pointless computations when converting number field elements

[mstreng]

In /sage/rings/number_field/number_field.py (as per #11869), when converting number field elements, the parts

• List of candidates for K(x)
• Find a common field F into which KF and LF both embed

are independent. The first can be very slow, while the second is likely to fail. So I propose to swap these two parts.

See #12269 for an example and for indicator code.

CC: @jdemeyer @dansme @sagetrac-JCooley

Component: number fields

Keywords: sd51

Author: Jenny Cooley

Branch: u/davidloeffler/ticket/12270

Reviewer: David Loeffler

Merged: sage-5.12.beta0

Issue created by migration from https://trac.sagemath.org/ticket/12270

[jdemeyer]

comment:1

I'm a but confused. Exactly where and when are roots unneccessarily computed?

[mstreng]

comment:2

Replying to @jdemeyer:

I'm a but confused. Exactly where and when are roots unneccessarily computed?

In /sage/rings/number_field/number_field.py (as per #11869) the parts

• "# List of candidates for K(x)" and
• "# Find a common field F into which KF and LF both embed."

are independent. The first can be very slow, while the second is likely to fail. So I propose to swap these two parts.

[jdemeyer]

comment:3

Thanks for the clarification. I agree.

[sagetrac-JCooley]

Changed keywords from none to sd51

[sagetrac-JCooley]

swaps the order of the two processes

[sagetrac-JCooley]

Attachment: trac_12270_swap_processes.patch.gz

fixes an error that came up in testing

[sagetrac-JCooley]

comment:8

Attachment: trac_12270_doc_fix.patch.gz

Sorry, newbie error, I did hg_commit() before I had tested the patch, hence there being two patches.

[loefflerd]

Branch: u/davidloeffler/ticket/12270

[loefflerd]

Reviewer: David Loeffler

[loefflerd]

comment:11

Does what it says on the tin. Positive review.

[loefflerd]

Author: Jenny Cooley

[jdemeyer]

Merged: sage-5.12.beta0