Dataset Viewer
Auto-converted to Parquet Duplicate
doc_id
stringlengths
10
62
tags
listlengths
1
3
principle
stringlengths
30
206
method
stringlengths
33
363
answer
stringlengths
1
51
TheoremQA_elainewan/econ_micro_11.json
[ "derivative computation" ]
Profit is maximized where its derivative is zero: revenue minus total cost, differentiated with respect to quantity.
1. Write revenue as price times quantity. 2. Subtract the cost function to get profit. 3. Set the derivative of profit to zero and solve for quantity. 4. Substitute back for the maximum profit.
200
TheoremQA_elainewan/econ_micro_12_2.json
[ "simultaneous linear equations" ]
In long-run competitive equilibrium price equals average cost including the output-dependent royalty; combine with demand to solve.
1. Write price = average cost + royalty as a function of Q. 2. Write the demand equation. 3. Substitute and solve for Q. 4. Substitute back for P.
11
TheoremQA_elainewan/econ_micro_14.json
[ "area formulas" ]
Consumer surplus under competition is the triangle under the demand curve above the price equal to marginal cost.
1. Set price equal to marginal cost and find the quantity from demand. 2. Find the choke price where quantity is zero. 3. Compute the triangle area, one-half times base times height.
1152
TheoremQA_elainewan/econ_micro_14_3.json
[ "derivative computation" ]
Profit is maximized where the derivative of revenue minus cost with respect to output is zero.
1. Invert demand to get price in terms of quantity 2. Write profit as price times quantity minus cost 3. Differentiate and set equal to zero 4. Solve for the quantity
5
TheoremQA_elainewan/econ_micro_18.json
[ "weighted average" ]
Expected utility is the probability-weighted average of payoffs minus the fixed disutility of effort.
1. Compute expected revenue as the probability-weighted average of outcomes. 2. Apply the agent's share to get expected payment. 3. Subtract the effort cost.
$250
TheoremQA_elainewan/econ_micro_3.json
[ "convexity arguments", "derivative computation" ]
An indifference curve is convex when the marginal rate of substitution diminishes along it, checked by the sign of its derivative.
1. Compute the marginal rate of substitution from partial derivatives 2. Differentiate it along the indifference curve 3. Check the sign to see if the MRS is diminishing 4. Conclude convexity
True
TheoremQA_elainewan/econ_micro_4.json
[ "derivative computation" ]
Maximize utility by substituting the budget constraint and setting the derivative to zero.
1. Write the budget constraint. 2. Solve for one good and substitute into the utility. 3. Differentiate and set to zero. 4. Solve for the quantity.
5
TheoremQA_elainewan/econ_micro_7.json
[ "logarithm laws" ]
A risk-averse log-utility bettor bets only if expected utility p ln(W + B) + (1 - p) ln(W - B) is at least ln W; solve for the break-even p.
1. Write expected utility of betting with win and loss wealth 2. Set it equal to utility of current wealth 3. Solve for p as a ratio of logarithms
0.526
TheoremQA_elainewan/econ_micro_7_2.json
[ "weighted average", "logarithm laws" ]
Expected utility is the probability-weighted average of the utility in each outcome.
1. List each outcome's spending and probability 2. Compute utility ln(Y) for each 3. Multiply each by its probability and sum
9.1839
TheoremQA_elainewan/math_abstact_algebra_2.json
[ "modular arithmetic" ]
A check-digit code is valid exactly when the weighted digit sum is congruent to zero modulo the given modulus.
1. List the digits with their weights 2. Compute the weighted sum 3. Reduce modulo the modulus 4. Valid if the residue is zero
True
TheoremQA_elainewan/math_abstact_algebra_7.json
[ "group order theorems" ]
The cyclic group of integers mod n has exactly n elements.
1. List the residues 0 to n - 1 2. Count them
18
TheoremQA_elainewan/math_abstact_algebra_7_3.json
[ "group order theorems" ]
The order of a direct product of groups is the product of the orders.
1. Find the order of each factor group 2. Multiply the orders
12
TheoremQA_elainewan/math_abstact_algebra_7_4.json
[ "group order theorems" ]
A set with an operation is a group if it has closure, associativity, an identity and inverses.
