Metric Spaces and Several Variables
CSIR NET real analysis: metric and normed spaces, compactness, Heine-Borel, connectedness, directional derivatives, inverse and implicit function theorems.
- 25 min
On this page · 15 sections
Prerequisites
- Sequences, Series, limsup and liminf: Cauchy sequences and Bolzano–Weierstrass, which this page generalises.
- Continuity, Differentiability and Bounded Variation: one-variable continuity and the MVT, used for every several-variable proof.
- Sequences and Series of Functions: \( C[a, b] \) with uniform convergence is the main infinite-dimensional example.
- Matrices, determinants and rank from linear algebra (Unit 1 of Mathematical Sciences).
The Hook: Closed, Bounded and Still Not Compact
Metric Spaces
A metric on \( X \) is \( d : X \times X \to [0, \infty) \) with \( d(x, y) = 0 \iff x = y \), \( d(x, y) = d(y, x) \), and \( d(x, z) \le d(x, y) + d(y, z) \).
| Space | Metric |
|---|---|
| \( \mathbb{R}^n \) | \( d_1 = \sum \lvert x_i - y_i \rvert \), \( d_2 = \left(\sum \lvert x_i - y_i \rvert^2\right)^{1/2} \), \( d_\infty = \max \lvert x_i - y_i \rvert \) |
| Any set | discrete: \( d(x, y) = 1 \) for \( x \ne y \) |
| \( C[a, b] \) | \( d_\infty(f, g) = \sup \lvert f - g \rvert \) or \( d_1(f, g) = \int_a^b \lvert f - g \rvert \) |
| \( \ell^p \), \( 1 \le p < \infty \) | \( \left(\sum \lvert x_n - y_n \rvert^p\right)^{1/p} \) on sequences with \( \sum \lvert x_n \rvert^p < \infty \) |
| Any metric space | \( \frac{d}{1 + d} \) and \( \min(d, 1) \): bounded metrics with the same open sets |
Open and closed sets
\( B(x, r) = \{y : d(x, y) < r\} \). \( U \) is open if every point has a ball inside \( U \); \( F \) is closed if its complement is open, equivalently if it contains all its limit points, equivalently if it contains the limit of every convergent sequence in it.
- Arbitrary unions and finite intersections of open sets are open. \( \bigcap_n (-\frac{1}{n}, \frac{1}{n}) = \{0\} \) shows infinite intersections can fail.
- \( \overline{A} = A \cup A' \) (closure = set plus limit points). \( A \) is dense if \( \overline{A} = X \).
- A set can be both open and closed (\( \emptyset \), \( X \), and in the discrete metric every set) or neither (\( [0, 1) \) in \( \mathbb{R} \)).
- In the discrete metric, \( B(x, 1) = \{x\} \) is closed, but the closed ball \( \{y : d(x, y) \le 1\} = X \). So the closure of an open ball need not be the closed ball (in normed spaces it always is).
Continuity. \( f : X \to Y \) is continuous iff \( f^{-1}(U) \) is open for every open \( U \), iff \( x_n \to x \Rightarrow f(x_n) \to f(x) \). Images of open sets need not be open (\( x \mapsto x^2 \) sends \( (-1, 1) \) to \( [0, 1) \)).
Completeness
\( X \) is complete if every Cauchy sequence converges in \( X \). A subset of a complete space is complete iff it is closed.
| Space | Complete? |
|---|---|
| \( \mathbb{R}^n \), closed subsets of \( \mathbb{R}^n \) | yes |
| \( \mathbb{Q} \), \( (0, 1) \), \( \mathbb{R} \setminus \{0\} \) | no |
| \( C[a, b] \) with \( d_\infty \) | yes (uniform limits of continuous functions are continuous) |
| \( C[0, 1] \) with \( d_1 \) | no (continuous functions approaching a step) |
| \( \ell^p \), \( 1 \le p \le \infty \) | yes |
| \( \mathbb{R} \) with \( \lvert \arctan x - \arctan y \rvert \) | no (\( x_n = n \) is Cauchy, no limit) |
| Any discrete metric space | yes (Cauchy sequences are eventually constant) |
Completeness is not a topological property: \( (0, 1) \) is homeomorphic to \( \mathbb{R} \) but not complete.
