Assignments: June 9th, 2026
We first recall the form of Fisher's inequality used below.
Remark 1.
Maybe there is something wrong. It seems that what we need is Fisher's inequality ?
Theorem 1 (Fisher's inequality).
Let
|Ai\cap Aj|=\lambda\quad\text{for all }i\ne j,
\qquad
|A_i|>\lambda\quad\text{for all }i.
\mathcal{L}_x={\ell\in\mathcal{L}:x\in\ell}.
|\mathcal{L}x\cap\mathcal{L}y|=1.
n=|P|\leq|\mathcal{L}|.
|A|\notin L\pmod p\quad\text{for every }A\in\mathcal{F},
|A\cap B|\in L\pmod p
\quad\text{for every two distinct }A,B\in\mathcal{F}.
|\mathcal{F}|\leq\sum_{k=0}^{|L|}\binom{n}{k}.
fA(x1,\ldots,xn) =\prod{\ell\in L}\left(\sum{i\in A}xi-\ell\right).
fA(\mathbf{1}B) =\prod_{\ell\in L}(|A\cap B|-\ell).
fA(\mathbf{1}A)\ne 0.
\sum{A\in\mathcal{F}}cAf_A=0
\left{\prod{i\in S}xi:S\subseteq[n],\ |S|\leq s\right}
\sum_{k=0}^{s}\binom{n}{k}.
Download the original write-up here.