Proof-based real analysis, online
Real analysis tutor online: learn to write the proofs, with Ajay Vatsyayan
Real analysis is the course where calculus has to be proved, and many students find the jump from computing to proving the hardest step of their degree. Ajay Vatsyayan teaches it one to one online: how to read a definition, how to plan an epsilon argument in scratch work, and how to write a proof a grader accepts, following your own course notes.

1:1
Online sessions
Proofs
Read, planned and written with you
Your notes
Definitions as your lecturer states them
I understand the lecture while it's happening, but faced with a blank page and 'prove that aₙ → 3' I don't know how to start.
My proofs get marked down for 'logic' and I can't see what's wrong with them.
I passed calculus easily, and now supremum, Cauchy sequences and uniform convergence feel like a different subject.
Why it feels hard
Why the first analysis course feels like a different subject
In calculus you were asked to find things: a derivative, an integral, a sum. In analysis you are asked to show that things are true, using only definitions and results already proved. Imperial College London describes university mathematics as being 'more about problem solving and rigorous proof than about tackling routine exercises and calculations', and real analysis is where most students feel that change first.
Strong calculus students are often surprised to struggle. The difficulty is rarely the ideas, which are about limits and continuity they already know. It is the language: quantifiers in a fixed order, inequalities instead of equations, and arguments where each line must follow from the one before. That language can be taught directly, and that is what sessions focus on.
- Reading 'for every ε > 0 there exists N' and not seeing that N is allowed to depend on ε, but ε is not allowed to depend on n.
- Writing the scratch work (working backwards from what you want) as if it were the proof.
- Negating a definition wrongly, so a proof by contradiction proves nothing.
- Using a picture or a calculus fact that the course has not yet proved.
- Stopping once an inequality looks plausible instead of finishing the chain to < ε.
Course mapping
Where real analysis sits in your degree
In UK mathematics degrees analysis usually starts in the first year. Imperial's first-year mathematics programme includes Analysis I, and the University of Manchester lists Real Analysis (MATH11112) and Mathematical Foundations & Analysis (MATH11121) among its Year 1 units for mathematics and statistics.
In the US, real analysis is usually an upper-level course taken after the calculus sequence and linear algebra, often after an introduction to proofs course. Titles and numbers vary between universities, so send us your course page and syllabus. In India, real analysis is a core part of B.Sc. and M.Sc. mathematics. Master's students who need the analysis behind a course or project can book sessions on just those ideas.
Most first courses follow a similar path: the real numbers and completeness, sequences, series, limits and continuity, differentiation, the Riemann integral, then sequences and series of functions. A second course may move into metric spaces or measure theory; tell us which topics your syllabus includes.
Topic map
Real analysis, topic by topic: what you must be able to prove
This is the map Ajay uses to plan a course. The third column is the kind of question that appears on problem sheets and finals; the fourth is where we most often see marks disappear in a student's written proof.
| Topic | What you must be able to prove | Typical question | Where marks go |
|---|---|---|---|
| Real numbers and completeness | Properties of supremum and infimum; the Archimedean property | Find sup and inf of a set and prove they are correct | Showing an upper bound but not that it is the least one |
| Convergence of sequences | aₙ → L from the ε-N definition | Prove (3n + 1)/(n + 2) → 3 | Choosing N that depends on n |
| Limit laws and the squeeze theorem | Sum, product and quotient of convergent sequences | Prove the product of convergent sequences converges | Not bounding one factor first |
| Monotone sequences and Bolzano–Weierstrass | Bounded monotone sequences converge; bounded sequences have convergent subsequences | Show a recursively defined sequence converges and find its limit | Assuming the limit exists before proving it |
| Cauchy sequences | Cauchy is equivalent to convergent in ℝ | Prove a sequence is Cauchy without knowing its limit | Confusing |aₙ₊₁ − aₙ| → 0 with the Cauchy condition |
| Series | Comparison, ratio and alternating series tests from the definitions | Prove Σ 1/n diverges; decide absolute or conditional convergence | Citing a test whose hypotheses are not checked |
| Limits of functions and continuity | ε-δ limits; sequential characterisation | Prove x² − x → 6 as x → 3 | Letting δ depend on x |
| Properties of continuous functions | Intermediate value and extreme value theorems | Show a polynomial has a root in an interval | Using the theorem on a set that is not closed and bounded |
| Uniform continuity | Uniform versus pointwise continuity; continuous on [a, b] implies uniform | Show 1/x is not uniformly continuous on (0, 1) | Negating the definition in the wrong order |
| Differentiation | Rolle's theorem, the mean value theorem, Taylor's theorem with remainder | Use the MVT to prove an inequality such as |sin a − sin b| ≤ |a − b| | Forgetting to state continuity and differentiability conditions |
| Riemann integral | Upper and lower sums; integrability of continuous functions; the fundamental theorem | Show a step function is integrable using partitions | Using antiderivatives before the theorem allows it |
| Sequences and series of functions | Pointwise versus uniform convergence; what uniform convergence preserves | Decide whether fₙ converges uniformly on a given set | Computing the pointwise limit but not the supremum of the error |
Writing proofs
How an analysis proof is built: scratch work first, proof second
Almost every epsilon proof is written in two passes, and most marking problems come from mixing them up. The first pass is private: you start from the inequality you want and work backwards to discover how large N, or how small δ, must be. The second pass is the proof itself, written forwards: given ε, here is N; take any n beyond it; here is the chain of inequalities ending in ε.
