Add TMVA support to ThOr

Now Moore depends on MVADictTools which is in LoKi. Ideally MVA cuts should be handled in ThOr as suggested by @apearce in Phys!919 (comment 4539230). This would allow Phys!919 (closed) to proceed.

added DPA-WP3 label

mentioned in merge request Phys!919 (closed)

mentioned in merge request Moore!931 (merged)

added ThOr label

mentioned in merge request Moore!1033 (merged)

Is there any time estimate on this issue? It would be nice to be able to include lines using MVAs in throughput tests, see Moore!1033 (merged).

Not sure I understand the thumbs up :D. That was a question to DPA as @mvesteri tells me it is under the attention of DPA.

:) I created this issue, indeed. But we will implement MVA when someone needs it. IF RTA needs it, you should define the schedule as it's more critical for you. We can add functors to be used offline any time.

I don't mind whether it is formally DPA or RTA since we are all friends . I just noticed this under the DPA WP3 actvities in the hackathon codimd

ah, this is why this comes up.

So do I get an answer? ;)

For us it's "best effort basis". So no due date.

OK it sounds like it may be more on the critical path for RTA, so we should discuss in RTA coordination to see if we find any effort...

Thanks for the answers.

Hi all, are there any news on this? From DPA we want to finalise on Phys#19 (closed) this year and will be discussing the how-to at our coords meeting tomorrow.

@nnolte does any of your recent work on the Topo overlap with this issue?

no i didn't do anything in this direction. what about this MR? !2588 (merged)

Hi @aalfonso, your work got pointed at in this discussion. Can you comment? Advance thanks.

In !2588 (merged) I ported the code from TMVATransform so that TMVA can now be invoked from the MVA functor. The code is essentially the same just with a few less unnecessary checks and of course adapted to live under MVA. I was able to test that the inputs were being fed correctly into a BDT and that the output was ok too.

The MR was reverted to draft just in case some other test was necessary and maybe we could add some debug/verbose behaviour. Otherwise it is ready to merge

Morning @aalfonso. Thank you for your quick reply. This sounds great and it would be good to see this merged and in use asap, for obvious reasons. But that means "replacement work" for RTA to ideally get rid of the dependency on MVADictTools (we are likely to have some work also on the offline side). @mvesteri, @poluekt, can this be given consideration and a push? First things first, let's get !2588 (merged) in .

Yes noted @erodrigu

(Sorry for the "push" but better to raise the hand now for the benefit of us all than the day before a beam test or alike, right? But acknowledged that priorities exist ... and are functions of time.)

mentioned in merge request !2588 (merged)

Closed with !2588 (merged)

closed