View Single Post
Old 2013-10-06, 17:09   #19
Just call me Henry
henryzz's Avatar
Sep 2007
Cambridge (GMT/BST)

173416 Posts

TheCount's calculations fail for b != 2 as base 2 is assumed in the calculation of 1751.

I have spent today delving though the paper linked in and have implemented the method it uses in the spreadsheet. It produces the same results as the original method.
I also extended both methods to support ranges of n based on integration. At some point in the future I might add ranges of k as well.

In the attached spreadsheet the first tab is the original method. The second is the new method and also includes a proth weight calculation. It wouldn't surprise me if either the new or old methods have regions where they give better/worse estimates than the other. If you find they are different for certain input then please let me know. I have been unable to find an example.
I left it set up for R620 using Gary's data sieved to 5T.
Attached Files
File Type: zip odds of (15.4 KB, 142 views)

Last fiddled with by henryzz on 2013-10-06 at 17:10
henryzz is online now   Reply With Quote