@SanathDevalapurkar it's related by a sequence of also-nontrivial facts. L_n-localization acts by rationalization on eilenberg-mac lane spectra, so a (spectral) map from Z/p into gl_1 E_n is sent to zero by postcomposition with gl_1 E_n --> L_n gl_1 E_n, so EinftySpaces(Z/p, GL_1 E_n) = Spectra(HZ/p, gl_1 En) = Spectra(HZ/p, F). then pi_j Spectra(HZ/p, F) = pi_0 Spectra(S^j HZ/p, F), and since F is n-cotruncated (by AHR 4.11), that last thing is = pi_0 Spectra((S^j HZ/p)(-infty, n], F)
if j exceeds n, then the source becomes contractible and the mapping space collapses
lurie's conjecture in the MO answer is an assertion about the Einfty space associated to Spectra(HZ/p, F), which i guess may have lots more coconnective information as a spectrum? but as a space he's guessing that it's a K(Z/p, n)
i actually wonder if that's a theorem now, last time i was in boston i heard that there had been a successful identification of the 'discrepancy spectrum' (which i think is F) (or maybe its connective truncation) with (a shift of) the brown-comenetz dualizing spectrum (or maybe its connective truncation)
p.s. is there a straightforward stable calculation that L_n = L_0 on E-M spectra? this follows from the unstable thing of course, but that's a considerably stronger theorem. maybe there's an abbreviated way to see just the stable statement
does the following construction have a name? start with a connected space X. write down the natural map X -> B H_1(X, Z) inducing an isomorphism on H_1. take its homotopy fiber X_1. write down the natural map X_1 -> B H_2(X_1, Z) inducing an isomorphism on H_2. take its homotopy fiber. etc.
in other words, it's like the whitehead tower, but you're trying to kill the homology of X rather than its homotopy
in particular, does the sequence of homology groups H_{n+1}(X_n, Z) you get this way have a name? (as far as I can tell it is not something familiar, or at least not to me)
in particular in particular, is it known what happens when you apply this construction to X = BS_n, at least stably in n?
(i have a vague suspicion that the answer, stably in n, is that you get the stable homotopy groups of spheres, and i'm trying to figure out whether this follows from barratt-priddy-quillen-segal or is a different statement)
user105491
1:39 AM
@EricPeterson Can I ask what (S^j HZ/p)(-infty, n] means?
S^j is suspension, and X(-infty, n] is my favorite notation for the object under X which is n-coconnective and whose homotopy groups agree with those of X up to & including n
(or: the fiber of the map X --> X(-infty, n] is n-connected & could be reasonably called X(n, infty))
user105491
Ah, thanks. That was the only part that confused me.
user105491
2:12 AM
@Eric Is this the discrepancy spectrum you were talking about?
the email thread, restored: cl.ly/2d1J2C1G1I3j . unfortunately i was asking a question that was premised on something false, and charles's reply is mostly about explaining the correct premise. still worth reading
user105491
Thanks! I'll read over it the day after tomorrow; I have a debate tomorrow at school...
i'm still tickled that eta in pi_* S acting on "eta" in pi_* gl_1 S has eta * "eta" = 0. bob bruner sent me? tried to send me? a paper by waldhausen that roughly said that the (pi_* S)-module structure of pi_* gl_1 R is essentially the same as that of pi_* R, except for etas, which appear and disappear only as they need to to make homotopy theory self-consistent
i have not yet tried to understand that claim at all
@QiaochuYuan I don't know if that construction has a name, sorry. I think that in the case of S_n (stably) you can, for n >> 0, inductively show that the map BS_n -> (BS_n)^+ induces homology isomorphisms at all stages of this tower, and conclude that you're effectively calculating the Whitehead tower of (BS_n)^+
so yes, for large n you're recovering the (positive-degree) stable homotopy groups of spheres due to homological stability for symmetric groups + Barratt-Priddy-Quillen
@EricPeterson do you happen to remember which Waldhausen paper that was?
@EricPeterson I would be VERY surprised if enough was known about the power operations on E_n to employ Charles' approach, but maybe others have been more clever.
@TylerLawson from bob's email: §2 of "The map BSG -> A(X) -> QS^0", Bökstedt and Waldhausen. again, I haven't looked for myself
@PeterNelson tobi tried to tell me how the Bostonians pulled it off, but I forget what he said. Maybe he'll send you a Private Communication? In any case it wasn't by Charles's method
Also re: the discrepancy spectrum result, mike wanted to know (last summer) whether the cech spectral sequence associated to the chromatic fracture hypercube gave a genuinely new way of conputing stable stable stems or if we would recognize it from somewhere else. That's soecifically whar Tobi and I talked about, his suspicion (last summer) was that we'd recognize it from somewhere else
@Tyler: great, thanks! is this a special property of BS_n or is it generally true that this construction ends up recovering the whitehead tower of the plus construction?
@QiaochuYuan what was necessary, so far as I can tell, is that it's the commutator subgroup [S_n, S_n] which is perfect. I had to use this to conclude that the fiber of X^+ -> K(H_1,1) was an (X_1)^+
you know what, that's not correct. to take a homology isomorphism X -> Y and conclude that fib(X -> K(A,1)) -> fib(Y -> K(A,1)) is a homology isomorphism, you need to know more; the assumptions only allow you to conclude something about the maps H_*(A; H_*(fib_1)) -> H_*(A; H_*(fib_2))
all right. I think that by being more explicit with a cell description, this does work: if X has perfect commutator subgroup of pi_1(X), then you can arange a map X -> X^+ so that taking the fiber of the map out to K(H_1 X, 1) gives a map Y -> Y^+.