« first day (767 days earlier)      last day (2050 days later) » 

1:59 AM
What is A^3? Is it ((A times A) times A so its elements are ((a,b),c) or can it be defined to be something else that is not (A times B) times C nor A times (B times C)?
 
 
2 hours later…
3:33 AM
@famesyasd you can define it as the set of functions from 3 to A
 
 
2 hours later…
5:54 AM
@famesyasd we usually don't really care about this question. Because $A\times (B\times C)\cong (A\times B)\times C$, and when $A=B=C$ then there is also isomorphism between the set of functions from $\{0,1,2\}$ to $A$ or the set of functions from $\{1,2,3\}$ to $A$
 
 
6 hours later…
11:57 AM
@user525966 Sorry I don't know what you mean by combine systems. You cannot combine different deductive systems, in the same sense that there is no well-defined notion of combining different games. However, I think you meant to say "combine proofs" or "combine axioms".
To answer that, recalled that I have defined what a theorem means:
20 hours ago, by user21820
@user525966 Short answer is "yes" Long answer: I might have forgotten to define what is a theorem in the Fitch-style system we have gone through. A theorem is a assertion/statement that is not under any context whatsoever. For example, in the first example proof here, there are two lines that are outside all the contexts, and those two are theorems.
Now in a deductive system you may also have axioms, which you are allowed to assert anywhere. I did not want to get to that yet, because we were still at propositional logic. Intuitively, "axiom" is supposed to mean something that you are assuming to be true, and hence it makes sense that we are allowed to assert an axiom anywhere. But if you want to talk about combining axioms it means you want to study logic from the outside, so we can consider any set of axioms, nonsensical or not.
We shall write "S |− Q" to mean "The formal system S proves (generates) the statement Q". We so far have a Fitch-style deductive system for PL (propositional logic), and so we can for example say that ( PL |− A or not A ) and ( PL |− A or ( A implies B ) ).
This definition can be applied to any other deductive system for propositional logic, because any one of them will prove the same statements (even though the proofs may look very different).
Now to add the notion of axioms, for a formal system S for propositional logic we shall write "S+X" to mean the same system plus the set X of axioms, meaning that you can use any statement in X as an axiom (besides the axioms S already has).
For example, it is false that ( PL |− A implies B ), but it is true that ( PL+{not A} |− A implies B ). If you want to be precise, it would be PL+{"not A"} |− "A implies B", but it is conventional to omit quotes if the reader can figure out what is a string and what is not (but when I omit quotes I usually enclose in brackets to make sure it is unambiguous).
Okay now we can talk about combining axioms. If you have two sets X,Y of axioms, then PL+X and PL+Y are compatible formal systems that both extend PL, in the sense that, if you have ( PL+X |− Q ) and ( PL+Y |− R ), then you will trivially get ( PL+X+Y |− Q ) and ( PL+X+Y |− R ).
I think this is what you want, because it says that, if from axioms X you can prove Q, and from axioms Y you can prove R, then from axioms X⋃Y you can prove both Q and R.
 
12:35 PM
Likewise, for first-order logic the Fitch-style deductive system can serve as the 'base' formal system FOL, and you will have things like ( FOL |− ∀x ( x=x ) ) and ( FOL |− ∃x ∀y ( y=x ) ⇔ ∃x ( x=x ) ∧ ∀y,z ( y=z ) ). And again we write "FOL+U" to denote the system FOL plus (first-order) axioms U, and so given (first-order) axiom sets U,V such that ( FOL+U |− Q ) and ( FOL+U |− R ) we again have ( FOL+U+V |− Q ) and ( FOL+U+V |− R ).
In many textbooks they will omit all mention of "PL" and "FOL", and rely on the reader figuring out which is meant based on the context.
@famesyasd In set theory, (A * A) * A is not the same as A * (A * A), and they are both different from the set of functions from {0,1,2} to A. So if you want to be precise you will have to use the correct one for your purposes.
@user525966: By the way, I suggest you finish with learning the Fitch-style system for first-order logic, because as you can now see it is not hard, and it will be extremely important later. Do you understand the 2 examples?
@famesyasd: Oh yea it is worth mentioning that the generalized product type cannot be expressed via cartesian product. Namely, if you have an arbitrary index set I and a function S from I to U, then you may want the set of functions { f : f in func(I,Union(U)) and forall k in I ( f(k) in S(k) ) }, which is also called a dependent product type. For example, if I = N then you get the set of infinite sequences where the k-th item is in S(k).
 
1:03 PM
@user21820 but they all have isomorphism from one to the other, so usually we don't really care which one to use, no?
 
@Holo In practical mathematics, it is of course common to just ignore the differences, but surely you will agree that they are ultimately not the same thing, and that no amount of cartesian products will get you the set of infinite sequences, so there is a fundamental difference between using cartesian product and using functions to 'encode' the product type.
It's partly because famesyasd seems to be interested in getting the technical details correct, and not just the abstract ideas grasped. Other than that, sure, they are just different implementations of the same intended interface.
 
