Calculate big numbers using gmp-dll

[funkey]

Hello,

I wrapped a few functions from the GMP library (http://gmplib.org/) to add and multiplicate huge integers and floats.

Maybe it is useful for someone.

Download: http://www.autoit.de/index.php?page=Attachment&attachmentID=16577&h=ffb11844ecb87e4fef88c05e9f9a3635d88d06f7

Edited November 14, 2012 by funkey

[funkey]

I updated the code for using with float numbers. But I have some troubles with changing the precision. It does not take effect. Can someone please have a look at it and tell me what I'm doing wrong? Thanks

[ValeryVal]

From Topic 3.7 Reentrancy of documentation

"mpf_set_default_prec and mpf_init use a global variable for the selected precision. mpf_

init2 can be used instead, and in the C++ interface an explicit precision to the mpf_class

constructor."

So you have to call mpf_set_default_prec before any init function.

1. Set precision.

2. Set or init value.

Other comment from 7.1 Initialization Functions:

void mpf_set_default_prec (mp bitcnt t prec) [Function]

Set the default precision to be at least prec bits. All subsequent calls to mpf_init will use

this precision, but previously initialized variables are unaffected.

#include "GMP.au3"
Global $sValue1 = "123456230290323489023430484039035832902904890289023482340234239023482349023483490890890859034809809309229342348972349"
Global $sValue2 = "100"
#region precision
_GMPf_SetDefaultPrec(512)
MsgBox(0,"","Default precision: " & _GMPf_GetDefaultPrec() & @LF)
Global $t_Value1 = DllStructCreate($tag_mpf_t), $t_Value2 = DllStructCreate($tag_mpf_t)
Global $p_Value1 = DllStructGetPtr($t_Value1), $p_Value2 = DllStructGetPtr($t_Value2)
_GMPf_Init_Set_Str($p_Value1, $sValue1)
_GMPf_Init_Set_Str($p_Value2, $sValue2)
MsgBox(0,"","Precision of Value1: " & _GMPf_GetPrec($p_Value1) & @LF)
$sText = StringFormat("%sn=n%snnn%sn=n%snnn", $sValue1, _GMPf_get_Str($p_Value1), $sValue2, _GMPf_get_Str($p_Value2))
MsgBox(64, "Test GMPf precision !?", $sText)
_GMPf_Clear($p_Value1)
_GMPf_Clear($p_Value2)
$t_Value1 = 0
$t_Value2 = 0
#endregion precision

[funkey]

Thanks for answering!

That's exactly the way I tried it, but without success!?

But the most important fact is, calculations with big integers work fine. Who needs floats?

[Baraoic]

I have updated your UDF and compiled the latest gmp dll. I initially added the parse functions from the BigNum UDF, but I have found that function has issues. It does all of one arithmetic operation first before doing the next operation, which will cause problems at times. I plan eventually attempting to fix that, but not right now.

Couple notes: by default integer functions only return 128 digits of precision, starting from most significant, but will return correct number of digits reguardless. So it will pad if needed.

Sometimes the float operations return repeating digits when it shouldn't.

Example _GMP_MulFloat("23", "23.4") gives .1999 repeating, but not _GMP_MulFloat("23", "23.5") and I am not sure why.

Included Functions:

_GMP_Parse

_GMP_ParseInt

_GMP_ParseFloat

_GMP_HexToInt

_GMP_IntToHex

_GMP_AddInteger

_GMP_SubInteger

_GMP_MulInteger

_GMP_DivInteger

_GMP_PowInteger

_GMP_SqrtInteger

_GMP_AddFloat

_GMP_SubFloat

_GMP_MulFloat

_GMP_DivFloat

_GMP_PowFloat

_GMP_SqrtFloat

GMP 1.0.au3

gmp-5.1.1.zip

Edited April 9, 2013 by Onichan

[Baraoic]

I have fixed the Parse functions and after some testing they seem to run quite a bit faster as a result. I tried to account for all the various inputs people might use in the Parse function, but I am no mathematician so I might have missed some. I do know it currently doesn't account for more than one level deep power of, for example 5^2^6 isn't calculated correctly as I haven't thought of a good way to do that yet.

If there is anything else I am missing just let me know.

GMP 1.1.au3

Edited April 9, 2013 by Onichan

[funkey]

Thank you for continuing this project! I'm glad someone likes it.

[AZJIO]

The license allows to use this DLL in commercial projects?

[Baraoic]

GMP uses the GNU Lesser General Public License, which from the GMP site says "The license gives freedoms, but also sets firm restrictions on the use with non-free programs." So I would assume it has restrictions for commercial products. No idea about the specifics though.

[Mat]

I do know it currently doesn't account for more than one level deep power of, for example 5^2^6 isn't calculated correctly as I haven't thought of a good way to do that yet.

Take a look I've done a lot of work on parsers since then, but I'm sure for simple expressions the code is still usable. There are C examples, but I seem to remember there being AutoIt code thrown around too.

[nullschritt]

Not to gravedig a thread, but yea, sorta....

It seems the calculations with this library return the same precision as normal autoit math? (as the dis-advantage of them taking several times as long)

 

I need to be able to calculate high precision decimals.

Edited February 6, 2014 by nullschritt

[jchd]

Can you describe your precise needs for arbitrary-precision arithmetic?

[nullschritt]

multi-level encryption. After just a few passes I'm dealing with numbers like 2.43434343452345e+69.

 

Eventually after enough passes the original value becomes irretrievable.

[jchd]

Better rely on proven algorithms scrutinized ad nauseam by professionnal cryptographers. Else you're going to silently shoot yourself in the foot with probability > 0.999999999999999999999

[nullschritt]

Better rely on proven algorithms scrutinized ad nauseam by professionnal cryptographers. Else you're going to silently shoot yourself in the foot with probability > 0.999999999999999999999

 

I know how to code a foolproof algorithm, thanks for the advice though(I took some lessons on cryptography). I would still trust the algorithm I've devoted over 2 years into developing, beyond industry standards, for it's sheer strength compared to them.

I simply want to make it even stronger. I currently only achieve about 81% accuracy when using my maximum desired encryption strength. though I can achieve 100% at levels below maximum.

I'm looking into the bignum udf.

Edited February 6, 2014 by nullschritt

[jchd]

 

I know how to code a foolproof algorithm, thanks for the advice though. I would still trust the algorithm I've devoted over 2 years into developing, beyond industry standards, for it's sheer strength compared to them.

Amen

[nullschritt]

Amen

 

My script contains a hybris AES option, simply to mock it(in comparison of speed). Believe it or not, encryption is just glorified mathematics. It's actually possible to write an algorithm in autoit that uses far more bits, and processes the data quicker. I'll be releasing a peer-to-peer instant messenger soon to showcase the algorithm. 

[Andrey_A_A]

Thanks for the great UDF!

Whether there is a examples Random Functions?

Random-Numbers

Structure gmp_randstate_t ...