# Sylow $p$ subgroups of $S_{2p}$ and $S_{p^2}$ - Dummit foote - $4.5.45; 4.5.46$

Question is to find

• Sylow $p$ subgroups of $S_{2p}$ for odd prime $p$ and show that this is an abelian group of order $p^2$

• Sylow $p$ subgroups of $S_{p^2}$ for odd prime $p$ and show that this is a non abelian group of order $p^{p+1}$

what I have done so far is :

$|S_{2p}|=(2p)!=(2p)(2p-1)\cdots (2p-p)\cdots 4\cdot3\cdot2\cdot1$

with out giving proper justification I would say that $p$ occurs only twice in factorization of $(2p)!$..

$p$ occurs first time in $2p$ and second time in $(2p-p)=p$..

(I do not know how to make above justification look better)

So, sylow subgroup of $S_{2p}$ is of order $p^2$..

I was first thought of seeing for some example May be something like $S_6=S_{2.3}$

Order of sylow $3$ subgroup of $S_6$ is $9$

only possibilities for orders of elements are $3$..

If i take some two arbitrary elements of order $3$ there is a possibility that they do not end up nicely i mean some thing like

$$(123)(124)=(13)(42)$$ would happen..

I would take two elements of order $3$ and then end up with an element of order $2$ which would be a problem as a group of order $9$ can not have an element of order $2$

So, I would make sure that those elements do not create any mess and commutes nicely and the choice that comes to my mind is $\{(123),(456)\}$ and the group generated by them is of order $9$

$$\{\langle(123),(456)\rangle\}=\{(123)^i(456)^j : 0\leq i,j\leq 2\}$$

I would like to repeat the same in terms of $S_{2p}$ I would consider first half of $\{1,2,3,\cdots,2p-2,2p-1,2p\}$ as one cycle and second half as another cycle. I mean $\{(123\cdots,p-1,p);(p+1,p+2,\cdots,2p)\}$

These two cycles commutes nicely and do not create any mess giving rise to a subgroup of order $p^2$ namely :

$$\{\langle(123\cdots,p-1,p);(p+1,p+2,\cdots,2p)\rangle\}$$

$$=\{(123\cdots,p-1,p)^i(p+1,p+2,\cdots,2p)^j : 0\leq i,j\leq p-1\}$$

So, I guess this is good enough..

$|S_{p^2}|=(p^2)!=(p^2)(p^2-1)\cdots (p^2-p)\cdots (p^2-2p)\cdots (p^2-(p-1)p) \cdots 4\cdot3\cdot2\cdot 1$

i.e., $|S_{p^2}|=p^2\cdot p(p-1)\cdot p(p-2)\cdots p(1)m$ where $p\nmid m$

i.e., we have $p^2\cdot p^{p-1}=p{p+1}$ number of $p$ in factorization of $(p^2)!$

So, now the order is known and we need to find the subgroup...

I would repeat the same idea as of $S_{2p}$

For the same reason as that of $S_{2p}$ I would take for $S_{9}=S_{3^2}$the set $\{(123),(456),(789)\}$.

But then $\{\langle (123),(456),(789)\rangle\}=\{(123)^i(456)^j(789)^k : 0\leq i,j,k\leq 2\}$ is giving me a subgroup of order $27$ where as i need a group of order $81$ (which is the order of sylow $3$ subgroup of $S_{9}$ i.e.$3^{3+1}$)

so, It sounds some thing like I should take an element of order $3$ which behaves nicely and does not create much problem and give a subgroup of order $81$

I am sure that any $3$ sylow subgroup of $S_9$ would have a copy isomorphic to $\{\langle (123),(456),(789)\rangle\}$ so only problem is i am not sure how to extend this to a subgroup of order $81$

More generally for $S_{p^2}$ I would consider

$$\{\langle(123\cdots p)(p+1\cdots 2p)\cdots (p^2-p+1\cdots p^2)\rangle\}$$

This would give a subgroup of order $p^p$ but then i need a subgroup of order $p^{p+1}$.

I would be thankful if some one can help me to see what would be the obvious element that one should multiply this with :O

Thank you.

• What part do you want explained? – Jack Schmidt Feb 17 '14 at 14:17

To find Sylow $p$-subgroups in a group in which you can generally find $p$-elements and normalizers the following algorithm is sometimes used:

Assume $P_i$ is a $p$-subgroup of $G$ (we can take $P_0 = \{1_G\}$ to be the trivial subgroup). If $P_i$ is a Sylow $p$-subgroup, then yay, return $P=P_i$. Otherwise we know by theory* that $P_i$ is not a Sylow $p$-subgroup of $N_G(P_i)$ either. Let $x_i$ be any $p$-element of $N_G(P_i)$ not contained in $P_i$, and set $P_{i+1} = \langle x_i, P_i \rangle$. This is still a $p$-subgroup, since* $x_i$ normalizes $P_i$. Hence we can repeat.

