bio  website  math.harvard.edu/~elkies 

location  Harvard University, Cambridge, MA 02138  
age  49  
visits  member for  4 years, 4 months 
seen  6 hours ago  
stats  profile views  24,841 
[Not sure yet what to use this space for...]
16h

comment 
2dimensional sublattices with all vectors having very big square (in absolute value)
But then it feels like the $N \cdot H$ construction or something much like it does work even when $\Lambda$ is not unimodular. 
1d

comment 
2dimensional sublattices with all vectors having very big square (in absolute value)
In any case this approach must fail in rank $n=3$ because then there are only finitely many $N$ for which a lattice of the form $N\cdot H$ can embed primitively into $\Lambda$. 
1d

comment 
Zero divisors with support of size 3 in group algebras of finite groups
Yes: let $G = \langle g \mid g^7 = 1 \rangle$ and $\alpha = g + g^2 + g^4$; then $\alpha (\alpha+1) = 0$. 
2d

answered  What is $\sum (x+\mathbb{Z})^{2}$? 
2d

awarded  Enlightened 
2d

awarded  Nice Answer 
2d

awarded  Necromancer 
2d

accepted  The coupon collector's earworm 
2d

comment 
The coupon collector's earworm
@kantelope try the new faster code. 
2d

comment 
The coupon collector's earworm
@Steven Stadnicki  I hadn't checked The Art Of Computer Programming; that's a good idea, but I cannot locate the problem in either the main text or the exercises. Were you able to find it there? 
2d

revised 
The coupon collector's earworm
edited mostly to report on the answer by **Kevin Costello** (and to improve the **gp** code at the end) 
2d

revised 
Does there exist a nonconstant, periodic, real analytic function with period 1 and rational Maclaurin coefficients?
minor edits 
Aug
31 
awarded  Necromancer 
Aug
31 
awarded  Revival 
Aug
31 
answered  Does there exist a nonconstant, periodic, real analytic function with period 1 and rational Maclaurin coefficients? 
Aug
30 
comment 
Congruences involving binary forms and primes of the form $x^2+y^2$
It might still help to see the original congruence, which could be more tractable than the form involving $\sum_{k=0}^s \left({s+k\atop k}\right) 2^k$. 
Aug
29 
comment 
The coupon collector's earworm
Thanks for this as well. I expect that I'll "accept" your answer soon (I already "upvoted" it), and will edit the question to reflect your answer. 
Aug
29 
comment 
The coupon collector's earworm
Thanks for this analysis; the appearance of $e$ as the asymptotic coefficient of $\log N$ is unexpected, and the results you quote almost give the asymptotic behavior of $M_N$ en route to describing the distribution near $e \log N$. I write "almost" because the distribution could be asymptotically concentrated at $e \log N$ but still have a heavy enough tail to shift the mean $M_N$ above $(e+o(1)) \log N$. Usually such "concentration" results come with error estimates that force the mean and mode to be quite close; is this the case here too? 
Aug
28 
awarded  Good Question 
Aug
27 
comment 
What is the exact statement of “there are 27 lines on a cubic”?
Even 27 rational lines (still with the same construction). 