# BigInteger exponents - how?

Stan Wilson

Greenhorn

Posts: 4

posted 11 years ago

- 0

Hi all, Have rattled my brain and cannot work out how to do a BigInteger a

to the power of e, where both a and e are BigIntegers.

Any help appreciated.

to the power of e, where both a and e are BigIntegers.

Any help appreciated.

Julian Kennedy

Ranch Hand

Posts: 823

posted 11 years ago

- 0

OK, well, here is some code:

Now, that sure as hell isn't the most efficient way to do it but it will work for positive numbers. You may need to go on holiday while it does its calculation for large values of exponent but it will eventually finish ... eventually. You might want to look up a more effective mathematical formula for calculating exponents if you really need this functionality. I expect my terminology's a bit suspect too, but hey! And I haven't exactly tested it thoroughly.

How does it look?

Jules

[ August 19, 2004: Message edited by: Julian Kennedy ]

Now, that sure as hell isn't the most efficient way to do it but it will work for positive numbers. You may need to go on holiday while it does its calculation for large values of exponent but it will eventually finish ... eventually. You might want to look up a more effective mathematical formula for calculating exponents if you really need this functionality. I expect my terminology's a bit suspect too, but hey! And I haven't exactly tested it thoroughly.

How does it look?

Jules

[ August 19, 2004: Message edited by: Julian Kennedy ]

Julian Kennedy

Ranch Hand

Posts: 823

posted 11 years ago

- 0

Having thought about this overnight it's quite likely that, for large exponents, the above code will not actually finish executing in our lifetime or, perhaps, ever.

Note that the time taken to execute BigInteger.pow() increases exponentially as the exponent increases (it's not linear as I thought my algorithm would theoretically be). on my box BigInteger.pow(100000) takes 1 second; pow(200000) takes 4 seconds and pow(500000), 27 seconds. My program does pow(100000) in 11 seconds and pow(200000) in 51 seconds.

It would therefore be more efficient to use BigInteger.pow(int) and use the modulus of Integer.MAX_VALUE in the above algorithm, rather than multiplying in a loop.

A really simple solution that would make sense for practical values (which, having gone through the hoops may be the answer you are looking for anyway) would be:

HTH

Jules

Note that the time taken to execute BigInteger.pow() increases exponentially as the exponent increases (it's not linear as I thought my algorithm would theoretically be). on my box BigInteger.pow(100000) takes 1 second; pow(200000) takes 4 seconds and pow(500000), 27 seconds. My program does pow(100000) in 11 seconds and pow(200000) in 51 seconds.

It would therefore be more efficient to use BigInteger.pow(int) and use the modulus of Integer.MAX_VALUE in the above algorithm, rather than multiplying in a loop.

A really simple solution that would make sense for practical values (which, having gone through the hoops may be the answer you are looking for anyway) would be:

HTH

Jules

I agree. Here's the link: http://aspose.com/file-tools |