Sequences and Series: Star problems
Ten optional challenges to stretch your reasoning. Work on paper, use hints when you need them, and check the answer or full solution when you are ready. You can skip these problems and continue the course.
- 1 of 3 stars: Stretch
- 2 of 3 stars: Challenge
- 3 of 3 stars: Deep challenge
Stars indicate difficulty within this set.
0 of 10 completed · 0 skipped
Progress saved in this browser.
Progress can't be saved in this browser, so your choices last for this visit only.
-
Problem 1 Two starting values
Difficulty: 1 of 3 stars, Stretch
A sequence satisfies , , and for every positive integer . Find all real for which every term is a positive integer and the entire sequence is strictly increasing.
Among those possibilities, the first terms sum to . Find , and determine whether this extra information forces the entire sequence to be arithmetic.
- Hint 1
Treat the odd-indexed and even-indexed terms separately.
- Hint 2
The gaps between consecutive terms alternate between two values. For the sum, pair with .
Answer
The admissible values are . The stated sum forces , giving , an arithmetic sequence.
Full solution
The recurrence gives and for every .
All terms are integers exactly when is an integer.
Strict increase requires the alternating gaps and to be positive.
Thus , giving the five reported integer values; positivity then follows from and strict increase.
For , a pair sums to
Therefore the first terms sum to
Equating this to gives .
Both alternating gaps are then , so the full sequence is arithmetic, with
For the other admissible values, the two gaps differ; two interleaved arithmetic subsequences need not themselves form one arithmetic sequence.
Answer
The admissible values are . The stated sum forces , giving , an arithmetic sequence.
Key idea
A recurrence that skips an index may describe interleaved sequences with different local behavior.
- Hint 1
-
Problem 2 Four integer terms
Difficulty: 1 of 3 stars, Stretch
Find every ordered four-term geometric sequence of positive integers whose sum is . The common ratio is not assumed to be an integer. Prove that your list is complete.
- Hint 1
If the ratio in lowest terms is , integrality of all four terms forces them to be for a positive integer .
- Hint 2
Factor the sum as . The inequality gives a small bound on .
Answer
The sequences are and .
Full solution
Reversing a sequence replaces its ratio by its reciprocal, so first consider ratios in lowest positive integer terms.
If the first term is , the fourth is .
Since are relatively prime, integrality forces to divide .
Thus the terms have the form with positive integer , and their sum is
Write .
From we get , hence
In particular , since .
Also divides , so only or remain.
For , gives sum , which cannot equal .
For with , the pairs are and .
They give sums and , respectively.
Only the second works, with .
This yields with ratio .
Reversing gives the other listed sequence, with ratio .
Both sum to , and the bound on proves that no unexamined ratios remain.
Answer
The sequences are and .
Key idea
Integer terms can have a noninteger geometric ratio; put that ratio in lowest terms before using divisibility.
- Hint 1
-
Problem 3 A missing exponent
Difficulty: 1 of 3 stars, Stretch
In the expansion of , the integer and the real number are unknown. The coefficient of is , and the coefficients of and are equal.
Find . Then prove exactly which coefficients in the expansion are largest, without listing all of them.
- Hint 1
Use the Binomial Theorem to write the first three coefficients in terms of and .
- Hint 2
For the largest-coefficient question, compare neighboring coefficients using their ratio.
Answer
and . The coefficients of and are the only largest coefficients; each is .
Full solution
The coefficient of is .
Equality of the next two coefficients gives
All canceled factors are positive because and , so .
Subtracting this from yields , hence and .
These values satisfy both original coefficient conditions.
Let , for
The Binomial Theorem gives for .
This ratio exceeds exactly when , or ; it equals at and is smaller than afterward.
Consequently
The two largest coefficients are therefore precisely , and
Answer
and . The coefficients of and are the only largest coefficients; each is .
Key idea
Ratios of neighboring terms can locate a maximum without expanding or computing the entire sequence.
- Hint 1
-
Problem 4 Each term reports an earlier average
Difficulty: 2 of 3 stars, Challenge
A sequence begins with . For every integer , its th term is three times the average of all previous terms:
Find an explicit formula for and for . Prove that your formulas hold for every positive integer index; do not assume a formula for a sum of squares.
- Hint 1
Write the defining relation at two consecutive indices and eliminate the earlier partial sum.
- Hint 2
Show that . Look for a product of two consecutive integers.
Answer
and .
Full solution
First
For , write and
Since , subtracting or substituting gives
Thus
The same ratio also holds for , because .
This ratio suggests : it gives the correct first term, and if it holds at , the recurrence gives
Induction proves the formula for every .
Positivity ensures all ratios used above are legitimate; it also follows directly from the original average rule.