1. Check closure. 2. Verify associativity by expanding both groupings. 3. Solve a * e = a for the identity. 4. Solve a * b = e for the inverse and check it lies in the set.
True
TheoremQA_elainewan/math_abstact_algebra_7_5.json
[ "modular arithmetic", "fermat's little theorem" ]
The order of an element in a group of units mod n is the smallest positive k with a^k congruent to 1 mod n.
1. List the units modulo n 2. Compute successive powers of the element modulo n 3. Stop at the first power equal to 1
2
TheoremQA_elainewan/math_abstact_algebra_7_6.json
[ "group order theorems" ]
A generator of a cyclic group is an element whose multiples produce every element; in the integers under addition only the units do so.
1. Recall that g generates the integers iff every integer is a multiple of g 2. Note this forces g to divide 1 3. List the units
{1, -1}
TheoremQA_elainewan/math_abstact_algebra_7_7.json
[ "modular arithmetic" ]
The kernel of a homomorphism on a group of units mod n is the set of elements mapping to the identity, found by evaluating the map modulo n.
1. List the elements of the group of units 2. Apply the map to each element and reduce modulo n 3. Collect the elements that map to 1
[1, 4]
TheoremQA_elainewan/math_abstact_algebra_7_8.json
[ "group order theorems" ]
Groups of the same order are isomorphic only if they share structural invariants such as being abelian.
1. Check whether each group is abelian 2. Compare other invariants such as element orders 3. Conclude non-isomorphic if any invariant differs
False
TheoremQA_elainewan/math_algebra_1.json
[ "simultaneous linear equations" ]
Solve one equation for one variable and substitute into the other to reduce to a single linear equation.
1. Rearrange the first equation to express y in terms of x 2. Substitute into the second equation 3. Collect like terms and solve for x
0
TheoremQA_elainewan/math_algebra_1_2.json
[ "simultaneous linear equations" ]
Each point on a polynomial graph gives a linear equation in the coefficients.
1. Substitute each point into the polynomial. 2. Solve the linear system by elimination. 3. Evaluate the polynomial at the requested input.
9
TheoremQA_elainewan/math_algebra_2.json
[ "eigenvalue analysis" ]
A square matrix is invertible exactly when its determinant is nonzero.
1. Compute the determinant. 2. Check whether it is zero. 3. Conclude invertibility.
True
TheoremQA_elainewan/math_algebra_3.json
[ "vector subspace test" ]
A set is a subspace only if it contains zero and is closed under addition and under every scalar multiple, including negative scalars.
1. Check the zero vector is in the set 2. Check sums of members stay in the set 3. Check scalar multiples, testing a negative scalar 4. Conclude not a subspace if any test fails
False
TheoremQA_elainewan/math_algebra_3_2.json
[ "simultaneous linear equations" ]
Vectors are linearly independent iff the homogeneous system (a combination of the vectors equal to zero) has only the trivial solution.
1. Set up the linear combination equal to the zero vector. 2. Write it as a homogeneous linear system. 3. Check for a dependency among the equations (or a zero determinant). 4. Conclude dependence if a nontrivial solution exists.
False
TheoremQA_elainewan/math_algebra_3_4.json
[ "simultaneous linear equations" ]
Vectors are dependent if the homogeneous system for their combination coefficients has a nontrivial solution.
1. Write a linear combination equal to the zero vector 2. Turn it into a homogeneous linear system 3. Eliminate variables to find a free variable 4. Exhibit a nontrivial solution or show none exists
False
TheoremQA_elainewan/math_algebra_3_5.json
[ "rank-nullity theorem" ]
AB = I_m forces rank(A) = m, and the n columns of A are independent only if n equals that rank.
1. Use rank of a product to bound rank(A) from below 2. Compare rank(A) with the number of columns 3. Decide independence from whether they match
Cannot be determined
TheoremQA_elainewan/math_algebra_4.json
[ "vector subspace test" ]
A set is a subspace only if it contains zero and is closed under addition and scalar multiplication; one counterexample disproves it.
