markusde

ALL HAIL LORD TRANS

I have acquired

Full body skeleton suit

I'm gonna do a fun project today

https://github.com/leanprover-community/iris-lean/pull/688

I've posted a few times about it but I think my work here is done. I am now very very pleased with this PR.

Fully swapped the theory underlying Iris with essentially zero changes for clients (or at least for the HeapLang client). Might be my favorite piece of code I've ever written, I feel like a proof ninja.

Extremely satisfying to work in the "putting things into their proper places" factory

Today's scripture verse comes from the proof irrelevance chapter

This change I'm helping shepherd into Iris-Lean is somehow radicalizing me even more.

We're working on changing the algebraic hierarchy to be based on ORA's instead of CMRA's, but because Lean has good typeclasses, we can actually do this swap Indiana Jones style with essentially zero impact on clients who use CMRA. It's kind of like how Iris-Rocq has a MRA construction, but you'll notice.... no MRA canonical structures. I understand that this kind of thing is very hard to do in that system (hard enough to necessitate a fork). Not a problem in Lean.

@ohad@mathstodon.xyz You know, I've been making two arguments that felt very different for some time now:

• Verification systems can be made more effective by using a worse but more conventional probability theory, because there are fewer open problems you need to solve before using the tool.
• The increasing expectations for papers to be formally verified will disadvantage interesting new ideas in favour of the status quo.

It is now occurring to me, through your post, that the former quite is a compelling example for the latter. @gallais@mamot.fr

As a true ultrafinitist I believe all ultrafinitist theories. Therefore there cannot be a maximum number N due to the fact that there is an (N+1) ultrafinitist theory which I believe in

Bird I have an idea

Providence is what I imagine ohio is like from all the rude jokes on Instagram

I don't think I truly appreciated how far ahead mathlib is compared to the other analysis libraries I have used. Half of the people at this workshop are just... regular mathematicians. And we're talking through theorems I understand maybe 5% of as a "realistic next step".

AND this whole workshop is being framed as "Analysis is less supported than algebra in mathlib let's change that". Your "less supported" is my "I hope to understand it before I die"

I'm picturing a "Keep Austin Weird" style campaign to "Make Providence Be Anything"

I caved and shelled out an additional $70 to not depart at 5:45 AM tomorrow (now I can leave my house at a luxurious 5:50 instead of a time starting with a 4)

In trying a thing where I read a textbook with sticky notes in hand, and I use it to fill in the missing little proofs. It's too early to tell for sure but I really like this! I've noticed it forces me to really understand the definitions eagerly and at a low level (instead of just getting the high level picture and being lost 20 pages later) which is especially useful in probability where the notation is horrible.

lol https://red-squares.cian.lol

Why do they call it main when it's a state full of side characters

Possibly this is tainted by my negative mood but I think Providence is the worst place on the planet and it should be sunk like a pathetic atlantis

I have the wonderful ability to be social xor sober

I am hatching a plan to get into the Rivoli as we speak

I have officially closed the Lean Zulip. Not opening it until Thursday unless the shakes get really bad

Brown has a lot of similarities to the Tufts Downhill Skatepark in terms of skatability

Feelin like making a change to View. Prepare for a shitstorm.