Both hypotheses matter. On \( X = (0, 1] \) (not complete), \( Tx = \frac{x}{2} \) has no fixed point. On \( [1, \infty) \), \( Tx = x + \frac{1}{x} \) satisfies \( \lvert Tx - Ty \rvert < \lvert x - y \rvert \) for \( x \ne y \) but has no fixed point: “strictly shrinking” is weaker than a contraction with a fixed \( c < 1 \).
Baire category theorem. A complete metric space is not a countable union of nowhere dense sets. Consequences: \( \mathbb{R} \) is uncountable; \( \mathbb{Q} \) is not a \( G_\delta \) set; a complete metric space without isolated points is uncountable.
Compactness
\( K \) is compact if every open cover has a finite subcover. In a metric space the following are equivalent:
- \( K \) is compact.
- Every sequence in \( K \) has a subsequence converging in \( K \) (sequential compactness).
- Every infinite subset of \( K \) has a limit point in \( K \).
- \( K \) is complete and totally bounded (for each \( \varepsilon > 0 \), finitely many \( \varepsilon \)-balls cover \( K \)).
Properties
- Compact sets are closed and bounded in every metric space. The converse is Heine–Borel, true only in special spaces such as \( \mathbb{R}^n \).
- A closed subset of a compact set is compact.
- The continuous image of a compact set is compact. So a continuous real function on a compact set is bounded and attains its max and min.
- A continuous function on a compact metric space is uniformly continuous.
- A continuous bijection from a compact space onto a metric space is a homeomorphism.
- Cantor intersection. A decreasing sequence of non-empty compact sets has non-empty intersection.
- Finite sets are compact; a convergent sequence together with its limit is compact.
| Set | Compact? |
|---|---|
| \( [0, 1] \), the Cantor set, \( \{0\} \cup \{\frac{1}{n}\} \) | yes |
| \( (0, 1] \), \( \{\frac{1}{n}\} \) without 0 | no (not closed) |
| \( \{(x, y) : \lvert x \rvert + \lvert y \rvert \le 1\} \), the unit sphere \( S^{n-1} \) | yes |
| \( \{(x, y) : xy = 1\} \) | no (closed, unbounded) |
| \( O(n) \), \( SO(n) \) in \( \mathbb{R}^{n^2} \) | yes (closed, entries bounded by 1) |
| \( GL(n, \mathbb{R}) \), \( SL(n, \mathbb{R}) \) for \( n \ge 2 \) | no |
| Closed unit ball of \( \ell^2 \) or \( C[0, 1] \) | no |
| An infinite discrete metric space | no (bounded and closed) |
Connectedness
\( X \) is connected if it is not the union of two disjoint non-empty open sets; equivalently, the only subsets that are both open and closed are \( \emptyset \) and \( X \). \( X \) is path-connected if any two points are joined by a continuous path \( [0, 1] \to X \).
- The connected subsets of \( \mathbb{R} \) are exactly the intervals (including single points).
- The continuous image of a connected set is connected. This is the intermediate value theorem in general form.
- Path-connected \( \Rightarrow \) connected. The converse fails: the topologist’s sine curve \( \{(x, \sin\frac{1}{x}) : 0 < x \le 1\} \cup (\{0\} \times [-1, 1]) \) is connected but not path-connected.
- An open connected subset of \( \mathbb{R}^n \) is path-connected (even by polygonal paths).
- The closure of a connected set is connected; a union of connected sets with a common point is connected.
| Set | Connected? |
|---|---|
| \( \mathbb{R}^n \setminus \{0\} \), \( n \ge 2 \) | yes (path-connected) |
| \( \mathbb{R} \setminus \{0\} \) | no |
| \( \mathbb{R}^2 \setminus \) (countable set), e.g. \( \mathbb{R}^2 \setminus \mathbb{Q}^2 \) | yes (path-connected) |
| \( \mathbb{Q} \), \( \mathbb{R} \setminus \mathbb{Q} \), the Cantor set | no (totally disconnected) |
| \( \{(x, y) : x^2 - y^2 = 1\} \) | no (two branches) |
| \( \{(x, y) : x^2 + y^2 = 1\} \) | yes |
| \( GL(n, \mathbb{R}) \) | no: \( \det \) is a continuous map onto \( \mathbb{R} \setminus \{0\} \); two components |
| \( GL(n, \mathbb{C}) \), \( SO(n) \) | yes |
| \( O(n) \) | no: \( \det \) takes only the values \( \pm 1 \) |
Using connectedness to show non-homeomorphism. \( \mathbb{R} \) and \( \mathbb{R}^2 \) are not homeomorphic: removing a point disconnects \( \mathbb{R} \) but not \( \mathbb{R}^2 \). Likewise \( [0, 1) \not\cong (0, 1) \) (removing 0 leaves the first connected; removing any point disconnects the second).