Finally, apply the original relation at index :
Substituting the formula for yields
This obtains the sum from the defining rule itself, rather than importing a separate sum-of-squares identity.
Each next term is determined by earlier ones, so the formulas describe the unique sequence.
Answer
and .
Key idea
When a recurrence contains a partial sum, compare consecutive equations to remove the accumulated history.
- Hint 1
-
Problem 5 A repeating pattern of signs
Difficulty: 2 of 3 stars, Challenge
For a real number , consider the infinite series whose signs repeat in blocks of two plus signs followed by two minus signs:
Find all real for which the series converges and has sum . Also explain why substituting into your simplified sum formula does not give the value of the original series.
Builds on Infinite Geometric Series
- Hint 1
First determine convergence from the magnitudes of the terms. Within the convergence range, group four consecutive terms.
- Hint 2
The four-term block factors as , and the blocks have ratio . Keep the original convergence restriction after canceling factors.
Answer
or . The sum formula is valid only for ; at the original series diverges.
Full solution
If , the term magnitudes do not approach zero, so the series cannot converge.
If , its tails are bounded in absolute value by the corresponding geometric tails, , which approach zero.
Thus it converges, and grouping consecutive blocks preserves its sum.
Each block is .
Summing the geometric series of blocks gives , where cancellation is valid because .
Equating this to gives
Both roots and lie in the convergence range, so both are valid.
At , the ordinary partial sums cycle through and never approach a limit.
The simplified rational expression happens to equal there, but it was derived only for .
Even observing that complete four-term blocks sum to zero would not prove convergence: it examines only one subsequence of partial sums.
Answer
or . The sum formula is valid only for ; at the original series diverges.
Key idea
Cancellation can extend an algebraic expression to values where the infinite process it represents is still undefined.
- Hint 1
-
Problem 6 Three partial sums in arithmetic order
Difficulty: 2 of 3 stars, Challenge
A geometric sequence has first term and a nonzero real common ratio . Write for the sum of its first terms. Suppose , in that order, form an arithmetic sequence.
Find every possible . For each, determine whether the infinite series converges, and find its sum when it does.
For the convergent case, prove that the terms in positions contribute exactly half the total infinite sum.
- Hint 1
Use directly, without dividing by .
- Hint 2
Your equation for simplifies . Compare each consecutive block of three terms with the first term in that block.
Answer
or . Only the first gives convergence, with sum ; the selected subseries sums to .
Full solution
The arithmetic condition is , which simplifies to
Since , this is , giving the two stated ratios.
Conversely, either root satisfies the original partial-sum relation, so neither can be discarded merely because it is negative.
The positive root lies between and because
Its geometric series converges and sums to
The negative root is less than , so the term magnitudes grow and its series diverges.
For the convergent root, the equation gives .
Each block therefore equals .
Summing complete blocks is valid here: both the full series and the subseries have ratios of magnitude below , and any unfinished tail tends to zero.
Hence the total is twice , proving the half-total claim and giving the selected sum .
Answer
or . Only the first gives convergence, with sum ; the selected subseries sums to .
Key idea
A relation among a few partial sums can become an identity for every block of the infinite series.
- Hint 1
-
Problem 7 Local steps and a global total
Difficulty: 2 of 3 stars, Challenge
A finite sequence satisfies , , and for .
Determine every possible value of . Identify all sequences attaining the smallest or largest value, and prove that every value in your answer can actually occur.
- Hint 1
Compare with its distance in steps from each endpoint. Also determine its parity.
- Hint 2
Encode increases as U and decreases as D. Replacing adjacent DU by UD changes only one intermediate height. Starting from the alternating sequence, what does each such replacement do to the total?
Answer
The possible totals are exactly . The unique minimum has for even and for odd ; the unique maximum has .
Full solution
Each step changes parity, so has the same parity as .
Since the terms are nonnegative, the six odd-indexed terms are at least , giving .
Equality forces all odd-indexed terms to be and all even-indexed terms to be , and this alternating sequence is valid.
The sum is even because it contains six odd terms and five even terms.
From the initial endpoint, ; from the final endpoint,
Thus and
Equality forces every term to reach its bound, giving the unique sequence that rises for six steps and then falls for six.
It remains to construct every even total between these bounds.
Encode the alternating minimum by UDUDUDUDUDUD.
Whenever adjacent steps are DU, replace them by UD.
This raises just the intermediate height by , leaves all other heights unchanged, and preserves nonnegativity and the endpoints.
Each swap moves a U past a preceding D, reducing the finite number of such out-of-order pairs by one.
Therefore repeated swaps eventually produce UUUUUUDDDDDD, the maximum sequence.
The total starts at , increases by exactly at every swap, and ends at .
