35 replies to this topic

[rjstephens]

i am working on my rpg engine, and i wanna know, how do i compress a matrix? I see you guys talking about compressing matrices. how have you guys done it and how can i do it?

[BiTwhise]

It's basically about packing more than 1 number into each cell, "shifting" the digits left by multiplying with a power of 10

Say that all you numbers are less than 10, you could pack 15 numbers into one cell (seeing as how Casio's varables holds 15 digits), plus you could even use the exponent to hold 2 numbers.
if you had the numbers 3, 2, and 8
you would put them at their designated cell by assigning it: 3 + 2 * 10 + 8 * 100, and so on
you could retreive them through a loop, extracting the most or least significan number, removing it from the mantissa, and "shifting" the mantissa up or down to make the next number ready for extracion

obviously, you could chose to multiply by 100, or 1000, or whatever size your numbers are, and naturally this will make less numbers fit in one cell

you don't have to chose a power of 10 multiplier either, you could use any number that might fit the size of your numbers

[CrimsonCasio]

I use my decimile notation, which stores numbers in the decimile places of each cell, then when needed you just look at the decimiles and get your number. its really quite easy.

this is the link to the thread on that topic: here

[ross8653]

i also used it in my tetris program it can take 16 digits per cell
do this if you dont understand
1234567890123456->A
A (enter) (to show that it wont display all the digits)

you can loop this part if you want
Frac (A/10) (you get the lowest digit 6, multiply by 10 for non decimal)
Int (A/10) -> A (cut the last digit off A, and repeat if you want)

thats basically how it's done both Frac and Int are found under NUM (option f.. something)

[rjstephens]

thanks heaps guys

think i'll write a tutorial or something.

[Mohamed]

Where's dscoshpe and his algorithms and software?
Maybe you'd look for his site.

[BiTwhise]

i also used it in my tetris program it can take 16 digits per cell

the calc variables hold 15 digits

9 byte packed bcd numbers
first byte:
bits:
0 - 3: exponent seccond digit
4: exponent first digit
5 - 7: sign, and format
seccond byte:
bits:
0 - 3: mantissa, first digit
4 - 7: exponent last digit
byte 3 - 9: mantissa


a packed bcd number holds 2 digits per byte, the mantissa is 7 bytes and 1 nibble, making for 15 digits of precision, if you want more, you have 2 digits and1 bit in the exponent + 1 bit in the sign

[rjstephens]

thanks bitwise. will incorporate this into tutorial

what's discosheep's site? anyone got a bookmark? what's his real name i'll look itup on CPA.

[Brazzucko]

Are you creating one Tutorial?

In the CASIO site there is various tutorials, see it......

see the related topic.....

[rjstephens]

i am creating a tutorial on how to use compression.

it is only a small tutorial.

which casio site are you reffering to?

[BiTwhise]

i am working on my rpg engine, and i wanna know, how do i compress a matrix?  I see you guys talking about compressing matrices.  how have you guys done it and how can i do it?

..
You have no clue how to compress stuff, and then, all of a sudden your making a tutorial..
doesn't sound right

[rjstephens]

i get confident with stuff pretty quickly.

especially something like this I already have all the necessary background knowledge to do, but just didn't now how to do it.

[BiTwhise]

yeah.. I mean, it's not like it's difficult or anything (it's freakin basic.. )

but, why make a tutorial, aren't there enough basic programming tutorials for basic programmin n00bs? (waist of time in'it?)

[Brazzucko]

This could be very helpful to me too

[rjstephens]

its a compression-specific tutorial. If you want i'll make a basic tutorial, but not yet. i'm still busy with writing about 5 programs at once.

it's holidays for me now though so i have more time.

[Pixter]

the calc variables hold 15 digits

weird, my calc (cfx9800gb+) seems to know 10 digits only

[BiTwhise]

It displays only 10 digits,

you can see the other 5 by doing something like this:

123456789012345->A
A_  //this outputs A, 10 first digits
Frac(A/1E5)_ //this outputs the 5 last digits

[Pixter]

nope, it outputs 0. Strange.