1. Check whether the zero element is in the set 2. Test closure under addition with a concrete pair 3. Test closure under scalar multiplication 4. Conclude false if any test fails
False
TheoremQA_elainewan/math_algebra_4_2.json
[ "rank-nullity theorem" ]
The left shift on sequences is onto: any sequence has a preimage by prepending an arbitrary first term.
1. Write the target sequence. 2. Build a preimage by prepending any value and shifting the target right. 3. Conclude the target is in the image.
True
TheoremQA_elainewan/math_algebra_4_3.json
[ "rank-nullity theorem" ]
Left multiplication by a matrix is an isomorphism of the matrix space exactly when the matrix is invertible, i.e. has nonzero determinant.
1. Compute the determinant of the multiplying matrix. 2. If nonzero, the map is injective and hence bijective. 3. Conclude it is an isomorphism.
True
TheoremQA_elainewan/math_algebra_5.json
[ "vector subspace test" ]
With an orthogonal spanning set, the projection onto a subspace is the sum of projections onto each basis vector, (x.u/u.u)u.
1. Check that the spanning vectors are orthogonal 2. Compute x.u and u.u for each basis vector 3. Form each projection (x.u/u.u)u 4. Add the projections
[0, 8, 2, -2]
TheoremQA_elainewan/math_algebra_6.json
[ "vector subspace test" ]
A map is linear only if it preserves addition and scalar multiplication; one failing case disproves linearity.
1. State additivity F(A + B) = F(A) + F(B) and homogeneity F(cA) = cF(A) 2. Test additivity with general or concrete inputs 3. Test homogeneity 4. Conclude not linear if either fails
False
TheoremQA_elainewan/math_algebra_6_2.json
[ "derivative computation" ]
Represent a linear map by its matrix on a basis (images expressed in that basis) and take the matrix's determinant.
1. Apply the map to each basis function 2. Express each image in the basis to get the matrix columns 3. Compute the determinant of the matrix
4
TheoremQA_elainewan/math_algebra_6_3.json
[ "arithmetic simplification" ]
A three-by-three determinant is computed by cofactor expansion along the first row: a(ei - fh) - b(di - fg) + c(dh - eg).
1. Label the matrix entries a through i. 2. Compute the three 2x2 minors of the first row. 3. Combine them with alternating signs weighted by the first-row entries.
-36
TheoremQA_elainewan/math_algebra_7.json
[ "eigenvalue analysis" ]
A common eigenvector of two matrices is an eigenvector of their sum, with eigenvalue the sum of eigenvalues.
1. Write Av = lambda v and Bv = mu v 2. Add to get (A + B)v = (lambda + mu)v 3. Note v is nonzero
True
TheoremQA_elainewan/math_algebra_7_2.json
[ "eigenvalue analysis" ]
Every nonzero vector is an eigenvector of the identity, so any basis is an eigenbasis.
1. Note I v = v for every vector 2. Conclude every basis consists of eigenvectors
True
TheoremQA_elainewan/math_algebra_additional_1.json
[ "vector subspace test" ]
A finite set of polynomials has bounded degree, so it can only span polynomials up to that degree, never the whole space of polynomials.
1. Find the maximal degree among the given polynomials. 2. Note linear combinations cannot exceed this degree. 3. Conclude whether they can span the full polynomial space.
It depends on the degrees (solution's answer)
TheoremQA_elainewan/math_calculus_1.json
[ "solving quadratic equations" ]
A quadratic has a real root iff its discriminant is nonnegative.
1. Substitute the parameter value 2. Compute the discriminant 3. Real root exists iff it is at least zero
False
TheoremQA_elainewan/math_calculus_11.json
[ "dominant term limits" ]
Divide numerator and denominator by the highest power of n so lower-order terms vanish.
1. Divide numerator and denominator by n (inside the root by n^2). 2. Let the vanishing terms go to zero. 3. Simplify the remaining expression.
1
TheoremQA_elainewan/math_calculus_12.json
[ "derivative computation" ]
The maximum height of a parametric trajectory occurs where the derivative of the vertical coordinate is zero.