It must therefore pass through every listed value, proving attainability as well as necessity.
Answer
The possible totals are exactly . The unique minimum has for even and for odd ; the unique maximum has .
Key idea
Bounds and parity restrict possible totals; a controlled local move can prove that none of the remaining values are missing.
- Hint 1
-
Problem 8 When two infinite codes agree
Difficulty: 3 of 3 stars, Deep challenge
A binary code , with every equal to or , represents the real number .
Give a complete description of when two distinct binary codes represent the same number. Prove your description by examining their first differing position.
Consequently, determine exactly which numbers in have two such codes, and prove that no number has more than two. Explain the endpoint cases and .
Builds on Infinite Geometric Series
- Hint 1
If the first difference is at position , the contribution there has magnitude . What is the largest contribution all later positions can make?
- Hint 2
Equality in the geometric tail bound forces every later digit in one code to be and every later digit in the other to be .
Answer
Two codes agree in value exactly when, after a common prefix, one has followed entirely by s and the other has followed entirely by s. Exactly the numbers with integers and have two codes. The endpoints each have one.
Full solution
Every code defines a convergent sum: the contribution after position is between and the geometric tail , which tends to zero.
Suppose two distinct codes first differ at position , with and .
If their values agree, then
Equality forces at every later position: any smaller digit difference would create a positive deficit that no other position can exceed its bound to repair.
Thus has only s afterward and only s.
Conversely, these tails have exactly the required difference, so this pattern always gives equal values.
The code with the at position terminates, so its value is an interior fraction .
Every such fraction has a finite binary representation: repeatedly divide the integer by , recording each remainder or , then read the remainders in reverse order and divide the resulting finite binary integer by .
Replace its last by and all later digits by to obtain a second code.
A terminating code can match a distinct code only by changing its last : changing an earlier position would violate the forced all-zero tail, and changing a later position is impossible.
Hence it has exactly one partner, and the first-difference argument shows that any repeated value has this terminating code.
No value has three codes.
Finally, requires all digits , and requires all digits , so neither endpoint has a partner in this indexing convention.
Answer
Two codes agree in value exactly when, after a common prefix, one has followed entirely by s and the other has followed entirely by s. Exactly the numbers with integers and have two codes. The endpoints each have one.
Key idea
An equality case in an infinite tail bound can classify all ambiguities in a numerical representation.
- Hint 1
-
Problem 9 Every third coefficient
Difficulty: 3 of 3 stars, Deep challenge
Let be a positive integer. In the expansion of , add the coefficients of . Find a closed formula for this sum and prove it without expanding all the terms.
You may use complex numbers, but do not use trigonometric forms or De Moivre's theorem. Derive the algebraic identities you need.
- Hint 1
Try evaluating the polynomial at and at the two nonreal roots of .
- Hint 2
For , show that is when divides and otherwise. Then find .
Answer
The sum is .
Full solution
Let
Direct substitution gives , so and hence
Since , its powers repeat as .
It follows that equals if divides , and otherwise equals
Write
Then , which is exactly three times the desired coefficient sum.
This finite identity needs no convergence argument.
Now
Also , so
Similarly, , so
Therefore , and division by gives .
The calculation explains why these three evaluations retain exactly the powers divisible by while canceling the others.
Answer
The sum is .
Key idea
Evaluating at roots of a simple algebraic equation can isolate selected coefficients of a polynomial.
- Hint 1
-
Problem 10 Powers just below integers
Difficulty: 3 of 3 stars, Deep challenge
Let . For , let be the greatest integer not exceeding . Prove that every is odd, and find a recurrence expressing in terms of and , together with the required initial terms.
Let be the least integer greater than or equal to . Determine the exact value of , and justify convergence. No decimal approximations to the powers are allowed.
Builds on Infinite Geometric Series
- Hint 1
Pair with its conjugate , which satisfies . Apply the Binomial Theorem to .
- Hint 2
The conjugate sum is an even integer, while lies strictly between and . Also and .
Answer
, , and for ; every is odd. The infinite sum is .
Full solution
Put , so .
The Binomial Theorem shows that is an even integer: the terms containing odd powers of cancel, and every remaining term is twice an integer.
Since and , we have
Thus is odd and .
The strict inequalities also prove that no is an integer.
Both and satisfy
Multiplying by and adding the resulting identities gives
Substituting yields
Directly, and , so and .
Finally,
The desired sum is therefore geometric, with ratio strictly between and .
Its tail after terms is , which tends to zero, so convergence is justified.
Its value is
Answer
, , and for ; every is odd. The infinite sum is .
Key idea
Conjugate powers can separate an exact integer from a small positive error, revealing both a recurrence and an infinite sum.
- Hint 1