Closed AndrewGibbs closed 3 days ago
`sing_accuracy' is also an endgame parameter. Every zero with condition number above that value is considered singular.
OK it appears that I was mistaken about the human cause of the issue - but there is still an issue. If I run
@var x y
F = System([x^2+y^2+1, 2x+3y-1], variables = [x, y])
solve(F, EndgameOptions(; sing_accuracy = 1e2))
Then I get the MethodError: no method matching EndgameOptions(; sing_accuracy=100.0)
Thanks!
This is indeed weird. Thanks for noticing. I will look into it...
What is the version of Hc.jl that you use? Because in earlier version sing_accuracy
was not implemented. In the most recent version of the software EndgameOptions(; sing_accuracy = 1e2)
goes through on my computer.
I'm using HomotopyContinuation v2.6.4. This was downloaded from the official Julia registry on Tuesday.
I think this is the problem. For some reason, the most recent version (2.9.3) is not automatically being installed on your device. I just tried a fresh installation and it gives me 2.9.3. There seems to be some compatibility conflict. Can you manually upgrade to 2.9.3?
Perhaps it is because I am using an older version of Julia (1.7.2); are there dependencies on newer features? Uninstalling and reinstalling gives this message:
The latest HomotopyContinuation version requires at least Julia 1.8
In the documentation (and corresponding docstring), the endgame parameter
singular_min_accuracy = 1e-6
appears incorrectly assing_accuracy = 1e-12
.