The Student Room Group

A Summer of Maths (ASoM) 2016

Scroll to see replies

Original post by drandy76
ImageUploadedByStudent Room1468610947.215133.jpg
Was looking at this and noticed something odd when I looked at the middle two/one, with the outer sides you can 'follow' the trail from top to bottom with no problems, but if you try the same for either of the middle pillar thingies it can't be done, does anyone know how they managed this?


Posted from TSR Mobile


They joined gaps on one end and respectively opposite on the other side the extra comes from gap between 4 is 3 spaces etc.
Proper weird though.


Posted from TSR Mobile
Original post by physicsmaths
They joined gaps on one end and respectively opposite on the other side the extra comes from gap between 4 is 3 spaces etc.
Proper weird though.


Posted from TSR Mobile


Yeah it was giving me a headache trying to trace it lol, thanks geometry nerd


Posted from TSR Mobile
Original post by drandy76
Yeah it was giving me a headache trying to trace it lol, thanks geometry nerd


Posted from TSR Mobile


No geometry in this one lad. Just using those eyes.


Posted from TSR Mobile
Original post by physicsmaths
No geometry in this one lad. Just using those eyes.


Posted from TSR Mobile


I have 1 1 vision bro


Posted from TSR Mobile
If a,b∈Zp a,b \in \mathbb{Z}_{p} for p prime, why are there p-1 choices for a,b such that ab=±1 ab = \pm 1 (mod p) ?
Original post by EnglishMuon
If a,b∈Zp a,b \in \mathbb{Z}_{p} for p prime, why are there p-1 choices for a,b such that ab=±1 ab = \pm 1 (mod p) ?


for each of 1,...,p-1 there is a multiplicative inverse to pair off with (e.g. for p = 3 you have (1,1), (2,2), for p = 5 you have (1,1), (2,3), (3,2), (4,4) )
edit: lol didn't notice the +/- tho
I presume it means we care either only about pairs such that ab = -1 (modp) or only about pairs such that ab = 1 (modp), or else there seem to be counterexamples. In this case proving the former is also fairly simple; multiplication is commutative and we can just multiply one of the elements in the inverse pair by p-1 (or -1 if you prefer) to get the corresponding pair that multiply to -1
(edited 7 years ago)
Original post by 1 8 13 20 42
for each of 1,...,p-1 there is a multiplicative inverse to pair off with (e.g. for p = 3 you have (1,1), (2,2), for p = 5 you have (1,1), (2,3), (3,2), (4,4) )
edit: lol didn't notice the +/- tho


Yep, thought so for the +1 case. Was the -1 case I wasnt sure how to explain though. thanks :biggrin:
Original post by EnglishMuon
Yep, thought so for the +1 case. Was the -1 case I wasnt sure how to explain though. thanks :biggrin:


dunno if you've seen the edit. But it's probably not worded amazingly; basically the point is that given each pair (a,b) for the +1 case, you can multiply b by -1 to find a pair for the -1 case (as you get a(-b) = -(ab) = -1), i.e. you can sort of set up a correspondence
Original post by 1 8 13 20 42
dunno if you've seen the edit. But it's probably not worded amazingly; basically the point is that given each pair (a,b) for the +1 case, you can multiply b by -1 to find a pair for the -1 case (as you get a(-b) = -(ab) = -1), i.e. you can sort of set up a correspondence


thanks yea that makes sense. My mistake was that after multiplying one of a or b but p-1 stupidly I forgot that that is congruent to one of the other elements in Zp.
Any hints on how to prove this? :

If G G is a group and H,K H, K are subgroups of G of finite index in G, prove that H∩K H \cap K is of finite index in G G .

Some initial thoughts:

Spoiler

Reply 390
Original post by EnglishMuon
Any hints on how to prove this? :

If G G is a group and H,K H, K are subgroups of G of finite index in G, prove that H∩K H \cap K is of finite index in G G .

Some initial thoughts:

Spoiler



G being finite.
Since you have proven that the intersection of H and K is a group, you are done! From the proof of Lagrange, this subgroup has a particular (finite) index.


