Does anyone know if they have a 64-bit Microsoft Office ?????

B

BTS

Thanks everyone for you informative advice.

Harlan, you are correct. Thanks for maxima.

For example, I have a value 1.0000000000077500000000000000, that should
change when some parameter is altered by 0.01, but I still get the same
value. In this case, anything beyond the 15th decimal place is zero, which is
not correct. Which tells me that the original value is probably not correct
either.

I found a C++ definition for 128 bit floating point (decimal128) that gives
33 digits, which is at least what I require - but don't want to get back into
programming.

Bernad, thanks for your input onf the Excel Add In. This looks most
promising - I have a large array of numbers and many iterations. I'll give it
a try and let you all know, how this turns out.

Thanks & Happy 4th.
Ben
 
D

Dana DeLouis

Hi. Although others are correct, Excel can do the following if interested.
The following is probably not what you want, but may give you some ideas.
I have a library of such Excel routines that I use often enough.
This will give you a taste. Again, this is put together real quick, as I
use slightly different techniques.

Sqrt of:
999,999,999,999,999,999,999,999,999
is:
31,622,776,601,683.7933199889354290
That number Squared is:
999,999,999,999,999,999,999,999,999.00000
= = = = =

Sub Demo()
'//By: Dana DeLouis
Dim n
Dim g
Dim T As Variant
Dim Ans As Variant

T = CDec(2)

n = "999999999999999999999999999"
g = CDec(Sqr(n))

Ans = g - (g * g - n) / (T * g)

Debug.Print "Sqrt of: "
Debug.Print FormatNumber(n, 0, , , vbTrue)
Debug.Print "is: "
Debug.Print FormatNumber(Ans, 16, , , vbTrue)
Debug.Print "That number Squared is: "
Debug.Print FormatNumber(Ans * Ans, 5, , , vbTrue)
Debug.Print " = = = = ="
End Sub
...I have a large array of numbers and many iterations.
Click to expand...

My favorite Excel program that I ever wrote is a Fourier Transform program
written at Excel's full 28 digit precission.
 
B

BTS

Thanks Dana.

I found the Bernad's suggestion
(http://digilander.libero.it/foxes/index.htm), the most appropriate for me.

I have and AMD 64-bit notebook running Office 2007, under Windows XP.
Previously I could not get beyong 15 significant digits (e.g.
1.000000000695880000000000000000), and a single Excel iteration takes about 3
seconds (?), maybe less.

With the Excel AddIn I do get upto 250 significant digits (
1.00000000069587869465092287704840406672297412319934712363234523078186967155436522990913289431011888241671681026138144200535216051949144401758976016014625095306872312244550223239253052159233354755536364053698830310490711335183529576049665606772468407)
Wow! but it takes anywhere from 4 to 12 hours for a single iteration.

I am very pleased with this precision, it is well worth waiting for.

Thank you all for all your suggestions.

Best,
Ben
 
D

Dana DeLouis

Wow! but it takes anywhere from 4 to 12 hours for a single iteration.

Ouch! Glad it's working though.
I'd be curious to learn what function you are trying to Solve.

As a side note, I would be curious to learn how long it takes to multiply
two 127-digit numbers with your program. Is it quick? As a hobby, I like
to mess around with code that multipies two 4,000 digits (or larger) numbers
via Fourier Transforms. I would be interested in your times, although I
don't know what algorithm it uses. Lately, I've been working on trying to
speed up my division function, which is done via repeated calls to a FFt
multiply routine. Thanks. :>)
 
B

BTS

Hi Dana,
I'd be curious to learn what function you are trying to Solve.
Click to expand...

I'm modeling the time dilation in a gravitational field, to determine how
time dilation affects the gravitational force - this is heresy by modern
physics, because if I'm correct, gravity may not have any quantum mechanical
properties. You can look up my published & non-proprietry work at
http://www.iseti.us/ (note I'll be revamping the website soon).

The model is not a function, but a numerical model consisting of 10 sheets
of formulae. A total of 780,000 cells of multiplications, divisions, ^2,
square roots, addition, and subtractions. This version takes between 4 to 8
hours. I have another version, that consists of 500,000 cells of formulae
which took 54 hours to complete, but I ran into an error. I'm not going to be
able to debug this model just yet without a second pc, because Excel halts if
I accidently click on it - and have done this too many times. Note that these
times are for a single Excel iteration.
As a side note, I would be curious to learn how long it takes to multiply
two 127-digit numbers with your program. Is it quick?
Click to expand...

I believe so, because the answer to the multiplication, division, square
root or even ^2 of two 250-digit numbers appears as soon as you complete
entering the formulae.
As a hobby, I like
to mess around with code that multipies two 4,000 digits (or larger) numbers
via Fourier Transforms. I would be interested in your times, although I
don't know what algorithm it uses. Lately, I've been working on trying to
speed up my division function, which is done via repeated calls to a FFt
multiply routine. Thanks. :>)
--
Click to expand...

I cannot answer this because I'm using someone else's ready built functions
(http://digilander.libero.it/foxes/index.htm), so don't know what algorithms
are being used.

Best,
Ben
 
Top