Running GGNFS

[jasonp]

you can start here, which is a fairly in-depth treatment of your question.

[em99010pepe]

Originally Posted by jasonp

you can start here, which is a fairly in-depth treatment of your question.

I already printed "Optimal Parameterization of SNFS" by R.D Silverman, that one will be next. Thank you.

[Jeff Gilchrist]

Originally Posted by Andi47

I just downloaded SVN 413 from Jeff's page. Is it intended, that the 15e binary says that it is SVN 406?

Not sure about the endless loop, but the SVN 406 seems to be correct. The source file gnfs-lasieve4e.c that it is built from was last modified with SVN revision 406.

Jeff.

[em99010pepe]

Attached an excel-table (updated today) which gives estimations for SNFS difficulty from 126 up to 252, based on factorizations I have done and factorizations that have been done by others.

[R.D. Silverman]

Originally Posted by em99010pepe

Attached an excel-table (updated today) which gives estimations for SNFS difficulty from 126 up to 252, based on factorizations I have done and factorizations that have been done by others.

My paper on optimal parameterization gives additional data.

Note that trying to fit a polynomial (and a fairly high degree one)
to a function known NOT to be polynomial makes very little
sense. It would be much better to fit the o(1) term in the known
asymptotic run time. I gave a rump session talk on this at crypto
back in the late 90's.

[em99010pepe]

Originally Posted by R.D. Silverman

My paper on optimal parameterization gives additional data.

Note that trying to fit a polynomial (and a fairly high degree one)
to a function known NOT to be polynomial makes very little
sense. It would be much better to fit the o(1) term in the known
asymptotic run time. I gave a rump session talk on this at crypto
back in the late 90's.

I printed your paper and it's been very helpful. At the beginning I was concentrated on chapter 4 but when I get more free time I will read the rest.

I know the polynomial fit makes almost no sense at all, I was just playing around with the variables.

Today I received more data to include on the excel-table, as soon as I get some free time I will share it (updated) on google pages.

[xilman]

Originally Posted by em99010pepe

I printed your paper and it's been very helpful. At the beginning I was concentrated on chapter 4 but when I get more free time I will read the rest.

I know the polynomial fit makes almost no sense at all, I was just playing around with the variables.

Today I received more data to include on the excel-table, as soon as I get some free time I will share it (updated) on google pages.

Note that a polynomial fit might actually be quite useful in practice for interpolation between well-characterized data points, as long as it is a reasonably low-degree polynomial. To that extent it makes sense, especially if it is much easier to implement than something more justified by theory. Using it for extrapolation is very likely to lead you astray.

Paul

[R.D. Silverman]

Originally Posted by xilman

Note that a polynomial fit might actually be quite useful in practice for interpolation between well-characterized data points, as long as it is a reasonably low-degree polynomial. To that extent it makes sense, especially if it is much easier to implement than something more justified by theory. Paul

Once you fit the o(1) term in the exponent for L(N, c) extrapolation
becomes quite easy on a hand calculator. You can ALSO fit an
extrapolation function for the factor base size, since it is just L(N, c/2)
with a different o(1) term.

[xilman]

Originally Posted by R.D. Silverman

Once you fit the o(1) term in the exponent for L(N, c) extrapolation
becomes quite easy on a hand calculator. You can ALSO fit an
extrapolation function for the factor base size, since it is just L(N, c/2)
with a different o(1) term.

Agreed.

It's the fitting that might be a little more difficult in practice. It's trivial to fit a polynomial to data in Excel or its Open Office counterpart. If you have a decent stats tool available, fitting other curves is fairly straightfoward but many people wouldn't want to track down, install and learn how to use such a tool when they already have Excel to hand. To be honest, I can't say I blame them if all they want is a one-off solution to a one-off problem.

Paul

[em99010pepe]

Attached an excel-table (updated today) which gives estimations for SNFS difficulty from 126 up to 252, based on factorizations I have done and factorizations that have been done by others
You can send me yours to em99010pepe at gmail dot com.

[em99010pepe]

The file is available here: https://sites.google.com/site/em99010pepe/snfs