Skip to main content
Notice removed Authoritative reference needed by CommunityBot
Bounty Ended with no winning answer by CommunityBot
Notice added Authoritative reference needed by Kuzja
Bounty Started worth 50 reputation by Kuzja
added 6 characters in body
Source Link
Sam Hopkins
  • 24.2k
  • 5
  • 97
  • 171

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ randomly permuted perfect matchings, i.e., we take uniformly at random $k$ permutationpermutations from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set [ E_k := \bigcup_{i=1}^k \pi_i(M)$E_k := \bigcup_{i=1}^k \pi_i(M)$. ]

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For k=1$k=1$, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}. $$$$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}, $$ Wewe can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the Nudelnoodle problem getinggetting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

SupposeMore precisely, we havework with a 2-dimensional complex with a full 1-dimensional skeleton, and suppose. Suppose we use a SreinerSteiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformly at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the follwoingfollowing complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is threethere a way to estimate the number of ``spheres''"spheres" in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ randomly permuted perfect matchings, i.e., we take uniformly at random $k$ permutation from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set [ E_k := \bigcup_{i=1}^k \pi_i(M). ]

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For k=1, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}. $$ We can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the Nudel problem geting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

Suppose we have a 2-dimensional complex with a full 1-dimensional skeleton, and suppose we use a Sreiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformly at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the follwoing complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is three a way to estimate the number of ``spheres'' in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ randomly permuted perfect matchings, i.e., we take uniformly at random $k$ permutations from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set $E_k := \bigcup_{i=1}^k \pi_i(M)$.

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For $k=1$, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}, $$ we can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the noodle problem getting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

More precisely, we work with a 2-dimensional complex with a full 1-dimensional skeleton. Suppose we use a Steiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformly at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the following complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is there a way to estimate the number of "spheres" in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?

typos
Source Link
Gerry Myerson
  • 39.9k
  • 10
  • 186
  • 247

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ rendomplyrandomly permuted perfect matchings, i.e., we take univormelyuniformly at random $k$ permutation from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set [ E_k := \bigcup_{i=1}^k \pi_i(M). ]

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For k=1, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}. $$ We can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the Nudel problem geting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

Suppose we have a 2-dimensional complex with a full 1-dimensional skeleton, and suppose we use a Sreiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformelyuniformly at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the follwoing complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is three a way to estimate the number of ``spheres'' in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ rendomply permuted perfect matchings, i.e., we take univormely at random $k$ permutation from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set [ E_k := \bigcup_{i=1}^k \pi_i(M). ]

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For k=1, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}. $$ We can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the Nudel problem geting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

Suppose we have a 2-dimensional complex with a full 1-dimensional skeleton, and suppose we use a Sreiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformely at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the follwoing complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is three a way to estimate the number of ``spheres'' in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ randomly permuted perfect matchings, i.e., we take uniformly at random $k$ permutation from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set [ E_k := \bigcup_{i=1}^k \pi_i(M). ]

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For k=1, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}. $$ We can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the Nudel problem geting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

Suppose we have a 2-dimensional complex with a full 1-dimensional skeleton, and suppose we use a Sreiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformly at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the follwoing complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is three a way to estimate the number of ``spheres'' in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?

Source Link
Kuzja
  • 81
  • 1
  • 4

Number of spheres in complexes constructed from a Steiner Triple System

Let $M$ be a perfect matching on an even number of vertices $n$, and let $\mathbb{S}_n$ be the symmetric group on $n$ vertices. Next, we construct a graph $G_k := ([n], E)$ using $k$ rendomply permuted perfect matchings, i.e., we take univormely at random $k$ permutation from $\mathbb{S}_n$, $\{\pi_i\}_{i=1}^k$, and apply them to the perfect matching $M$ getting the edge set [ E_k := \bigcup_{i=1}^k \pi_i(M). ]

Now, we would like to compute the expectation on the number of cycles in $G_k$, or its first homological group over $\mathbb{Z_2}$. For k=1, it is obviously zero. For $k \geq 3$, we know that the graph is connected with high probability, and since we know that the expectation on the number of edges is $$ \mathbb{E}[|E_k|] = 1 - \left(1 - \frac{1}{n-1}\right)^k \sim \frac{k}{n-1}. $$ We can use Euler characteristic and compute asymptotically that $\text{rank}~H_1(G_k;\mathbb{Z_2}) = O(n)$.

The border case of $k=2$ can be reduced to the Nudel problem geting us $\text{rank}~ H_1(G_2;\mathbb{Z_2}) = O(\log n)$.

My question is, can we compute the same number for dimension 2?

Suppose we have a 2-dimensional complex with a full 1-dimensional skeleton, and suppose we use a Sreiner Triple System (STS) $S \subseteq \binom{[n]}{3}$ on $n$ elements instead of a perfect matching, where $n$ is of the form $6m+1$ or $6m + 3$ for some $m$. Note that in an STS each pair of elements is contained in exactly one triplet.

Chose uniformely at random 3 permutations $\pi_1, \pi_2, \pi_3 \in \mathbb{S}_n$, and construct the follwoing complex $$ K_3 = \binom{[n]}{1}\cup\binom{[n]}{2}\cup\bigcup_{i=1}^{3}\pi_{i}(S). $$

Question: is three a way to estimate the number of ``spheres'' in $K_3$, or more formally $\mathbb{E}[\text{rank}~H_2(K_3;\mathbb{Z_2})]$?