It would be an ugly picture. Zip files would never be all zeros. Not text files. No audio format would have all zeros, or 3d format. Let me reiterate my definition of a 'random file':

--------------------------------------------------------
See above.

already did.

quote:
--------------------------------------------------------

By 'random data is not compressible', we mean that every file of length n is not compressible by the same algorithm. This also means that on average, over all files of the same length, you would see no gain. End of story.
----------------------------------------------------------

How can you infer your second statement from your first? By your first, a situation where only one file is non-compressible and all others compress very well fits the statement. And that would definitely average out to a gain.

You're right, the first statement does not infer the second.
However, you are also wrong; you are infering that he is claiming that only one file can't be compressed. He didn't say that! He said that not all files can be compressed. And there are a lot more than one of them.

quote:
Furthermore, if your algorithm can see that for the particular file there is no gain, and then just doesn't compress it, then all of the files put together still average out to a gain.

Since there are a lot more than one uncompressible files, they ean easily average out to no gain statistically.

Edited by - CobraA1 on 4/26/00 12:55:38 PM

quote:Original post by CobraA1

answer to question: extremely rarely, even with randomly generated files.

answer to statement: Huh? The chances of identical files (especially large ones) generated randomly are close to zero.

Exactly, so all zeros would occur just as frequently as any other pattern.

quote:
I think what the person is trying to say is that no particular pattern appears more than others, so compression algorithms that depend on frequency would fail.

i.e. if you use 2 bits, ''00'' ''01'' ''10'' and ''11'' should appear with the same frequency, with none of them appearing to have a statistical advantage over the others.

I know, but what I''m saying is can you even _have_ a file where every bit pattern of each size occurs of the same frequency as other patterns of the same size? Like is it possible to find a string wherein you have:

00 6
01 6
10 6
11 6

but also:

000 3
001 3
010 3
011 3
100 3
101 3
110 3
111 3

and so on, for every byte size? And I''m talking without overlapping patterns, just every three or four or whatever bits is a byte (unlike ga''s results from the arj file.)

quote:
You''re right, the first statement does not infer the second.
However, you are also wrong; you are infering that he is claiming that only one file can''t be compressed.

I wasn''t inferring that, I was just using that to prove that you can''t get the first statement directly from the second.

quote:
He said that not all files can be compressed. And there are a lot more than one of them.

Since there are a lot more than one uncompressible files, they ean easily average out to no gain statistically.

How do you figure? If you have several numbers from 0 and up, even if half of them are 0 (gain) it still averages out positive.


Lack

Christianity, Creation, metric, Dvorak, and BeOS for all!

quote: Original post by Loqi

To keep this ridiculous thread going...

Original post by ga

Anti-particles have negative mass, but negative mass is positive energy.
And if negative mass were positive energy, wouldn't Einstein's little equation be stated: E = /m/ c^2 ? (The pipes don't show up right, but I mean absolute value).


Edited by - Loqi on 4/25/00 10:55:34 AM

Edited by - Loqi on 4/25/00 10:56:27 AM

Einsteins equation is correct, it's the amount of the mass, so if the mass = -1, the energy is /-1/c^2 = 1*c^2 = c^2, positive energy.
btw, the correct equation for moving particles is E^2 = m^2*c^4 + p^2*c^2 and m^2 will always be positive.

Visit our homepage: www.rarebyte.de.st

GA

Edited by - ga on 4/27/00 9:45:26 AM

Loqui, I don''t know if Einstein''s original equation was with /m/, but I don''t think he knew of the negative mass of positrons.

Visit our homepage: www.rarebyte.de.st

GA