number fail help..i get 5.56e+031

[wipped]

Hello

I am making a script of collatz.

But at some high numbers the number transforms in ...e+...

how can i make sure that this wont happen or how can i transform it back to an normal number?

Thanks anyways

Robin Jan

[MvGulik]

whatever Edited February 7, 2011 by MvGulik

[Melba23]

wipped,

The problem arises because AutoIt cannot deal with integers above 0x7FFFFFFF (2147483647). Anything above this is displayed in the exponential format which will not produce the degree of accuracy you require. Exponential notation works in powers of 10:

100 = 1.0 e2 = 1 times 10 to the power 2

Looking at the Collatz conjecture in my maths textbook:

"Starting with 27 takes 111 steps to reach 1 with a maximum of 9232"

So I can well believe that you could easily breach the Autoit limit with a not unreasonably sized seed number. And using exponential notation will not give you the level of accuracy you need to continue with the calculation.

Sorry to be the bearer of bad tidings, but AutoIt is just not suited to this type of mathematical calculation - you need another language.

M23

[Mat]

wipped,

The problem arises because AutoIt cannot deal with integers above 0x7FFFFFFF (2147483647). Anything above this is displayed in the exponential format which will not produce the degree of accuracy you require. Exponential notation works in powers of 10:

100 = 1.0 e2 = 1 times 10 to the power 2

Looking at the Collatz conjecture in my maths textbook:

"Starting with 27 takes 111 steps to reach 1 with a maximum of 9232"

So I can well believe that you could easily breach the Autoit limit with a not unreasonably sized seed number. And using exponential notation will not give you the level of accuracy you need to continue with the calculation.

Sorry to be the bearer of bad tidings, but AutoIt is just not suited to this type of mathematical calculation - you need another language.

M23

try the BigNum UDF. It's amazing and surprisingly speedy.

Mat

[wipped]

tricky ...

- The number your getting is just displayed in a way that makes sens for large numbers. use stringformat() to get it displayed in a other way.

- I don't think you can do a collatz script with floats(or AutoIt double).

- To get passed the normal in64 limitation your probably going to have to setup your own math functions.

float_1()
Func float_1()
    Local $n, $sOut
    For $i = 1 To 129 Step 32
        $n = 2 ^ $i
        $sOut = String($i) & ') ' & String($n) & ', ' & StringFormat('%20.0f', $n)
        ConsoleWrite($sOut & @CRLF)
    Next
EndFunc
;; output:
;~ 1) 2,                    2
;~ 33) 8589934592,           8589934592
;~ 65) 3.68934881474191e+019, 36893488147419103000
;~ 97) 1.58456325028529e+029, 158456325028528680000000000000
;~ 129) 6.80564733841877e+038, 680564733841876930000000000000000000000

Looks pretty heavy...

i will try making this function working

Thanks for answering

PS if anybody has an other solution just post it

[Melba23]

wipped,

try the BigNum UDF.

Good call from Mat - give it a try.

M23

[wipped]

try the BigNum UDF. It's amazing and surprisingly speedy.

Mat

Hmm thanks for the reply but i dont really know how to use it can somebody help me?

Robin

[wipped]

i read the inside but it are kinda big scripts, i think my collatz calculator will be pretty slow if i use it

[MvGulik]

whatever Edited February 7, 2011 by MvGulik

[Mat]

Hmm thanks for the reply but i dont really know how to use it can somebody help me?

Robin

All the numbers are stored as strings, so there is no limit to their size other than the windows string size and the memory (I haven't tried though). The only problem is that you have to call functions _BigNum_* rather than use the in-built operators. I haven't used it for a while but I believe there are all the functions you'll need there. Jennico wrote some very rapid root functions if you read further down the thread.

[Mat]

i read the inside but it are kinda big scripts, i think my collatz calculator will be pretty slow if i use it

It's big, but its fast. I tried making my own using arrays etc but it couldn't scratch it for performance.

If you want speed you're using the wrong language. If you still want to use AutoIt you might want to try making a c++ dll with an AutoIt wrapper.

[wipped]

It's big, but its fast. I tried making my own using arrays etc but it couldn't scratch it for performance.

If you want speed you're using the wrong language. If you still want to use AutoIt you might want to try making a c++ dll with an AutoIt wrapper.

im just using it for fun so i will try all the stuff tomorrow but now im gonna get some sleep...

cya

[Mat]

im just using it for fun so i will try all the stuff tomorrow but now im gonna get some sleep...

cya

I don't know what you're trying to do unfortunately or i'd help more... I might do some googling tonight

[wipped]

I don't know what you're trying to do unfortunately or i'd help more... I might do some googling tonight

lol i thougt it was like 12 pm but its only 10

do you have xfire or msn so i can add you then we can discus some.

[Mat]

lol i thougt it was like 12 pm but its only 10

do you have xfire or msn so i can add you then we can discus some.

Nope... at least I don't think so...

It is my bed time though, so I'm off. If you want then pm me. I've read wikipedia and it sounds like a simple enough program, a bit like projecteuler.net would give you, so I don't mind helping out

If not I'll pm you if I get anything done, I'm done with tests on friday so I'll have plenty of time.

Mat

[jchd]

The problem arises because AutoIt cannot deal with integers above 0x7FFFFFFF (2147483647). Anything above this is displayed in the exponential format which will not produce the degree of accuracy you require.

I beg to differ.

AutoIt will switch from signed 32-bit to signed 64-bit integers when needed for arithmetical operations. Several operations of functions still can't cope with those 64-bit ints, notably bitwise functions, Hex(), ...

Local $n = 0x7fffffff
For $i = 1 to 1234567890 Step 54321
    ConsoleWrite($n & @LF)
    $n += $i
Next

[Melba23]

jchd,

Oops!

Read the wrong line in the Help file:

Number range (integers): 64-bit signed integer

Hexadecimal numbers: 32-bit signed integer (0x80000000 to 0x7FFFFFFF)

So I claim half a point because I did specify the limit in Hex!

M23

[jchd]

0x4772616E74656421

[Mat]

jchd,

Oops!

Read the wrong line in the Help file:

Number range (integers): 64-bit signed integer

Hexadecimal numbers: 32-bit signed integer (0x80000000 to 0x7FFFFFFF)

So I claim half a point because I did specify the limit in Hex!

M23

Ermm... Where did you see that? Only I did a bit of testing using this:

$n = 2 ^ 48
$i = 2 ^ 49
While 1
    While Not StringInStr(String($i + $n), "e")
        $i += $n
        ToolTip($i)
    WEnd
    If $n = 1 Then ExitLoop
    $n /= 2
WEnd

ToolTip("")
ClipPut($i) ; 999999999999999

And got 999999999999999 which is 0x38D7EA4C67FFF in hex and 11100011010111111010100100110001100111111111111111 in binary and as you would have guessed not a normal computer max, more a human one.

That number itself is somewhere between 2^49 and 2^50.

So. You're both (nope... reading it again gets just Melba i'm afraid ) wrong, reading helpfile entries that are slightly outdated - talking about converting from v2 and most noticeably the line length limit (now removed as of 3.3.4.0).

I win

Edit: Woops, beaten to the post (literally) there, and I think I got some of conversions a bit wrong...

Edited June 16, 2010 by Mat

[MvGulik]

whatever Edited February 7, 2011 by MvGulik