Normed Linear Spaces
A norm on a real vector space \( V \) satisfies \( \lVert x \rVert = 0 \iff x = 0 \), \( \lVert \lambda x \rVert = \lvert \lambda \rvert \lVert x \rVert \), and \( \lVert x + y \rVert \le \lVert x \rVert + \lVert y \rVert \). It induces the metric \( d(x, y) = \lVert x - y \rVert \). A complete normed space is a Banach space.
| Result | Statement |
|---|---|
| Finite dimension | all norms are equivalent: \( c\lVert x \rVert \le \lVert x \rVert' \le C\lVert x \rVert \) |
| Finite dimension | every normed space is complete; every linear map is continuous; Heine–Borel holds |
| Riesz | the closed unit ball is compact iff the space is finite-dimensional |
| Linear maps | continuous \( \iff \) continuous at 0 \( \iff \) bounded (\( \lVert Tx \rVert \le M\lVert x \rVert \)) |
| Operator norm | \( \lVert T \rVert = \sup_{\lVert x \rVert \le 1} \lVert Tx \rVert \) |
| Parallelogram law | \( \lVert x + y \rVert^2 + \lVert x - y \rVert^2 = 2\lVert x \rVert^2 + 2\lVert y \rVert^2 \) holds iff the norm comes from an inner product |
In \( \mathbb{R}^n \): \( \lVert x \rVert_\infty \le \lVert x \rVert_2 \le \lVert x \rVert_1 \le \sqrt{n}\,\lVert x \rVert_2 \le n\lVert x \rVert_\infty \).
Infinite dimensions break all of this. On \( C[0, 1] \), \( \lVert \cdot \rVert_\infty \) and \( \lVert \cdot \rVert_1 \) are not equivalent: \( f_n = x^n \) has \( \lVert f_n \rVert_\infty = 1 \) and \( \lVert f_n \rVert_1 = \frac{1}{n+1} \). Differentiation \( D : (C^1[0, 1], \lVert \cdot \rVert_\infty) \to (C[0, 1], \lVert \cdot \rVert_\infty) \) is linear and unbounded: \( \lVert x^n \rVert_\infty = 1 \), \( \lVert D x^n \rVert_\infty = n \). And the \( \lVert \cdot \rVert_1 \), \( \lVert \cdot \rVert_\infty \) norms on \( \mathbb{R}^2 \) fail the parallelogram law (try \( x = (1, 0) \), \( y = (0, 1) \)), so they come from no inner product.
Functions of Several Variables
Limits and continuity
\( \lim_{(x, y) \to (0, 0)} f = L \) means \( \lvert f(x, y) - L \rvert < \varepsilon \) whenever \( 0 < \lVert (x, y) \rVert < \delta \). To show a limit fails, find two paths with different limits; to show it exists, bound \( \lvert f - L \rvert \) by a function of \( r = \sqrt{x^2 + y^2} \) that tends to 0.
| \( f(x, y) \), \( (x, y) \ne 0 \) | Limit at 0 |
|---|---|
| \( \frac{xy}{x^2 + y^2} \) | none: \( \frac{m}{1 + m^2} \) along \( y = mx \) |
| \( \frac{x^2y}{x^4 + y^2} \) | none: 0 along every line, \( \frac{1}{2} \) along \( y = x^2 \) |
| \( \frac{x^2y}{x^2 + y^2} \) | 0: \( \lvert f \rvert \le \lvert y \rvert \) |
| \( \frac{x^2y^2}{x^2y^2 + (x - y)^2} \) | none: both iterated limits are 0, but the value is 1 along \( y = x \) |
| \( x\sin\frac{1}{y} + y\sin\frac{1}{x} \) (\( xy \ne 0 \)) | 0, although neither iterated limit exists |
Iterated limits and the double limit are independent: either can exist without the other, as the last two rows show.