1. Differentiate the vertical component with respect to t 2. Set the derivative to zero and solve for t 3. Substitute t into the vertical component
2041 meters
TheoremQA_elainewan/math_calculus_14.json
[ "derivative computation" ]
A curve is a line only if its velocity vector keeps a constant direction.
1. Differentiate each component 2. Check whether r'(t) is a scalar multiple of a fixed vector for all t 3. Conclude it is not a line if the direction changes
False
TheoremQA_elainewan/math_calculus_15.json
[ "derivative computation" ]
Linear approximation: f(x) near a is f(a) plus the gradient at a dotted with the displacement.
1. Write the expression as a function of several variables. 2. Choose a nearby point with easy values. 3. Compute the partial derivatives there. 4. Add f(a) and the sum of partials times the small displacements.
33.44
TheoremQA_elainewan/math_calculus_16.json
[ "riemann sum approximation" ]
A double Riemann sum adds f at sample points times the subrectangle area.
1. Compute subinterval widths 2. List the lower-left sample points 3. Evaluate f at each 4. Multiply the sum by the cell area
2.8125
TheoremQA_elainewan/math_calculus_1_2.json
[ "trigonometric identities" ]
The half-angle formula cos(x/2) = sqrt((1 + cos x)/2) evaluates cosines of half of standard angles.
1. Write the angle as half of a standard angle. 2. Apply the half-angle cosine formula. 3. Simplify the nested radical and compare.
True
TheoremQA_elainewan/math_calculus_2.json
[ "surds and rationalization" ]
A 0/0 limit with a square root is resolved by multiplying by the conjugate and cancelling the common factor.
1. Check direct substitution gives 0/0. 2. Multiply numerator and denominator by the conjugate of the radical expression. 3. Cancel the common factor. 4. Substitute the limit point.
6
TheoremQA_elainewan/math_calculus_2_10.json
[ "dominant term limits" ]
A horizontal asymptote exists only if the function has a finite limit at infinity; find it by keeping the dominant terms.
1. Drop terms that vanish as x grows. 2. Take the limit of the dominant part. 3. If it diverges, there is no horizontal asymptote.
False
TheoremQA_elainewan/math_calculus_2_11.json
[ "intermediate value theorem" ]
A continuous difference that changes sign on an interval has a root there.
1. Define the difference of the two sides. 2. Evaluate it at two endpoints of an interval. 3. Observe a sign change. 4. Conclude a root exists.
True
TheoremQA_elainewan/math_calculus_2_2.json
[ "algebraic identities" ]
Factor the difference of cubes to combine the fractions; the limit exists only if the numerator also vanishes.
1. Rewrite both terms over the common denominator 2. Factor the difference of cubes 3. Require the numerator to vanish at the limit point 4. Solve for the constant
3
TheoremQA_elainewan/math_calculus_2_4.json
[ "bounding with inequalities" ]
A factor tending to zero times a bounded oscillating factor tends to zero (squeeze).
1. Bound the oscillating sine factor by 1 in absolute value 2. Bound the product by the vanishing factor 3. Apply the squeeze theorem
0
TheoremQA_elainewan/math_calculus_2_5.json
[ "bounding with inequalities" ]
A bounded factor times a factor tending to zero tends to zero.
1. Note that the oscillating factor stays between -1 and 1. 2. Note the other factor tends to zero. 3. Conclude by squeezing that the product tends to zero.
0
TheoremQA_elainewan/math_calculus_2_6.json
[ "trigonometric identities", "l'hopital's rule" ]
Rewrite csc - cot as (1 - cos x)/sin x and resolve the 0/0 form.
1. Express both terms over sin x 2. Recognize a 0/0 limit 3. Differentiate numerator and denominator 4. Evaluate the limit
0
TheoremQA_elainewan/math_calculus_2_7.json
[ "bounding with inequalities" ]
Since -|f| <= f <= |f|, if |f| tends to 0 then f is squeezed to 0.
1. Write -|f(x)| <= f(x) <= |f(x)| 2. Take limits of both bounds 3. Conclude by the squeeze theorem
0
TheoremQA_elainewan/math_calculus_3_3.json
[ "derivative computation" ]
Differentiate a reciprocal of a function with the chain rule: d/dx [1/u] = -u'/u^2.
