New phi for homogeneous Cunningham numbers

[wpolly]

I have modified akruppa's phi program to work with a^n +/- b^n numbers.
here is the source file, compile with GCC with -lgmp.

[VolMike]

As usual,compilation problems appear.
Trying to compile with dev-C++ 4...
Steps done:

1) Unpack gmp 4.2.1 archieve
2) Rename gmp-h.in to gmp.h
3) Path to gmp.h was added as "include" directory

While compiling, the errors line "line XX ,parse error" appear.

How can I compile this code?

[wpolly]

Originally Posted by VolMike

As usual,compilation problems appear.
Trying to compile with dev-C++ 4...
Steps done:

1) Unpack gmp 4.2.1 archieve
2) Rename gmp-h.in to gmp.h
3) Path to gmp.h was added as "include" directory

While compiling, the errors line "line XX ,parse error" appear.

How can I compile this code?

try convert the file into DOS mode(CRLF)..

[VolMike]

Originally Posted by wpolly

try convert the file into DOS mode(CRLF)..

I have tried with ConvertCRLF - doesn't help - the same errors.
If you are under Windows XP, would you share binary file compiled on this platform?

[wpolly]

This binary is compiled under Cygwin - thus it need cygwin1.dll and cyggmp-3.dll to run.

[chris2be8]

Is there a later version of phi available anywhere? I've think I've found a bug in this version:

$ phi -deg4 17 313616599 1 52768558671518362205861309316257680653426149189122032710729071712003321614920605612513 n: 52768558671518362205861309316257680653426149189122032710729071712003321614920605612513
# 313616599^17-1^17, difficulty: 169.93, skewness: 2356672.75, alpha: 0.00
# cost: 8.99361e+15, est. time: 4.28 GHz days (not accurate yet!)
skew: 2356672.753
c4: 1
c0: -30845876999193307850169799
Y1: -1
Y0: 3033857681268212293634352084667698572482999
m: 3033857681268212293634352084667698572482999
type: snfs

Which is ridiculous. A poly like the following will be much better:

n: 52768558671518362205861309316257680653426149189122032710729071712003321614920605612513
c4: 313616599
c0: -1
m: 9673779037659330951530253934893601
# m is 313616599^4

Chris

PS. Is there a way to search for attachment names matching a string? The forum search engine doesn't appear to.

[Batalov]

You could have fun with an octic! ;-)

n: 52768558671518362205861309316257680653426149189122032710729071712003321614920605612513
type: snfs
skew: 1
# t^8+t^7-7*t^6-6*t^5+15*t^4+10*t^3-10*t^2-4*t+1, t=(x^2+1)/x
c0: 1
c1: -4
c2: -10
c3: 10
c4: 15
c5: -6
c6: -7
c7: 1
c8: 1
Y1: 313616599
Y0: -98355371168326802

but of course SIQS will do her in a few minutes, anyway!

[R.D. Silverman]

Originally Posted by Batalov

You could have fun with an octic! ;-)

n: 52768558671518362205861309316257680653426149189122032710729071712003321614920605612513
type: snfs
skew: 1
# t^8+t^7-7*t^6-6*t^5+15*t^4+10*t^3-10*t^2-4*t+1, t=(x^2+1)/x
c0: 1
c1: -4
c2: -10
c3: 10
c4: 15
c5: -6
c6: -7
c7: 1
c8: 1
Y1: 313616599
Y0: -98355371168326802

but of course SIQS will do her in a few minutes, anyway!

Sign me confused. I thought that the number to be factored is
313616599^17-1; At 145 digits this is much too big for SIQS.

[bsquared]

Originally Posted by R.D. Silverman

Sign me confused. I thought that the number to be factored is
313616599^17-1; At 145 digits this is much too big for SIQS.

The cofactor is only 86 digits.

[CRGreathouse]

Originally Posted by R.D. Silverman

Sign me confused. I thought that the number to be factored is
313616599^17-1; At 145 digits this is much too big for SIQS.

Sure, but its cofactor 52768558671518362205861309316257680653426149189122032710729071712003321614920605612513 is just a C86.

Edit: bsquared beat me to it.

[CRGreathouse]

Factors as 40198613071456517734490522684447 * 1312696002165987096298388275017136243160743778951559679.