Definitions, claims, and proof checks by unit
Each unit names the actual skills, demonstrates one complete case, and identifies the mistake most likely to change the result.
Real numbers
You can use the completeness property to guarantee a least upper bound exists. You can also handle absolute values and inequalities with real care.
Claims to justify
- Find the supremum of the set of all numbers below 3.
- Prove there is a rational number between any two reals.
- Use the triangle inequality on a sum of absolute values.
Worked example: Find the supremum and the maximum of the set of x with x < 3
- Every member sits below 3, and no smaller number bounds them all
- So the supremum is 3
- Supremum 3, but there is no maximum, since 3 is not in the set
Unsupported step: Assuming the supremum belongs to the set: claiming the maximum is 3, even though every single member is strictly below 3.
Sequences
You can prove a sequence converges straight from the epsilon-N definition. You can also argue convergence from monotone and bounded when no formula for the limit is handy.
Claims to justify
- Given an epsilon, produce an N that works for the sequence 1/n.
- Apply the monotone convergence theorem.
- Pull a convergent subsequence out of a bounded sequence.
Worked example: Show 1/n converges to 0. Take epsilon = 0.01.
- We need 1/n to be under 0.01
- That means n has to exceed 100
- Take N = 101, and every n at or beyond N works
Unsupported step: Choosing N before epsilon: a fixed N = 100 fails the moment epsilon drops to 0.001, because N has to depend on epsilon.
Series
You can decide whether an infinite sum converges and separate absolute from conditional convergence. You can also explain why reordering some series changes the total.
Claims to justify
- Apply the comparison test against a p-series.
- Use the alternating series test with the shrinking condition.
- Explain how a conditionally convergent series can be rearranged to any sum.
Worked example: Does the alternating sum of 1/n converge? Does it converge absolutely?
- Terms alternate and shrink to 0, so it converges to -ln 2
- But taking absolute values gives the harmonic series, which diverges
- Converges conditionally, not absolutely
Unsupported step: Treating conditional convergence as if it were absolute: reordering the terms freely can be made to produce any total you like, including infinity.
Continuity
You can prove continuity with epsilon and delta and use it to guarantee roots and extreme values. You can also tell uniform continuity from plain continuity.
Claims to justify
- Produce a delta for a given epsilon on a linear function.
- Apply the intermediate value theorem to trap a root.
- Show 1/x is continuous on (0, 1) but not uniformly continuous there.
Worked example: Show f(x) = 3x is continuous at x = 2, with epsilon = 0.06
- The gap in outputs is 3 times the gap in inputs
- So we need the input gap under 0.06/3
- delta = 0.02 works
Unsupported step: Setting delta equal to epsilon: delta = 0.06 lets x = 2.05 through, where the output gap is 0.15 and blows past the required 0.06.
Differentiation
You can prove differentiability from the limit definition and apply the mean value theorem. You can also exhibit a function that is continuous but has no derivative.
Claims to justify
- Use the mean value theorem to bound how much a function can change.
- Show the absolute value function has two different one-sided slopes at zero.
- Prove that differentiable implies continuous.
Worked example: For f(x) = x squared on [1, 3], find the mean value theorem point c
- Average slope = (9 - 1)/(3 - 1) = 4
- Set f'(c) = 2c = 4
- c = 2, which does lie inside (1, 3)
Unsupported step: Running the implication backwards: assuming continuity forces differentiability, though the absolute value function is continuous everywhere and has no derivative at 0.
Riemann integration
You can define the integral through upper and lower sums and say which functions are integrable. You can also link the integral back to the derivative.
Claims to justify
- Compute upper and lower sums for a simple partition.
- State the criterion that upper minus lower sums can be squeezed as small as you like.
- Apply the fundamental theorem to evaluate a definite integral.
Worked example: Lower sum for f(x) = x on [0, 1] with two equal slices
- Slices [0, 0.5] and [0.5, 1], with minimums 0 and 0.5
- 0 x 0.5 + 0.5 x 0.5
- Lower sum 0.25, while the true integral is 0.5
Unsupported step: Taking right endpoints for a lower sum: the values 0.5 and 1 give 0.75, which is the upper sum sitting above the true integral, not below it.
Metric spaces
You can measure distance in more general settings and talk about open sets, compactness and completeness there. You can also check whether a proposed distance rule is really a metric.
Claims to justify
- Verify the triangle inequality for a proposed distance rule.
- Show a set is open by fitting a small ball inside it around every point.
- Explain why the open interval (0, 1) is not complete.
Worked example: Is d(x, y) = (x - Y) squared a metric on the real line?
- Test the triangle inequality with 0, 1 and 2
- d(0, 2) = 4, but d(0, 1) + d(1, 2) = 1 + 1 = 2
- 4 is bigger than 2, so it is not a metric
Unsupported step: Checking only that distances are non-negative and vanish when the points match, then declaring a metric without ever testing the triangle inequality.
Read, attempt, challenge, and revise an argument
Short cycles create evidence that the idea is becoming usable rather than merely familiar.
Define
Write the definitions and assumptions that the claim can use.
Attempt
Build the argument without consulting a model proof.
Challenge
Search for a missing direction, edge case, or unsupported implication.
Revise
Repair the earliest logical gap and present the dependency clearly.
Trace the dependency of each claim
Definitions and earlier results supply the reasons used later, so repair the first unsupported step.
Questions about studying Real Analysis
Where should I begin in this course?
Start with the earliest unit whose definitions or notation you cannot explain from memory. Later units often depend on that foundation even when the current assignment looks different.
When should I move to the next unit?
Move on after you can choose the method on a changed problem, complete it without copying a model, and check the answer independently.
How should I use the practice workspace?
Begin without a hint, record the first uncertain decision, repair only that connection, and then answer a fresh question to test whether the repair transferred.