1. Set u to the denominator 2. Differentiate u with the power rule 3. Apply -u'/u^2 4. Substitute the given point
-3/8
TheoremQA_elainewan/math_calculus_3_6.json
[ "derivative computation" ]
Acceleration is the second derivative of position with respect to time.
1. Differentiate position to get velocity 2. Differentiate velocity to get acceleration 3. Evaluate at the given time
-120
TheoremQA_elainewan/math_calculus_3_8.json
[ "derivative computation" ]
For motion on a circle, implicitly differentiate x^2 + y^2 = r^2 with respect to time to relate the coordinate rates.
1. Write the circle equation. 2. Differentiate both sides in t: 2x x' + 2y y' = 0. 3. Substitute the position and the known rate. 4. Solve for the unknown rate.
5/3
TheoremQA_elainewan/math_calculus_5.json
[ "derivative computation" ]
Evaluate a definite integral of a polynomial by the antiderivative at the bounds (fundamental theorem of calculus).
1. Integrate term by term with the power rule 2. Evaluate the antiderivative at the upper and lower limits 3. Subtract F(lower) from F(upper)
392/6
TheoremQA_elainewan/math_calculus_5_2.json
[ "derivative computation" ]
Integrate a separable first-order equation and fix the constant with the initial condition.
1. Integrate both sides with respect to the variable 2. Add a constant of integration 3. Use the initial condition to find the constant 4. Evaluate the solution at the requested point
17/3
TheoremQA_elainewan/math_calculus_7_3.json
[ "derivative computation" ]
The derivative of an inverse function at a point is the reciprocal of the original derivative at the corresponding input.
1. Differentiate f. 2. Use g'(f(x)) = 1/f'(x). 3. Find the input x with f(x) equal to the target value. 4. Evaluate the reciprocal.
1.137
TheoremQA_elainewan/math_real_analysis_additional_1.json
[ "recurrence relations" ]
If a recursively defined sequence converges, its limit is a fixed point L = g(L) of the recursion.
1. Replace x_n and x_{n+1} by L 2. Solve the resulting equation 3. Keep the root consistent with the sign of the terms
sqrt(2) ≈ 1.41
TheoremQA_elainewan/math_real_analysis_additional_2.json
[ "function substitution" ]
A rational function is continuous wherever its denominator is nonzero, so the limit equals the function value there.
1. Check the denominator is nonzero at the point. 2. Evaluate the function at the point. 3. Conclude the limits from both sides match the value.
True
TheoremQA_elainewan/math_real_analysis_additional_3.json
[ "picking numbers", "bounding with inequalities" ]
A universal inequality is refuted by a single counterexample with specific admissible values.
1. Pick admissible values of x and n 2. Evaluate both sides 3. Show the inequality fails
False
TheoremQA_elainewan/math_real_analysis_additional_5.json
[ "bounding with inequalities" ]
Uniform continuity follows from bounding |f(x) - f(y)| by a constant times |x - y| (a Lipschitz bound).
1. Write f(x) - f(y) over a common denominator. 2. Factor out x - y. 3. Bound the remaining factor by a constant. 4. Choose delta in terms of epsilon from that constant.
True
TheoremQA_jianyu_xu/Binomial_1.json
[ "casework", "combinations" ]
Split by the exact composition meeting the 'at least' conditions and multiply the selections within each case.
1. List the compositions meeting all the minimums 2. Count each with products of nCr 3. Add the cases
60
TheoremQA_jianyu_xu/Binomial_2.json
[ "multiset permutations" ]
Arrangements of a word with repeated letters equal n! divided by the factorial of each letter's multiplicity.
1. Count the total letters n 2. Count how many times each letter repeats 3. Compute n! divided by the product of the repeat factorials
60
TheoremQA_jianyu_xu/Binomial_3.json
[ "combinations", "multiplication principle" ]
Choosing a fixed number from each separate group is the product of the group-wise combinations.
1. Compute C(group size, needed) for the first group 2. Compute it for the second group 3. Multiply the counts
210
TheoremQA_jianyu_xu/Binomial_4.json
[ "combinations" ]
Choosing an unordered group of r from n is C(n, r) = n!/(r!(n - r)!).
