New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Merged by Bors] - feat(measure_theory/integral/lebesgue): Add Markov inequalities for tsum and card using counting measure. #17588
Conversation
…sum and card using counting measure.
Co-authored-by: Floris van Doorn <fpvdoorn@gmail.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I am not happy that these proofs about tsum
are done using measure theory, and it's definitely not good that they are in the measure theory library (instead of in a file topology/algebra/infinite_sum/something
).
Since similar arguments are already in mathlib without using measure theory, can you try to mimic/use those arguments?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't mind having some inequalities on infinite sums depending on measure theory, as it's often the most efficient way to prove them. But the file containing these inequalities should definitely not be in the measure_theory folder, if the statements of the inequalities don't involve measures. Could you create a new file (next to already existing files in the topology folder on infinite sums) for the results whose statements do not mention measures?
(And in this way, if someone wants to refactor later the proof to give direct elementary proofs, this shouldn't break anything).
Unsurprisingly, I wholeheartedly agree with this! (And besides many inequalities, something like a dominated convergence for summation would be an easy consequence of measure theory but not nice to develop from scratch).
For the last statements of this PR of Markov's inequality which don't have counting measure in the hypotheses or the conclusion, I can actually refactor right away and make these not use measure theory --- proposal below (golfs welcome!). I think the proofs are slightly worse, but the imports are lighter. Unless someone disagrees, I could switch to these and put them in the appropriate files for now. (I mainly care about unblocking the remaining portmanteau implications, so I would settle for the least controversial approach and argue about measure theory or not later.)
|
Co-authored-by: sgouezel <sebastien.gouezel@univ-rennes1.fr>
…n their statements.
bors r+ |
…sum and card using counting measure. (#17588) Add Markov inequalities for `tsum` and `finset.card` using the counting measure to translate the Markov inequality from measure theory to these. Co-authored-by: kkytola <39528102+kkytola@users.noreply.github.com>
Pull request successfully merged into master. Build succeeded: |
Add Markov inequalities for
tsum
andfinset.card
using the counting measure to translate the Markov inequality from measure theory to these.I don't know if it is considered bad to do cardinality proofs via importing measure theory. But I tend to believe the measure theory library is very convenient to use compared to cardinalities directly, so I'd prefer to allow this. Markov's inequality is an example of where I think there is a small advantage, but other tools including convergence theorems will be convenient to translate from
lintegral
s totsum
s.Perhaps the reviewers can comment on whether this import-heavy approach is acceptable/desirable at all.