Professional Documents
Culture Documents
Instructor: S Kumaresan
We shall try to be as faithful as possible to the title of the course “Differential Calculus on
Rn ". We shall start with basic results and go up to the inverse and implicit function theorems.
Up to this point, we shall be working in the context of Banach spaces. After a brief review
of integration on Rn , we shall prove the change of variable formula. We shall outline the
modern approach to the Stokes theorem. Depending on the interest of the audience we shall
either discuss submanifolds of Rn or the degree theory.
There will be periodical assignments. You are expected to submit it on or before the
deadline. You may consult any book or discuss with anybody on the assignments, but you
are to write on your own. No shoddy submission will be accepted and it will be treated as
non-submission.
Out of the total 100 marks for the course, 40 will be earmarked for the assignments.
There will be two oral examinations: one midsemester and the other final. Each will carry
30 marks each. The oral examination will focus on your understanding of the subject, will
assess how honest you have been in the assignments and on your problem solving abilities.
We may replace one oral examination by a seminar.
Grading policy:
1
Assignment 1
2. With the notation as above, let p (resp. q) be equivalent norms on X (resp. on Y ). Prove
that f : (X , k k) → (Y, k k) is differentiable at a iff f : (X , p) → (Y, q) is differentiable
at a with the same derivative.
3. Define the directional derivative Dv f (a). Show that if f is differentiable, the directional
derivatives exists for any v ∈ X . State and prove the relation between D f (a) and
Dv f (a).
(a) A 7→ A2 on B L(X ).
(b) A 7→ A−1 on G L(X ). (You also need to show that G L(X ) is open.)
The assignment is due on 8 August 2019. Write your name, Id etc., legibly along with the
date. Staple the pages before submission. Make sure that the submission is not shoddy and
that you are proud of it.
2
Assignment 2
2. Let H be a (real) Hilbert space. Consider the map f (x) := 〈x, x〉. Prove that f is
differentiable and find its gradient at v ∈ H.
More generally, let A ∈ B L(H) be self-adjoint. Let g(x) := 〈Ax, x〉. Prove that g is
differentiable and find its gradient at any point.
6. Let Hbe a Hilbert space. Let γ: R → H be a function such that 〈γ(t), γ(t)〉 = 1 for
t ∈ R. How are γ0 (t) and γ(t) are related? What does it mean if H = Rn ?
3
Assignment 3
4
Assignment 4
newpage Assignment-6
1. Let U ⊂ Rn be a star-shaped open set containing 0. Establish 2nd order Taylor expan-
sion for C 3 functions f : U → R
2. State and prove the 2nd derivative test for local extrema. Apostol may be use here! ,
3. Formulate and prove the basic existence and uniqueness theorem for initial value prob-
lems of an ODE. (The version should talk of the existence of integral curves for any
x ∈ B(x 0 , r) with a common domain in R.)
4. Identify Tp S where (i) S is an open set in Rn , (ii) S is the plane defined by 〈x, a〉 = b
where a ∈ Rn is a fixed nonzero vector and b ∈ R, and (iii) S is a graph surface of a
C 1 -function.
7. Use Lagrange multiplier rule to prove the AM-GM Pinequality. Consider the function
f (x) := (x 1 x 2 · · · x n ) with the constraint g(x) := k x k − 1 = 0.
2 2
a b
8. Find out the minimum of f (A) := a + b + c + d where A :=
2 2 2 2
∈ S L(2, R).
c d
10. Let An×m be a real matrix. Consider it as a linear map from Rm to Rn . Assume that the
spaces are equipped with Euclidean norm. Then the operator norm kAk is the square
root of (necessarily nonnegative) maximum eigenvalue of the m × m symmetric matrix
AT A.
1. State and prove the inverse function theorem (as was done in the class). To start with
write down the mean value inequality (without proof) which is needed often in the
proof. Do not consult any book for writing the statement of the theorem. If you have
understood it well, you should be able to write it on your own.
2. Compare your proof with the one in Rudin’s Principles of Mathematical Analysis. There
will be certainly some questions related to this during the Oral exams.
5
4. Let f : Rn → Rn be C 1 . Assume that k f (x) − f ( y)k ≥ k x − y k. Prove that f is a C 1 -
diffeomorphism of Rn onto itself. Hint: It is clear that f is one-one. Let B := f (Rn ). It is
easy to show that B is closed. Use the ‘expansive’ property of f and the differentiabilty
of f to prove that for each x ∈ Rn , there exists δ > 0 such that k D f (x)hk ≥ 12 khk for
khk < δ. Hence D f (x) is nonsingular and we can invoke IMT to conclude that B is
open.
Note that this argument works for any finite dimensional normed linear space.
5. State and prove the implicit function theorem. Compute the derivative of “g”.
6
Assignment 5
K f (x) := k(x, y) f ( y) d y
0
is a continuous linear map on the Banach space C(I). Assume that λ ∈ / spec(K) := {λ ∈
C : (K − λI) ∈
−1
/ B L(C(I))}. Prove that there exists an " > 0 such that the nonlinear
equation
Z1
λ f (x) = k(x, y) f ( y) + [ f ( y)]2 d y + g(x) (1)
0
Prove that there exist U and V neighbourhoods of 0 such that for each y ∈ V , there
exists a unique x ∈ U with the following properties:
(i) x is a solution of the equation (2)
(ii) y 7→ x( y) is C 1
(iii) x 0 (0) = 1.
7
Assignment-6
1. Let U ⊂ Rn be a star-shaped open set containing 0. Establish 2nd order Taylor expan-
sion for C 3 functions f : U → R
2. State and prove the 2nd derivative test for local extrema. Look up Apostol! ,
3. Formulate and prove the basic existence and uniqueness theorem for initial value prob-
lems of an ODE. (The version should talk of the existence of integral curves for any
x ∈ B(x 0 , r) with a common domain in R.)
4. Identify Tp S where (i) S is an open set in Rn , (ii) S is the plane defined by 〈x, a〉 = b
where a ∈ Rn is a fixed nonzero vector and b ∈ R, and (iii) S is a graph surface of a
C 1 -function.
7. Use Lagrange multiplier rule to prove the AM-GM Pinequality. Consider the function
f (x) := (x 1 x 2 · · · x n )2 with the constraint g(x) := k x k2 − 1 = 0.
a b
8. Find out the minimum of f (A) := a + b + c + d where A :=
2 2 2 2
∈ S L(2, R).
c d
10. Let An×m be a real matrix. Consider it as a linear map from Rm to Rn . Assume that the
spaces are equipped with Euclidean norm. Then the operator norm kAk is the square
root of (necessarily nonnegative) maximum eigenvalue of the m × m symmetric matrix
AT A.
8
Assignment-7 & 8
1. Define the 1-form d f . Find an explicit expression for d f in terms of the basic 1-forms
d xi.
7. Define the exterior differenal operator d : Λk → Λk+1 . How does it genetralise the
classical vector analysis operators gradient, curl and divergence?
8. Prove that d ◦ d = 0.
10. Give an example of a closed 1-form on R2 \ {(1, 0)} which is not exact .
11. Let ϕ be a k-form. We define the interior multipliction of ϕ by a vector field X by setting
15. We say that a form ϕ is exact if there exists a form η such that dη = ϕ. Prove that if ϕ
is exact and ψ is closed, then ϕ ∧ ψ is exact.