Ajay teaches the two passes as separate skills. Students keep a scratch column and a proof column on the shared screen, and in the early weeks every proof is checked line by line for three things: is every symbol introduced before it is used, are the quantifiers in the order the definition demands, and does each inequality follow from something already stated.
1. State the goal
Write out the definition you must satisfy with the actual sequence or function substituted in.
2. Scratch work
Simplify |aₙ − L| or |f(x) − L|, bound the awkward factor, and find the condition on n or δ.
3. Write it forwards
Let ε > 0. Choose N (or δ). Suppose n ≥ N (or 0 < |x − a| < δ). Then a chain of inequalities ending below ε.
4. Conclude
Say which definition is now satisfied. Graders look for that final sentence.
Worked examples
Three complete proofs in the style of a first analysis course
These are original problems. The steps are written as a final proof, with the scratch work summarised where it decides the choice of N or δ.
An ε-N proof for a sequence
Prove from the definition that aₙ = (3n + 1)/(n + 2) converges to 3.
- Scratch work: |aₙ − 3| = |3n + 1 − 3(n + 2)|/(n + 2) = 5/(n + 2), and 5/(n + 2) < 5/n, so it is enough to make 5/n < ε, that is n > 5/ε.
- Proof. Let ε > 0. By the Archimedean property, choose a natural number N with N > 5/ε.
- Let n ≥ N. Then |aₙ − 3| = 5/(n + 2) < 5/n ≤ 5/N < ε.
- Since ε > 0 was arbitrary, aₙ → 3 by the definition of convergence.
Answer: Proved, with N any natural number greater than 5/ε (for ε = 0.1, N = 51 works).
N depends only on ε. Any natural N greater than 5/ε − 2 also works; the cruder bound 5/n just makes the choice easier. What matters is that N is fixed before n is chosen.
An ε-δ proof for a limit of a function
Prove that the limit of f(x) = x² − x as x → 3 is 6.
- Scratch work: |f(x) − 6| = |x² − x − 6| = |x − 3|·|x + 2|. The factor |x + 2| must be controlled. If |x − 3| < 1 then 2 < x < 4, so 4 < x + 2 < 6 and |x + 2| < 6.
- Proof. Let ε > 0 and put δ = min(1, ε/6). Then δ > 0.
- Suppose 0 < |x − 3| < δ. Since δ ≤ 1, we have |x − 3| < 1, so |x + 2| < 6.
- Then |f(x) − 6| = |x − 3|·|x + 2| < δ·6 ≤ (ε/6)·6 = ε.
- So for every ε > 0 there is δ > 0 with |f(x) − 6| < ε whenever 0 < |x − 3| < δ, which is the definition of the limit being 6.
Answer: Proved, with δ = min(1, ε/6).
The min is the key idea: the first restriction (δ ≤ 1) bounds |x + 2|; the second makes the product small.
Uniform convergence of a sequence of functions
Let fₙ(x) = x/(1 + nx²) for x in ℝ. Show that fₙ → 0 uniformly on ℝ.
- Pointwise limit: for each fixed x, fₙ(x) → 0 as n → ∞ (if x = 0 every term is 0; otherwise the denominator grows like nx²).
- Uniform bound: by the AM-GM inequality, 1 + nx² ≥ 2√n·|x|. So for x ≠ 0, |fₙ(x)| = |x|/(1 + nx²) ≤ |x|/(2√n·|x|) = 1/(2√n), and the bound also holds at x = 0.
- So sup over x in ℝ of |fₙ(x) − 0| ≤ 1/(2√n), with equality at x = 1/√n, where fₙ = (1/√n)/2.
- Let ε > 0 and choose N with N > 1/(4ε²). For n ≥ N, 1/(2√n) ≤ 1/(2√N) < ε, so |fₙ(x)| < ε for every x at once.
- That is the definition of uniform convergence to 0 on ℝ.
Answer: fₙ → 0 uniformly on ℝ, since sup |fₙ| = 1/(2√n) → 0.
Contrast gₙ(x) = xⁿ on [0, 1]: its pointwise limit jumps from 0 to 1 at x = 1, so the convergence cannot be uniform, because a uniform limit of continuous functions is continuous.