[huhn_m]

This may sound silly but did you press the EXP key for the E and
not input the letter? YOu should press the EXP key! a smaller E will appear.

[BiTwhise]

Some older models carry less digits of pressision, but even the fx8000 has 12 or 13 digits..

strange,
perhaps you did something wrong (like inputting the letter E, instead of exonent E.. )

-- EDIT --
sorry huhn_m, didn't see ur post

[CrimsonCasio]

It displays only 10 digits,

you can see the other 5 by doing something like this:

<!--QuoteEBegin-->123456789012345->A<!--QuoteEBegin-->A_  //this outputs A, 10 first digits<!--QuoteEBegin-->Frac(A/1E5)_ //this outputs the 5 last digits<!--QuoteEBegin-->

sorry for the late reply, but I've been trying to get this to work and it wont so I just wrote my own, mabey Pixter is having the same problem I am.

.0123456789123456->A
Lbl 0
10A->A
A-Int A)->A
Int (10A)_     //output sign [SHIFT]+[VARS]+[F4]
Goto 0

that should output all the digits.

[BiTwhise]

Strange, maybe that command doesn't work on the cfx.. only have my afx here, and it works perfectly..

[CrimsonCasio]

I have an AFX...

[BiTwhise]

then it should work fine.. printing the fraction of 123456789012345/100000 should give you 0.12345, surely you must do something wrong.. (like input alpha E instead of exponent E, or input 0.123456789012345, instead of 123456789012345...)

[CrimsonCasio]

oki, I got the .12345 thing, I cant get more though... if it only gets the first 5 what use is it, we can see 10 already?

[BiTwhise]

it's the last 5.... (the 5 you dont see...)

[CrimsonCasio]

ohhhhh............
now it makes sense... :iws:
oh well, mine works to so

[BiTwhise]

yeah, I used your kind of algorithm as well when I did compression way back.. it's more useful for extracting data
just thought this was an easier way of demonstrating how to display more digits of a number than what is printed on screen..

say, for pi, you could do something like this:

pi_                          //displays the first 10 digits
Frac (pi x 1e5)         //gets rid of the first 5 of the fraction + the integer part, all in all the 6 first digits of the number and displays the last 8 (pi is stored only with 14 digits of precission..)

[Pixter]

It still doesn't work, whatever I try
It always displays 0 for the last 5 digits, even if I type in
Frac (1234567890.12345)

[CrimsonCasio]

try my method, mabey it will work.

[rjstephens]

pi related sites:
http://www.angio.net/pi/piquery#find
http://www.angio.net...igits/10000.txt
http://www.hepl.phys...if1000c.data.gz

if anyone finds these useful or interesting, please tell me.

[Pixter]

pi related sites:
http://www.angio.net/pi/piquery#find
http://www.angio.net...igits/10000.txt
http://www.hepl.phys...if1000c.data.gz

if anyone finds these useful or interesting, please tell me.

Too few digits, how can I know the surface of my circle precisely enought with that few digits! lol

[rjstephens]

2 billion!!! come on, you couldn't measure it that precisely!

[Pixter]

lol my text editor crashes when I open that file, I think it's too big and my pc is not fast enough to open it. (but my computer sucks). I will give it a try on the laptop tonight.

I am trying to put as much digits of pi as possible on my calc (just for fun :S).
for every byte I have, I can store one digit.
But now something more interesting: is it possible to compress numbers?
I know that you can write 1000 as 1E3 but I'm looking for better ways to compress.

[huhn_m]

use 1K for 1000 1M for 1,000,000
OPTN->NUM->ESYM

(only for makeing the program smaller.

[Pixter]

Sorry, but maybe I was little bit unclear .
I mean that I'm thinking about how I can put more numbers in one byte. A programm like winzip or so.


btw, can someone send me a program or a game that uses the 15 digits in one matrix-cell trick? I can't get that working . Maybe I'm just too stupid

My email is rick_heuijerjans AT hotmail.com

thank you very much