Theory*: If $P_i$ is a not a Sylow $p$-subgroup, then it is properly contained in an as-yet-unknown Sylow $p$-subgroup $P$. Since $P$ is nilpotent, it satisfies the normalizer condition and so $P_1$ is a proper subgroup of the $p$-subgroup $N_P(P_1) \leq N_G(P_1)$. Hence $P_1$ is not a Sylow $p$-subgroup of $N_G(P)$.

Since*: If $H,K \leq G$ and $H$ normalizes $K$, then $HK=KH$ is a subgroup and $|HK| \cdot |H \cap K| = |H| |K|$. In our case these numbers are all powers of $p$.

Application to the symmetric group: You already have $$P_p = \left\langle (1,2,3,\dots,p), (p+1,p+2,p+3,\dots,2p),\dots,(p^2-p+1,p^2-p+2,\dots,p^2-1,p^2) \right\rangle$$ so you just need to find some element that normalizes this group. Hence you need some order $p$ automorphism of the group, say one that cycles the basis elements $$x_i=(p(i-1)+1,p(i-2),\dots,pi)$$ amongst themselves, cyclically, in a cycle of length $p$. Such an element is exactly $$x_p = (1,p+1,2p+1,\dots,p^2-p+1)(2,p+2,2p+2,\dots,p^2-p+2)\cdots(p,2p,3p,\dots,p^2)$$

Generally speaking normalizers can be hard to find, so this algorithm is only used when the more advanced algorithms are not available (which unfortunately is still pretty often AFAIK). However, it just works fine here.

• +1 for telling me why should i choose normalizing element.. :) that is what my actual question was and i could not ask it properly... If i have normalizing element $x$ of order $p$ then i would consider $M=\langle x \rangle$ and as $M\unlhd P$ I see that $MP$ is a subgroup of $G$ and then I have $|MP|=\frac{|M||P|}{|M\cap P|}$ but then $|M\cap P|$ is possibly $1$ (I am sorry i have not checked it yet, I am so excited :)) we would end up with a subgroup $MP$ such that $|MP|=\frac{|M||P|}{1}=|M||P|=p^p.p=p^{p+1}$ and this is what we need... Please tell me if i have misunderstood anything... – user87543 Feb 18 '14 at 5:09
• That is mostly correct ($M \unlhd P$ typically means that $P$ normalizes and contains $M$; we want $M$ to normalize $P$ which can be written $M \leq N_G(P)$). As long as $x_i \notin P_i$ we get that $P_i < P_{i+1}$, and so it must go up. In this case, we definitely get $P_p \cap \langle x_p \rangle = 1$ since $x_p \notin P_p$ and $|x_p| = p$, so the end of your argument is exactly correct. – Jack Schmidt Feb 18 '14 at 14:46
• Yes Yes... It makes perfect sense now.... Thank you so much :) – user87543 Feb 18 '14 at 15:58

The additional generator you could take to generate a Sylow $p$-subgroup of $S_{p^2}$ is

$(1,p+1,2p+1,\ldots,(p-1)p+1)(2,p+2,2p+2,\ldots,(p-1)p+2) \cdots (p,2p,3p,\ldots p^2)$

For example, for $p=3$, this would be $(1,4,7)(2,5,8)(3,6,9)$.

This normalizes the elementary abelian subgroup of order $p^p$ that you have already found: check that it conjugates $(1,2,\ldots,p)$ to $(p+1,\ldots,2p)$ to $(2p+1,\ldots,3p)$, etc.

The resulting nonabelian group of order $p^{p+1}$ is called the wreath product of $C_p$ and $C_p$, written $C_P \wr C_p$.

You get a Sylow $p$-subgroup of $S_{p^n}$ by taking and iterated wreath product $C_p \wr C_p \wr \cdots \wr C_p$ $n$ times.

• It does help me for $p=3$ but is it really obvious to guess that $(1,4,7)(2,5,8)(3,6,9)$ would work so nicely.. :O I would be thankful if you would like to say something more about choosing this... it is not arbitrary i guess... and i am not familiar with wreath products :O I would try to read it... Thank you :) – user87543 Feb 15 '14 at 9:11
• If you know that your are looking for a permutation that conjugates $(1,2,3)$ to $(4,5,6)$, $(4,5,6)$ to $(7,8,9)$ and $(7,8,9)$ to $(1,2,3)$, then it is easy to guess it! – Derek Holt Feb 15 '14 at 11:41
• but how does it help i mean how does knowing conjugating element helps any how :O... – user87543 Feb 15 '14 at 12:41
• Because this conjugating element normalizes the abelian subgroup of order $p^p$, and it has order $p$ itself,so together with the abelian subgroup it generates a group of order $p^{p+1}$. I learnt all of this a long time ago from Marshall Hall's book on group theory. First he defines and studies wreath products, and then he finds Sylow $p$-subgroup of $S_n$ (for all $p$ and $n$) as an application. It seems strange to me that you are doing this as an exercise, particularly if you are not already familiar with wreath products. – Derek Holt Feb 15 '14 at 13:40
• @PraphullaKoushik Wreath products are briefly introduced in Dummit and Foote in Exercise 23 of section 5.5, p. 187. – André 3000 Feb 15 '14 at 14:55