How sessions work
How we teach real analysis online
Sessions are online with handwritten proofs on a shared screen, at a time agreed by your time zone. Because analysis depends on the exact wording of definitions, we use your lecturer's versions, not a textbook's.
1. Before the first session
Send your lecture notes or textbook (Abbott, Bartle and Sherbert or your own department's notes), the current problem sheet, a marked piece of work with the grader's comments, and exam dates.
2. Diagnostic
You talk through two short proofs aloud. Hearing how you read the definition tells Ajay whether to start with logic and quantifiers or go straight to the course material.
3. Weekly pattern
One definition or theorem in depth, a model proof built from scratch work, then a proof you write while Ajay comments line by line.
4. Exam preparation
A list of definitions and theorems you must state exactly, the standard proofs your course expects you to reproduce, and timed past papers.
5. After the exam
We go through the marked script and plan for the next course, whether that is a second analysis course, metric spaces or complex analysis.
First month
Our usual first month with a new analysis student
This is our usual approach for a student who joins a few weeks into the term. It is a typical plan, not a promise, and we change it to match your syllabus.
| Week | Focus | By the end you can |
|---|---|---|
| 1 | Logic, quantifiers and negation; supremum and infimum | Negate any definition correctly and prove a number is a supremum |
| 2 | ε-N proofs for sequences, including limit laws | Write a complete convergence proof without notes |
| 3 | Monotone convergence, subsequences and Cauchy sequences | Prove a recursive sequence converges and find its limit |
| 4 | Series and the first ε-δ limits | Prove convergence tests and write a clean ε-δ proof using a min for δ |
Fit
Who this suits, and when another option is better
It suits students in a first or second real analysis course who are willing to attempt proofs between sessions and want feedback on how they write them, and students in an introduction to proofs course who want to be ready before analysis begins. Ajay holds an M.Tech and has 14+ years teaching maths, and he teaches real analysis online to students in India and abroad.
It does not suit anyone who wants proofs written for submission. Students who only need calculus methods, not proofs, will be better served by our calculus or engineering mathematics pages. Home tuition is available in Gurgaon (Gurugram). Students elsewhere in India and abroad learn with us online.
Academic integrity
We teach proof-writing; we never write your proofs for credit
Proof-based problem sheets are often graded, so this matters more in analysis than anywhere else. We explain definitions and theorems, work through parallel problems and give detailed feedback on proofs you have written yourself. We do not write graded proofs or assignments, take online tests or exams in your place, or write dissertations or other assessed work. If a question is from a graded sheet, Ajay teaches the technique on a different example and you write your own proof.
Study guides
Free guides for this course
Worked examples and exam advice from the Maths Bodhi team, topic by topic.
Related Ajay routes
Continue with the most relevant next page
University maths tutor online
All of Ajay's university courses on one page.
Linear algebra tutor online
Often the first course with proofs, alongside analysis.
US college calculus tutor
Calculus I to III before an analysis course.
Calculus 3 tutor online
Multivariable and vector calculus.
UK university maths tutor
Year 1 analysis and the move from A Level.
Differential equations tutor online
ODEs and an introduction to PDEs.
Book a free demo class
Bring one proof you are stuck on.
Ajay Vatsyayan
Profile and every subject he teaches.
FAQ
Questions students ask about real analysis tutoring
How do I start a real analysis proof?
Write down the definition you need to satisfy with your sequence or function substituted in, then do scratch work backwards to find N or δ. The proof is then written forwards.
Is real analysis harder than calculus?
It is harder in a different way. The ideas are familiar from calculus, but you must prove every step, and most students need a few weeks to get used to that.
Can you teach epsilon-delta proofs from the beginning?
Yes. We start with quantifiers and negation, then ε-N proofs for sequences, then ε-δ for functions.
Which textbook do you use?
Yours. Send your lecture notes or the book your course follows, and we use your lecturer's definitions.
I'm in an intro to proofs course now. Should I start before analysis?
Many students find it helps. A few sessions on logic, induction and writing short proofs make the first weeks of analysis much easier.
Do you teach metric spaces or a second analysis course?
Share the syllabus. Ajay teaches real analysis, and we will tell you honestly which topics in a later course he covers.
How many sessions a week do students take?
Usually one or two through the term, so there is time to attempt proofs between sessions. More are common just before exams.
Do you work with students outside India?
Yes. Sessions are online, and times are agreed by your time zone when you enquire.
Will you write my problem sheet proofs?
No. We teach the method on parallel examples and review proofs you have written yourself.
Next step
Send us one proof you are stuck on
Share your university, course name, lecture notes or textbook, your time zone and a proof you could not finish. Maths Bodhi will confirm whether Ajay is the right fit and suggest session times.