G being infinite. A good example of this that I can think of is G = Z with the + operation and subgroups H,K being the {3n | n in Z} and {4n | n in Z}. These have finite indexes in G. (Eg, the left cosets of H are {3n | n in Z}, {3n+1 | n in Z}, and {3n+2 | n in Z}.) The intersection of H and K is {12n | n in Z} which has index 12 in G - clearly finite.

The only problem is that it would seem difficult to generalise this example to other infinite groups. Here I would turn to contradiction.

Suppose that H and K have finite indexes in G but their intersection has an infinite index. The intersection of H and K is a subgroup of H. At this point, I'm going to point out that K is irrelevant - K is only here to give us a subset of H (intersection of H,K), which is actually a group (so that we can talk about the index of this subset). If a subset (subgroup) of H has infinite index... How can H possibly have a finite index?


Posted from TSR Mobile
Original post by Ecasx
G being finite.
Since you have proven that the intersection of H and K is a group, you are done! From the proof of Lagrange, this subgroup has a particular (finite) index.


G being infinite. A good example of this that I can think of is G = Z with the + operation and subgroups H,K being the {3n | n in Z} and {4n | n in Z}. These have finite indexes in G. (Eg, the left cosets of H are {3n | n in Z}, {3n+1 | n in Z}, and {3n+2 | n in Z}.) The intersection of H and K is {12n | n in Z} which has index 12 in G - clearly finite.

The only problem is that it would seem difficult to generalise this example to other infinite groups. Here I would turn to contradiction.

Suppose that H and K have finite indexes in G but their intersection has an infinite index. The intersection of H and K is a subgroup of H. At this point, I'm going to point out that K is irrelevant - K is only here to give us a subset of H (intersection of H,K), which is actually a group (so that we can talk about the index of this subset). If a subset (subgroup) of H has infinite index... How can H possibly have a finite index?


Posted from TSR Mobile

Thanks, and Indeed the integers was the example I was drawn to to help get a feel for whats happening!

I have the following argument so far:

Let x1,...,xm x_{1},...,x_{m} be the elements that account for all distinct Hx Hx
and
y1,...,yn y_{1},...,y_{n} be the elements that account for all distinct Ky Ky .
As any two subgroups are either equal or disjoint,
G=G∩G=(∪i=1mHxi)∩(∪i=1nKyi)[br]=∪1≤i≤m,1≤j≤n(Hxi∩Kyj)[br] G= G \cap G= ( \cup ^{m} _{i=1} Hx_{i} ) \cap ( \cup ^{n} _{i=1} Ky_{i}) [br]= \displaystyle\cup _{ 1 \leq i \leq m ,1 \leq j \leq n} (Hx_{i} \cap Ky_{j})[br]

But He=H He=H is a rightcoset so we can choose our xi, yj such that wlog x1=y1=e x_{1} = y_{1}=e . Still cant see the final argument from this though...

p.s apologies for this disgusting formatting, if someone knows how to 'displaysyle' the intersections and have 2 lines of subscript one under the other, id appreciate it.
(edited 7 years ago)
Ah I might have it:

Basically I think you can show Hxi=Kyj=Hg,g∈G Hx_{i}= Ky_{j} = Hg , g \in G if Hxi∩Kyj Hx_{i} \cap Ky_{j} isnt the empty set. So Hxi∩Kyj=Hg∩Hg=(H∩K)g Hx_{i} \cap Ky_{j} = Hg \cap Hg = (H \cap K) g
hence
G=∪i=1nm(H∩K)gi G= \cup ^{nm} _{i=1} (H \cap K) g_{i} for gi∈G g_{i} \in G i.e. there are less than or equal to mn right cosets of H n K in G:

iG(H∩K)≤iG(H)iG(K) i_{G} (H \cap K) \leq i_{G} (H) i_{G} (K) (as (H∩K)gi (H \cap K) g_{i} may equal some (H∩K)gj,gi≠gj (H \cap K) g_{j}, g_{i} \not= g_{j} )
Reply 393
Original post by EnglishMuon

p.s apologies for this disgusting formatting, if someone knows how to 'displaysyle' the intersections and have 2 lines of subscript one under the other, id appreciate it.