Partial, directional and total derivatives
For \( f : \mathbb{R}^n \to \mathbb{R}^m \) and \( a \in \mathbb{R}^n \):
- Directional derivative along a unit vector \( u \): \( D_u f(a) = \lim_{t \to 0}\frac{f(a + tu) - f(a)}{t} \). Partial derivatives are \( D_{e_i} f \).
- Differentiable at \( a \): there is a linear \( T \) (the derivative \( Df(a) \)) with
Its matrix is the Jacobian \( \left(\frac{\partial f_i}{\partial x_j}(a)\right) \).
| Example at the origin (\( f(0, 0) = 0 \)) | Partials | All \( D_u f \) | Continuous | Differentiable |
|---|---|---|---|---|
| \( \frac{xy}{x^2 + y^2} \) | yes (both 0) | no | no | no |
| \( \frac{x^2y}{x^4 + y^2} \) | yes (both 0) | yes | no | no |
| \( \frac{x^2y}{x^2 + y^2} \) | yes (both 0) | yes, \( D_u f = u_1^2u_2 \) (not linear) | yes | no |
| \( \sqrt{\lvert xy \rvert} \) | yes (both 0) | no | yes | no |
| \( (x^2 + y^2)\sin\frac{1}{\sqrt{x^2 + y^2}} \) | yes, not continuous at 0 | yes (all 0) | yes | yes |
For \( \frac{x^2y}{x^4 + y^2} \), along \( u = (\cos\theta, \sin\theta) \) with \( \sin\theta \ne 0 \): \( \frac{f(tu)}{t} = \frac{\cos^2\theta\sin\theta}{t^2\cos^4\theta + \sin^2\theta} \to \frac{\cos^2\theta}{\sin\theta} \). So every directional derivative exists, yet \( f \) is discontinuous.
Chain rule. \( D(g \circ f)(a) = Dg(f(a)) \circ Df(a) \): Jacobian matrices multiply.
Mixed partials
Schwarz (Clairaut) theorem. If \( f_x \), \( f_y \) and \( f_{xy} \) exist near \( a \) and \( f_{xy} \) is continuous at \( a \), then \( f_{yx}(a) \) exists and equals \( f_{xy}(a) \). In particular this holds for \( C^2 \) functions.
Extrema
At a critical point (\( \nabla f(a) = 0 \)) of a \( C^2 \) function of two variables, with \( D = f_{xx}f_{yy} - f_{xy}^2 \) at \( a \):
| Condition | Conclusion |
|---|---|
| \( D > 0 \), \( f_{xx} > 0 \) | local minimum |
| \( D > 0 \), \( f_{xx} < 0 \) | local maximum |
| \( D < 0 \) | saddle point |
| \( D = 0 \) | no conclusion (\( x^4 + y^4 \): min; \( x^4 - y^4 \): saddle) |
In \( n \) variables: Hessian positive definite \( \Rightarrow \) local minimum; negative definite \( \Rightarrow \) maximum; indefinite \( \Rightarrow \) saddle. For constrained extrema of \( f \) on \( g = 0 \) with \( \nabla g \ne 0 \): \( \nabla f = \lambda\nabla g \) (Lagrange multipliers).
Inverse function theorem
Let \( f : U \to \mathbb{R}^n \) be \( C^1 \) on an open \( U \subseteq \mathbb{R}^n \), with \( Df(a) \) invertible (\( \det J_f(a) \ne 0 \)). Then there are open \( V \ni a \) and \( W \ni f(a) \) such that \( f : V \to W \) is a bijection with \( C^1 \) inverse, and \( D(f^{-1})(f(a)) = [Df(a)]^{-1} \).
- Local, not global. \( f(x, y) = (e^x\cos y, e^x\sin y) \) has \( \det J = e^{2x} \ne 0 \) everywhere but is not injective (\( 2\pi \)-periodic in \( y \)).
