Let V be a vector space, a #~{metric} on V is a function _ d#: V # V --> &reals.^+ _ such that:
V is then said to be a _ #~{metric space} _ with metric d, and is represented by the pair (V, d)
(V, d) is a metric space. The _ #~{closed ball} _ in V with centre _ ~a _ radius _ ~r _ _ (~r &in. &reals.) _ is _ &set. ~x &in. V | d( ~a , ~x ) =< ~r &xset. _ denoted _ B[ ~a , ~r ]
The _ #~{open ball} _ in V with centre _ ~a _ radius _ ~r _ _ (~r &in. &reals.), _ is _ &set. ~x &in. V | d( ~a , ~x ) < ~r &xset. _ denoted _ B( ~a , ~r )
Let _ (A, d_A) _ and _ (B, d_B) be metric spaces. (We will write 'd' for both 'd_A' and 'd_B', it should be clear from context which is which).
f#: A --> B _ is a #~{continuous function} at a point ~a &in. A _ if given &epsilon. > 0 _ &exist. &delta. > 0 such that _ d( ~a, ~x ) < &delta. _ => _ d( f ( ~a ), f ( ~x ) ) < &epsilon. .
f is said to be #~{continuous (on A)} if it is continuous at every point of A.
If _ f#: A --> B _ and _ g#: B --> C _ are continuous, then _ g &comp. f #: A --> C is also continuous.
Proof:
Let _ f ( ~a ) = ~b _ for some ~a &in. A. Then given &epsilon. > 0 , _ &exist. &eta. > 0 _ such that _ d( g(~b), g ( ~x ) ) < &epsilon. _ if _ d( ~b, ~x ) < &eta. .
In particular if ~x is such that _ ~x = f ( ~y) _ some ~y &in. A , _ then &exist. &delta. > 0 _ such that _ d( f ( ~a ), f ( ~y) ) < &eta. _ if _ d( ~a, ~y ) < &delta. .
So given &epsilon. > 0 _ we have found &delta. > 0 _ such that _ d( (g &comp. f)( ~a ), (g &comp. f)( ~y ) ) < &epsilon. _ if _ d( ~a, ~y ) < &delta..
In the sequel we will consider functions whose co-domain is either &reals.^~n or &complex.^~n with the usual metric,
_ _ _ _ _ i.e. _ d( ~x , ~y ) _ = _ | ~x - ~y | _ = _ &sqrt.( &sum._~i (~x_~i - ~y_~i)^~n ) _ _ ~x , ~y &in. &reals.^~n _ ( similarly &complex.^~n ).
In particular we have _ | ~x - ~y | _ = _ | (~x + ~a) - ( ~y + ~a) | _ _ ("parallel shift").
If _ f#: A --> &reals. _ and _ g#: A --> &reals. _ are continuous then so are
Proof:
Proof: