Solution time a shade over 1 second. I thought I’d get it under 1 second by running in XL 2000, but it was almost exactly the same time (to my surprise).

[VB}
Function pandigital(ByVal val As Double, LorR As String) As Boolean
Dim DigitA(1 To 9) As Long, i As Long, j As Long, k As Long, CheckVal As Double, NCheckval As Double

k = (Log(val) / Log(10))
CheckVal = val

If k .NE. 9 Then
If LorR = "R" Then
pandigital = False
Exit Function
Else
CheckVal = val * 10 ^ (9 - k)
End If
End If

For i = 1 To 9
NCheckval = Int(CheckVal / 10)
j = CheckVal - NCheckval * 10
CheckVal = NCheckval

If j = 0 Or j .GT. 9 Then
pandigital = False
Exit Function
End If

If DigitA(j) = 1 Then
pandigital = False
Exit Function
Else
DigitA(j) = 1
End If

Next i
pandigital = True

End Function
[VB]

I could never get those BigNum routines to work. They are called with 3 parameters, with the 3rd one not optional. I could never decipher why you needed three parameters to operate on two numbers (or strings). Still can’t. The author did not assign any meaningful names to his variables!

And not being able to load the form made me unable to check out anything.

Is it straight forward? I suppose somewhat. I get the additions certainly, and The Foxes website walks through a large multiplication at
http://digilander.libero.it/foxes/MultiPrecision.htm#Packet

I’d find it more straight forward for understanding the whole process if both their multiplier and their multiplicand had to be packetized.

…mrt

Interesting stuff, I spent a little bit of time reading up on this, this PM. There is an addin which claims to work with over 2 billion digits, so there you go! (the link to the site is at the bottom of Tushar’s pages)

Is it not a relatively rival task to just use 2 string in stead of one? (looking at Tushar’s code away). Having said that I am extreamly confident that your understanding of this is much much much much^10000! better than mine, and there is a reason why this is not straight forward?

The Euler problems are famous for asking for answers that require precision longer than 32 bits. Math done that way can be done on strings, and the answers are strings. For instance, AddAsStrings takes two numbers and adds them in columns, with carrys, from the right side, just as you did in elementary school. I do it one column at a time, Tushar does many at a time (his is much, much the faster), but for both of us, the result is a string representation of the number. The Foxes team addin can do math up to 250 significant digits in several hundred esoteric functions. All of it is string manipulation.

The problem here is that the Fibonacci number of the Answer is longer than Excel can handle as a string, and larger than Excel can handle as a double, and we need to be accurate at the front and the back. The middle is of no consequence to Euler . If you look up the Excel limit for string size, it’s about 64K characters. The length of the answer is longer than that, and we need to look at it’s first nine digits in the doubles, and its last nine digits in the longs. If there is a LargeNumber API for us, I’ve not stumbled across it.

…mrt