- The Jacobian condition is sufficient, not necessary, for injectivity. \( x \mapsto x^3 \) is a bijection of \( \mathbb{R} \) with derivative 0 at 0; its inverse is not differentiable at 0.
- \( C^1 \) cannot be dropped. \( f(x) = x + 2x^2\sin\frac{1}{x} \), \( f(0) = 0 \), has \( f'(0) = 1 \) but is not injective on any neighbourhood of 0 (\( f' \) takes negative values arbitrarily close to 0).
Implicit function theorem
Let \( F : \mathbb{R}^{n+m} \to \mathbb{R}^m \) be \( C^1 \) near \( (a, b) \) (\( a \in \mathbb{R}^n \), \( b \in \mathbb{R}^m \)), with \( F(a, b) = 0 \) and the \( m \times m \) matrix \( D_yF(a, b) = \left(\frac{\partial F_i}{\partial y_j}\right) \) invertible. Then near \( a \) there is a unique \( C^1 \) function \( g \) with \( g(a) = b \) and \( F(x, g(x)) = 0 \).
For one equation \( F(x, y) = 0 \) with \( F_y(a, b) \ne 0 \):
$$\boxed{\frac{dy}{dx} = -\frac{F_x}{F_y}}, \qquad \text{and for } F(x, y, z) = 0:\ \frac{\partial z}{\partial x} = -\frac{F_x}{F_z}.$$The condition is sufficient, not necessary: \( F(x, y) = y^3 - x \) has \( F_y(0, 0) = 0 \), yet \( y = x^{1/3} \) is a (non-differentiable) solution.
Counter-examples Table
| False claim | Counter-example |
|---|---|
| Closed and bounded \( \Rightarrow \) compact (any metric space) | unit ball of \( \ell^2 \) or \( C[0, 1] \); \( \mathbb{R} \) with \( \min(\lvert x - y \rvert, 1) \) |
| Closure of \( B(x, r) \) = closed ball | discrete metric, \( r = 1 \) |
| Completeness is preserved by homeomorphism | \( (0, 1) \cong \mathbb{R} \) |
| \( d(Tx, Ty) < d(x, y) \) on a complete space \( \Rightarrow \) fixed point | \( x + \frac{1}{x} \) on \( [1, \infty) \) |
| Connected \( \Rightarrow \) path-connected | topologist’s sine curve |
| Continuous maps send open sets to open sets | \( x^2 \) on \( (-1, 1) \) |
| All norms on a vector space are equivalent | \( \lVert \cdot \rVert_1 \), \( \lVert \cdot \rVert_\infty \) on \( C[0, 1] \) |
| Every linear map between normed spaces is continuous | differentiation on \( (C^1[0, 1], \lVert \cdot \rVert_\infty) \) |
| Partials exist \( \Rightarrow \) continuous | \( \frac{xy}{x^2 + y^2} \) |
| All directional derivatives exist \( \Rightarrow \) continuous | \( \frac{x^2y}{x^4 + y^2} \) |
| Continuous with all directional derivatives \( \Rightarrow \) differentiable | \( \frac{x^2y}{x^2 + y^2} \) |
| Differentiable \( \Rightarrow \) partials continuous | \( (x^2 + y^2)\sin\frac{1}{\sqrt{x^2 + y^2}} \) |
| \( f_{xy} = f_{yx} \) always | \( \frac{xy(x^2 - y^2)}{x^2 + y^2} \) |
| Iterated limits equal \( \Rightarrow \) limit exists | \( \frac{x^2y^2}{x^2y^2 + (x - y)^2} \) |
| Non-zero Jacobian everywhere \( \Rightarrow \) globally injective | \( (e^x\cos y, e^x\sin y) \) |
| Injective \( C^1 \) \( \Rightarrow \) non-zero Jacobian | \( x^3 \) |
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| closed and bounded (necessary for compact) | compact (needs Heine–Borel to be equivalent) | |
| bounded (depends on the metric) | totally bounded (what compactness needs) | |
| connected | path-connected (stronger; equal for open subsets of \( \mathbb{R}^n \)) | |
| complete (metric property) | closed (relative property: \( (0, 1) \) is closed in itself) | |
| existence of partial derivatives | differentiability | |
| local inverse (IFT) | global inverse |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper. In Level 3, as in Part C, one or more options may be correct.
