Recipe
Compute a coset of a subgroup
Use when: you're given a subgroup H, a representative g, and a side (left or right), and asked for the coset gH or Hg
Goal: the complete coset as a set of |H| elements, with none skipped and none miscomputed
Taught in Left and right cosets
Quick recipe
-
1
Write out all of H's elements explicitly; the coset will have exactly this many entries, no more and no fewer.
Example
H=\{e,(1\ 2)\} in S_3: two elements to multiply through.
Why: a coset gH is defined by multiplying g by every element of H, so missing an element of H directly means missing an element of the coset.
-
2
For a left coset gH, compute g\cdot h for every h\in H, applying h first then g. For a right coset Hg, compute h\cdot g instead, applying g first.
Example
Left coset (1\ 2\ 3)H: (1\ 2\ 3)\cdot e=(1\ 2\ 3), and (1\ 2\ 3)\cdot(1\ 2) traces to (1\ 3), applying (1\ 2) first, then (1\ 2\ 3).
Why: gH and Hg genuinely differ in a non-abelian group, so multiplying on the wrong side produces the wrong set entirely, not just a relabeling.
-
3
Gather every product from Step 2 into one set; a coset always has exactly |H| elements, since h\mapsto gh (or h\mapsto hg) is a bijection from H onto it.
Example
(1\ 2\ 3)H=\{(1\ 2\ 3),(1\ 3)\}, two elements, matching |H|=2.
- every element of H was actually multiplied through — the set has |H| elements, and the coset is complete.
- only the identity of H was multiplied through, stopping early — the "coset" found has just 1 element, a sign a step was skipped.
Why: stopping before every element of H is processed produces a subset of the true coset instead of the whole thing.
-
4
Repeat the process on the other side (Hg if you computed gH, or vice versa), and check whether the two sets are identical.
Example
H(1\ 2\ 3)=\{(1\ 2\ 3),(2\ 3)\}, which differs from (1\ 2\ 3)H=\{(1\ 2\ 3),(1\ 3)\} found above, since S_3 is non-abelian.
Why: the two sides always agree automatically in an abelian group, so genuinely comparing both is the only way to confirm which situation applies.
Common mistakes
- Stopping after multiplying g by just the identity of H, producing a "coset" of size 1.
- Multiplying on the wrong side, using gh when the problem asked for Hg, or vice versa.
- Assuming left and right cosets always agree, which only holds in an abelian group.
- Forgetting that g itself is one element of any coset containing it, never the entire coset alone.