Scribd
Upload
17 views2 pages

Rectifiablity

1. The document discusses rectifiable curves and defines the length of a curve C as the supremum of the lengths of polygonal paths approximating C. 2. A curve is rectifiable if its coordinate functions are of bounded variation. Functions of bounded variation have finite total variation, which is analogous to the length of a curve. 3. If the curve C is given by a C1 function g, then the length of C exists and is equal to the integral of the absolute value of the derivative of g over the interval.

Uploaded by

Nidhi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
17 views2 pages

Rectifiablity

1. The document discusses rectifiable curves and defines the length of a curve C as the supremum of the lengths of polygonal paths approximating C. 2. A curve is rectifiable if its coordinate functions are of bounded variation. Functions of bounded variation have finite total variation, which is analogous to the length of a curve. 3. If the curve C is given by a C1 function g, then the length of C exists and is equal to the integral of the absolute value of the derivative of g over the interval.

Uploaded by

Nidhi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

RECTIFIABLE CURVES

1. Discussion of the definition


Assume that the curve C is given by the graph of g, C = g([a, b]) Given a partition
{t0 , ..., tJ } of [a, b], the length of a polygonal path through C is
J
X
(1) LP (C) = |g(tj ) − g(tj−1 )|
1

Since a straight line gives the shortest distance between two points, the length of C, if
it exists, is always larger than LP (C). Note that LP is increasing in P : if P 0 ⊃ P is a
subpartition of P then, by the triangle inequality (think of the finer polygon), LP (C) ≤
LP 0 (C). Thus, the sup over all partitions L(C) is in a good sense the limit of LP when
the partition becomes finer and finer and it is natural to call L(C) the length of C.

2. When are curves rectifiable?


Let f be a scalar function defined on [a, b]. Its total variation is defined very similarly
to L(C):
J
X
(2) Vba (f ) = sup |f (xj ) − f (xj−1 )| < ∞
P
1

where P are partitions of [a, b]. The space of functions of bounded variation BV is defined
as the space of functions for which Vba (f ) is finite:

(3) f ∈ BV ([a, b]) ⇐⇒ Vba (f ) < +∞

These notions were introduced by Camille Jordan (the same who introduced what we now
call the Jordan measure).

Note 1. If g = (g1 , ..., gn ) it is easy to see that g is rectifiable on [a, b] iff gi ∈ BV ([a, b])
for each i = 1, 2, ..., n.

This is because of the equivalence of Euclidian norms, (1.3) on p.6 in the text. Note
that Vba (gi ) is the length of the graph of gi on [a, b]. It is known that functions in BV
are differentiable almost everywhere (that is, except perhaps on a set of zero Lebesgue
measure).
If |g0 | is continuous, the construction of its Riemann integral is closely related to that
of L(C), and the proof below follows this connection.

Exercise 2 (Finite differences versus derivatives). Consider the function



 g(y) − g(x)
; if x 6= y
(4) h(x, y) = y−x
 g0 (x); if x = y
R1
Show that h(x, y) = 0 g0 tx + (1 − t)y dt. Use this to show that h is continuous in

2
(x, y) ∈ [a, b] .
1
2 RECTIFIABLE CURVES

3. The length formula when g ∈ C 1


Rb
Theorem 3. If g be C 1 on (a0 , b0 ) ⊃ [a, b], then L(C) exists and equals a
|g0 (t)|dt.
Note 4. Using the exercise, the idea is completely straightforward: with mj the minimum
point of g between tj−1 and tj we can write
(5) g(tj ) − g(tj−1 ) = g0 (mj )(tj − tj−1 ) + j (tj − tj−1 )
where j is uniformly small on [a, b]. Thus for a fine enough partition P , LP (C) is arbi-
trarily close to the lower Riemann sum sP (|g0 |) (we could have equally well worked with
the maximum point Mj and the upper sum SP ).
Proof of the theorem. The proof merely consists of writing down the details in the note
carefully.
Since |g0 | is continuous, it is integrable on [a, b] and there is a partition P 0 s.t.
Z b X Z b
0
sP 0 (|g |) − |g0 (t)|dt = |g0 (mj )| − |g0 (t)|dt < 

a 16j6J 0 a

Since [a, b] is compact, by the exercise above h is uniformly continuous on [a, b]2 .
2

Thus for any  there is a δ such that |(x, y) − (x0 , y 0 )| < δ ⇒ |h(x, y) − h(x0 , y 0 )| < . In
particular,
(6) |(x, y) − (x, y 0 )| < δ ⇒ |h(x, y) − g0 (x)| < 
Choose any partition P and take a refinement, if necessary, to arrange that tj −tj−1 < δ.
For each j, choose mj ∈ [tj−1 , tj ] to be the point where |g0 (m j )| is minimum. For any
Rb
partition P 00 finer than both, P 00 ⊃ P, P 00 ⊃ P 0 , we then have sP 00 (|g0 |) − a |g0 (t)|dt < .

(Review the notions if you forgot them.) By (6) and the choice of P 0 ,
(7) g(tj ) − g(tj−1 ) − g0 (mj )(tj − tj−1 ) = j (tj − tj−1 ); where |j | < 
Thus, by summing,
(8)
X X
LP (C) ≤ LP 00 (C) = |g(tj ) − g(tj−1 )| = |g0 (mj )|(tj − tj−1 ) + 0 (b − a)
16j6J 00 16j6J 00
Z b
= sP 00 (|g0 |) + 0 (b − a) = |g0 (t)|dt + 00 + 0 (b − a); where |0 |, |00 | < 
a
Since P was arbitrary and we can take  arbitrarily small, it follows immediately from
these inequalities that
Z b
(9) L(C) = sup LP (C) = |g0 (t)|dt
P a

You might also like