Level 1: Foundation
- In the discrete metric on an infinite set \( X \), which subsets are open? Which are compact?
- Is \( (0, 1) \) with the usual metric complete?
- State the condition in the implicit function theorem for \( F(x, y) = 0 \) to define \( y = g(x) \) near \( (a, b) \).
Answers
- Every subset is open (\( \{x\} = B(x, 1) \)). Exactly the finite subsets are compact: the cover by singletons has no finite subcover of an infinite set.
- No. \( \frac{1}{n} \) is Cauchy with no limit in \( (0, 1) \).
- \( F \) is \( C^1 \) near \( (a, b) \), \( F(a, b) = 0 \) and \( F_y(a, b) \ne 0 \).
Level 2: Part B Style
Practice Q1. Let \( f(x, y) = \frac{xy}{x^2 + y^2} \) for \( (x, y) \ne (0, 0) \) and \( f(0, 0) = 0 \). At the origin:
(a) \( f \) is continuous (b) both partial derivatives exist, but \( f \) is not continuous (c) \( f \) is differentiable (d) neither partial derivative exists
Answer: (b)
\( f(x, 0) = 0 \) and \( f(0, y) = 0 \), so \( f_x(0, 0) = f_y(0, 0) = 0 \). Along \( y = x \), \( f = \frac{1}{2} \ne 0 \), so \( f \) is not continuous, hence not differentiable.
Practice Q2. Which subset of \( \mathbb{R}^2 \) is compact?
(a) \( \{(x, y) : x^2 + y^2 < 1\} \) (b) \( \{(x, y) : xy = 1\} \) (c) \( \{(x, y) : \lvert x \rvert + \lvert y \rvert \le 1\} \) (d) \( \{(x, \sin\frac{1}{x}) : 0 < x \le 1\} \)
Answer: (c)
(c) is closed (preimage of \( [0, 1] \) under a continuous map) and bounded. (a) is not closed; (b) is unbounded; (d) is bounded but not closed (its closure adds \( \{0\} \times [-1, 1] \)).
Practice Q3. Which of the following metric spaces is complete?
(a) \( \mathbb{Q} \) (b) \( (0, 1] \) (c) \( C[0, 1] \) with \( d(f, g) = \sup\lvert f - g \rvert \) (d) \( C[0, 1] \) with \( d(f, g) = \int_0^1 \lvert f - g \rvert \)
Answer: (c)
A sup-Cauchy sequence of continuous functions converges uniformly, and the limit is continuous. (a) misses \( \sqrt{2} \); (b) misses 0; in (d), continuous ramps converging in \( d_1 \) to a step function form a Cauchy sequence with no continuous limit.
Practice Q4. The equation \( x^2 + y^2 + z^2 = 3 \) defines \( z \) as a \( C^1 \) function of \( (x, y) \) near \( (1, 1, 1) \). Then \( \frac{\partial z}{\partial x}(1, 1) \) equals:
(a) \( -1 \) (b) 1 (c) 0 (d) \( -\frac{1}{2} \)
Answer: (a)
\( F = x^2 + y^2 + z^2 - 3 \), \( F_z = 2z = 2 \ne 0 \). \( \frac{\partial z}{\partial x} = -\frac{F_x}{F_z} = -\frac{2x}{2z} = -1 \).
Level 3: Part C Style
Practice Q5. Which statements are true?
(a) Every closed and bounded subset of every metric space is compact. (b) The continuous image of a connected metric space is connected. (c) Every compact subset of a metric space is closed and bounded. (d) The closed unit ball of \( C[0, 1] \) with the sup norm is compact.
Answer: (b) and (c)
- (a) False. An infinite set with the discrete metric is closed and bounded but not compact.
- (b) True. If \( f(X) = A \cup B \) were a separation, \( f^{-1}(A) \), \( f^{-1}(B) \) would separate \( X \).
- (c) True. Compact sets are totally bounded (so bounded) and contain the limits of their convergent sequences (so closed).
- (d) False. By Riesz’s theorem the closed unit ball of an infinite-dimensional normed space is not compact; the disjoint tents of the hook have mutual distance 1.
