Quadruples $ ( X; A, B, x _ {0} ) $,
where $ X $
is a topological space and $ A $
and $ B $
are subspaces of it such that $ A \cup B = X $
and $ x _ {0} \in A \cap B $.
The homotopy groups of triads, $ \pi _ {n} ( X; A, B, x _ {0} ) $,
$ n \geq 3 $(
for $ n = 2 $,
it is just a set), have been introduced and are used in the proof of homotopy excision theorems. There is also an exact Mayer–Vietoris sequence connecting the homology groups of the spaces $ X $,
$ A $,
$ B $,
$ A \cap B $(
cf. Homology group).
For a triple $ ( X ; A, B) $ consisting of a space $ X $ and two subspaces $ A , B \subset X $, one defines the path space $ \Omega ( X; A, B) $ as the space of all paths in $ X $ starting in $ A $ and ending in $ B $,
$$ \Omega ( X; A, B) = \{ {p : [ 0, 1] \rightarrow X } : { p( 0) \in A , p( 1) \in B } \} . $$
If there is a distinguished point $ * $ in $ A \cap B $, the constant path at $ * $ is taken as a distinguished point of $ \Omega ( X; A, B) $( and is also denoted by $ * $).
The relative homotopy groups (cf. Homotopy group) $ \pi _ {n} ( X, A, * ) $, $ * \in A \subset X $, can also be defined as $ \pi _ {n-} 1 ( \Omega ( X; A, * ) , * ) $. Now let $ ( X; A, B, * ) $ be a triad. The homotopy groups of a triad are defined as the relative homotopy groups
$$ \pi _ {n} ( X; A, B , * ) = \ \pi _ {n-} 1 ( \Omega ( X; B, * ),\ \Omega ( A; A \cap B , * ), * ). $$
Using the long homotopy sequence of the triplet $ ( \Omega ( X; B, * ) , \Omega ( A; A \cap B, * ), * ) $ there results the (first) homotopy sequence of a triad
$$ {} \dots \rightarrow \pi _ {n+} 1 ( X; A, B, * ) \mathop \rightarrow \limits ^ \partial \pi _ {n} ( A, A \cap B, * ) \rightarrow $$
$$ \rightarrow \ \pi _ {n} ( X, B, * ) \rightarrow \pi _ {n} ( X; A, B, x _ {0} ) \mathop \rightarrow \limits ^ \partial \dots , $$
so that the triad homotopy groups measure the extend to which the homotopy excision homomorphisms
$$ \pi _ {n} ( A, A \cap B, * ) \rightarrow \pi _ {n} ( X, B, * ) $$
fail to be isomorphisms. The triad homotopy groups can also be defined as
$$ \pi _ {n} ( X; A, B, * ) = \pi _ {n-} 1 ( \Omega ( X; A, B), * ). $$
[a1] | S.-T. Hu, "Homotopy theory" , Acad. Press (1955) pp. Chapt. V, §10 |
[a2] | B. Gray, "Homotopy theory. An introduction to algebraic topology" , Acad. Press (1975) pp. 88 |
[a3] | R.M. Switzer, "Algebraic topology - homotopy and homology" , Springer (1975) pp. §6.17 |