\bigcup: ⋃1≤i≤m,1≤j≤n\displaystyle \bigcup_{1 \leq i \leq m, 1 \leq j \leq n}
Reply 394
Original post by EnglishMuon
Ah I might have it:

Basically I think you can show Hxi=Kyj=Hg,g∈G Hx_{i}= Ky_{j} = Hg , g \in G if Hxi∩Kyj Hx_{i} \cap Ky_{j} isnt the empty set. So Hxi∩Kyj=Hg∩Hg=(H∩K)g Hx_{i} \cap Ky_{j} = Hg \cap Hg = (H \cap K) g
hence
G=∪i=1nm(H∩K)gi G= \cup ^{nm} _{i=1} (H \cap K) g_{i} for gi∈G g_{i} \in G i.e. there are less than or equal to mn right cosets of H n K in G:

iG(H∩K)≤iG(H)iG(K) i_{G} (H \cap K) \leq i_{G} (H) i_{G} (K) (as (H∩K)gi (H \cap K) g_{i} may equal some (H∩K)gj,gi≠gj (H \cap K) g_{j}, g_{i} \not= g_{j} )


I don't quite see what you are trying to do. I don't think your first sentence is true: 'Basically I think you can show...'. If a coset of H, say Hx, and a coset of K, say Ky, are to be equal, then H and K must be of equal size, which is not guaranteed in the question.

I would suggest looking for a contradiction. We have H and K being finite-indexed subgroups. We have T = (H intersection K) being a subgroup of H. If T has infinite index (infinitely many different cosets), then surely H must have infinite index too - why?

The benefit of contradiction here is that there is no detailed considerations needing to be made, in contrast to m, n being the number of cosets of H, K in your attempt, for example.


Posted from TSR Mobile


Ok. whats wrong with the first sentence? We know that any two right cosets in a group are either equal or disjoint. So if we are looking at the intersection of 1 right coset Hxi Hx_{i} and another Kyi Ky_{i} and this set is not empty, they must both equal some Hg Hg for some g∈G g \in G . There was a typo actually in the 2nd line as I meant to say Hxi∩Kyj=Hg∩Kg=(H∩K)g Hx_{i} \cap Ky_{j} = Hg \cap Kg = (H \cap K) g but I don't see why the rest of argument doesn't work.
Original post by Zacken
\bigcup: ⋃1≤i≤m,1≤j≤n\displaystyle \bigcup_{1 \leq i \leq m, 1 \leq j \leq n}


Thanks :smile:
Reply 397
Original post by EnglishMuon
Ok. whats wrong with the first sentence? We know that any two right cosets in a group are either equal or disjoint. So if we are looking at the intersection of 1 right coset Hxi Hx_{i} and another Kyi Ky_{i} and this set is not empty, they must both equal some Hg Hg for some g∈G g \in G . There was a typo actually in the 2nd line as I meant to say Hxi∩Kyj=Hg∩Kg=(H∩K)g Hx_{i} \cap Ky_{j} = Hg \cap Kg = (H \cap K) g but I don't see why the rest of argument doesn't work.


With the (Z, +) example, subgroups {3n} and {4n}, the intersection of the cosets {3n+1} and {4n+1} are certainly non empty, but this does not imply that Hx = Ky as you've written.


Posted from TSR Mobile
Original post by Ecasx
With the (Z, +) example, subgroups {3n} and {4n}, the intersection of the cosets {3n+1} and {4n+1} are certainly non empty, but this does not imply that Hx = Ky as you've written.


Posted from TSR Mobile


lol yea that seems true (e.g. 13 occurs in both 3n+1, 4n+1) but I swear
"any two left cosets of H in G are either identical or disjoint. In other words every element of G belongs to one and only one left coset and so the left cosets form a partition of G.[3] Corresponding statements are true for right cosets." must hold?? what on earth am i doing wrong here XD
Actually yeah its cus they are different subgroups
(edited 7 years ago)
Reply 399
Original post by EnglishMuon
lol yea that seems true (e.g. 13 occurs in both 3n+1, 4n+1) but I swear
"any two left cosets of H in G are either identical or disjoint. In other words every element of G belongs to one and only one left coset and so the left cosets form a partition of G.[3] Corresponding statements are true for right cosets." must hold?? what on earth am i doing wrong here XD
Actually yeah its cus they are different subgroups


Yeah they have to be cosets of the same subgroup.


Posted from TSR Mobile

Quick Reply

Latest