Practice Q6. Which of the following subsets are connected?
(a) \( \mathbb{R}^2 \setminus \mathbb{Q}^2 \) (b) \( \{(x, y) \in \mathbb{R}^2 : x^2 - y^2 = 1\} \) (c) \( GL(2, \mathbb{R}) \), as a subset of \( \mathbb{R}^4 \) (d) \( \{(x, \sin\frac{1}{x}) : 0 < x \le 1\} \cup (\{0\} \times [-1, 1]) \)
Answer: (a) and (d)
- (a) Connected. Any point of \( \mathbb{R}^2 \setminus \mathbb{Q}^2 \) lies on a horizontal or vertical line with an irrational coordinate, entirely inside the set; the line \( y = \sqrt{2} \) meets all vertical lines \( x = \alpha \) (\( \alpha \) irrational), and all horizontal lines \( y = \beta \) (\( \beta \) irrational) meet \( x = \sqrt{2} \). So the set is path-connected.
- (b) Not connected. \( x^2 = 1 + y^2 \ge 1 \), so the set splits into the disjoint closed pieces \( x \ge 1 \) and \( x \le -1 \).
- (c) Not connected. \( \det \) maps it continuously onto \( \mathbb{R} \setminus \{0\} \), which is disconnected.
- (d) Connected. It is the closure of the connected graph of \( \sin\frac{1}{x} \) on \( (0, 1] \). (It is not path-connected, but the question asks only about connectedness.)
Practice Q7. Let \( f(x, y) = \frac{x^2y}{x^4 + y^2} \) for \( (x, y) \ne (0, 0) \) and \( f(0, 0) = 0 \). Which statements are true?
(a) All directional derivatives of \( f \) at the origin exist. (b) \( f \) is continuous at the origin. (c) \( f \) is differentiable at the origin. (d) \( \frac{\partial f}{\partial x}(0, 0) = \frac{\partial f}{\partial y}(0, 0) = 0 \).
Answer: (a) and (d)
- (a) True. For \( u = (\cos\theta, \sin\theta) \) with \( \sin\theta \ne 0 \), \( \frac{f(tu)}{t} \to \frac{\cos^2\theta}{\sin\theta} \); for \( \sin\theta = 0 \), \( f(tu) = 0 \) and the derivative is 0.
- (b) False. Along \( y = x^2 \), \( f = \frac{x^4}{2x^4} = \frac{1}{2} \).
- (c) False. Differentiable functions are continuous. (Also, the directional derivatives are not linear in \( u \).)
- (d) True. \( f(x, 0) = 0 \) and \( f(0, y) = 0 \).
Quick Revision Table
| Idea | Fact |
|---|---|
| Compact (metric) | \( \iff \) sequentially compact \( \iff \) complete + totally bounded |
| Heine–Borel | in \( \mathbb{R}^n \): compact \( \iff \) closed and bounded |
| Riesz | unit ball compact \( \iff \) finite-dimensional |
| Banach fixed point | contraction on complete space: unique fixed point |
| Baire | complete space is not a countable union of nowhere dense sets |
| Connected in \( \mathbb{R} \) | intervals |
| Path-connected | \( \Rightarrow \) connected; converse fails (sine curve) |
| Finite-dim normed | norms equivalent, complete, linear maps continuous |
| Linear map | continuous \( \iff \) bounded |
| Differentiability | \( C^1 \Rightarrow \) diff \( \Rightarrow \) cont + \( D_uf = \nabla f\cdot u \) |
| Clairaut | \( f_{xy} \) continuous \( \Rightarrow f_{xy} = f_{yx} \) |
| Inverse function | \( C^1 \), \( \det J \ne 0 \) \( \Rightarrow \) local \( C^1 \) inverse |
| Implicit function | \( F_y \ne 0 \Rightarrow y = g(x) \), \( g' = -\frac{F_x}{F_y} \) |
Key Takeaways
What’s Next
- Real Analysis: Quick Revision: the whole chapter on one page.
- Back to the start: Real Numbers, Sets and Countability, where completeness first appears.
- Riemann, Improper and Lebesgue Integration for the \( L^1 \) metric and its completion.
- The Mathematical Sciences overview for how this unit fits the paper.