@user21820 so yes, they are not technically the same, and the set of functions definition is more general
 
Yea. Though it's interesting to note that many practical formal systems still have cartesian product types even if they can form the general product types. Seems like the same reason it is easy for us to imagine finite tuples as literal finite strings with items dangling off them than to imagine them as functions from an initial segment of N.
 
 
1 hour later…
2:28 PM
@LeakyNun: You have just appeared in a HNQ:
62
Q: Which mathematical definitions should be formalised in Lean?

Kevin BuzzardThe question. Which mathematical objects would you like to see formally defined in the Lean Theorem Prover? Examples. In the current stable version of the Lean Theorem Prover, topological groups have been done, schemes have been done, Noetherian rings got done last month, Noetherian schemes h...

 
indeed
 
I don't like the implications of the comments:
Lean 4 is a different story, not least because it does not exist. We are patiently waiting for the Lean developers to release Lean 4, and when it is made public there are rumours that certain critical pieces of infrastructure (for example the "simp" tactic which proves certain equalities automatically, saving the user from having to supply trivial proofs) will be absent. So it will be a while before we are ready for Lean 4. Porting will occur, at first slowly, and then more quickly as we begin to understand what needs to be done. — Kevin Buzzard yesterday
One of the criticisms of Lean on the blog of Hales is incredible: "Lean is not backwards compatible. Lean 3 broke the Lean 2 libraries, and old libraries still haven’t been ported to Lean 3. After nearly 2 years, it doesn’t look like that will ever happen. Instead new libraries are being built at great cost. Lean 4 is guaranteed to break the Lean 3 libraries when it arrives." So a big chunk of your work may be wrecked when the next version of Lean comes out. — KConrad yesterday
@ScottMorrison I know nothing about theorem prover systems, but it seems like a stupendously bone-headed idea to create a system with this type of "feature" at every single upgrade. — KConrad yesterday
It would not be a problem if each new version also comes together with a trans-compiler.
 
well I can't do anything about that
 
Yea.
But I have the same gripe with document processors.
It's not specific to this. I just detest it when documents I have fail to even open in new versions of the document processor.
It's particular annoying that it also happens with LaTeX, which is peculiarly hard to debug.
 
3:09 PM
> It is not possible to get a usable installation of Lean + library under Windows without following instructions in Kevin's blog post − Neil Strickland
This is also ridiculous, more so than the non-backward-compatibility.
One thing which Lean, or at least its libraries, are not good at is playing with axioms. That is, showing what holds in the absence of the axiom of choice, for example. Although there is a mechanism for finding out if a theorem uses a particular axiom, and AC is one of these, it's no help if all the basic theorems in the library assume AC (in particular when Diaconescu is used to prove LEM from AC). So while it's possible to define your category, Hahn-Banach is true from lean's POV, and so it collapses to the usual category. ... — Mario Carneiro yesterday
This too. I wonder who was the one who thought it was a good idea to use AC to prove LEM?!
 
but that's Diaconescu's theorem
so, Diaconescu was the one
@user21820 I understand, but you see, we have quite a situation
I would not like to discuss this in public
 
@LeakyNun Diaconescu merely observed that you can use AC over IZF to prove LEM. I was referring to the issue of doing this in a practical formal system, which seems really a wrong decision.
 
hmm
 
@LeakyNun Why is it not possible to just wrap the whole thing in some auto-installer for windows 7/8/10? In general, I think if one wants to get something widely used, one should make it as easy as possible. Security considerations aside, that means auto-installation. Presumably these logicians aren't concerned about security anyway, since 1Gb is impossible for anyone to check for hidden malware.
 
 
4 hours later…
6:58 PM
@user21820 To some extent, kind of
I'm a little lost on the overall context of fitch-style in the first place, i know you answered before but I didn't quite understand
like is this another system we can place "alongside" hilbert style, ND style, sequent style, etc?
or is it a "proof writing style" we can apply to hilbert systems, ND systems, sequent systems, etc?
i can't tell if fitch-style is its own "logic system" or if it's more or a "writing style / notation" that accomplishes the same thing as just writing out lines like I did here: math.stackexchange.com/questions/2927979/…
 
 
5 hours later…
11:53 PM
@user21820 Do you know whether the class of functions computable by a quantum turing machine is strictly larger than the class of functions computable by a classical turing machine?
It certainly appears to effect efficiency.
For example, https://en.wikipedia.org/wiki/Grover%27s_algorithm
https://en.wikipedia.org/wiki/Shor%27s_algorithm
However faster does not necessarily mean stronger in this context.
 

« first day (767 days earlier)      last day (2050 days later) »