1. Identify n and r 2. Compute C(n, r) as the falling product over r!
1140
TheoremQA_jianyu_xu/Binomial_5.json
[ "combinations", "multiplication principle" ]
Choose each subgroup independently with nCr and multiply the counts.
1. Compute C(group 1, needed 1) 2. Compute C(group 2, needed 2) 3. Multiply
840
TheoremQA_jianyu_xu/Binomial_6.json
[ "combinations" ]
Choosing an unordered subset of toppings is counted by nCr.
1. Identify n available and r chosen 2. Compute n! / (r!(n - r)!)
56
TheoremQA_jianyu_xu/Burnside_2.json
[ "burnside's lemma" ]
Count colourings up to rotation by averaging, over the rotation group, the number of colourings each rotation fixes.
1. List the cube's rotation classes and their sizes 2. For each, count face cycles and fixed colourings colours^cycles 3. Sum weighted counts and divide by the group order
57
TheoremQA_jianyu_xu/Catalan_1.json
[ "catalan numbers" ]
The number of balanced arrangements of n pairs of parentheses is the nth Catalan number C(2n, n)/(n + 1).
1. Identify n as the number of pairs. 2. Compute C(2n, n)/(n + 1).
132
TheoremQA_jianyu_xu/Catalan_2.json
[ "catalan numbers" ]
Monotone lattice paths from origin to (n, n) that do not cross the diagonal are counted by the nth Catalan number.
1. Recognize the non-crossing monotone path condition. 2. Compute C_n = (1/(n + 1)) * C(2n, n).
16796
TheoremQA_jianyu_xu/Cayley_3.json
[ "multiplication principle" ]
Cayley's formula: there are n^(n-2) labeled trees on n vertices (a Prufer sequence of n - 2 entries, each with n choices).
1. Identify the number of labeled vertices n. 2. Compute n^(n-2).
125
TheoremQA_jianyu_xu/Chinese_Remainder_Theorem_2.json
[ "modular arithmetic" ]
Solve a system of congruences with pairwise coprime moduli by the Chinese Remainder Theorem: x = sum of a_i M_i y_i mod M.
1. Compute M as the product of the moduli. 2. For each modulus, compute M_i = M/m_i and its inverse y_i mod m_i. 3. Form the sum of a_i M_i y_i and reduce mod M. 4. Verify the result.
8
TheoremQA_jianyu_xu/Graph_2.json
[ "multiplication principle" ]
Cayley's formula: there are n^(n-2) labeled trees on n vertices (a Prufer sequence of n - 2 entries, each with n choices).
1. Identify the number of labeled vertices n. 2. Compute n^(n-2).
1296
TheoremQA_jianyu_xu/Multinomial_1.json
[ "multinomial coefficients", "combinations" ]
Splitting people into labelled groups of given sizes is a product of successive combinations, n!/(a! b! c!).
1. Choose the first group from all people. 2. Choose the next group from those remaining. 3. The last group is what remains. 4. Multiply the counts.
4200
TheoremQA_jianyu_xu/Multinomial_2.json
[ "set partition counting", "combinations" ]
Choose the groups one after another with binomial coefficients, then divide by the factorial of the number of groups because identical groups are unordered.
1. Choose members for the first group with nCr 2. Choose the next group from the remaining people, and so on 3. Multiply these counts 4. Divide by k! for k indistinguishable groups
280
TheoremQA_jianyu_xu/Multinomial_3.json
[ "multiset permutations" ]
Arrangements of a multiset of digits are n! divided by the factorial of each digit's multiplicity.
1. Count the total digits and each repeated digit's frequency. 2. Compute n!/(n1! n2! ...).
420
TheoremQA_jianyu_xu/Multinomial_4.json
[ "multiset permutations" ]
Arrangements of n items with repeated groups equal n! divided by the factorial of each repetition count.
1. Count total items n 2. Count how many times each distinct item repeats 3. Compute n!/(n1! n2! ...)
105
TheoremQA_jianyu_xu/Multinomial_6.json
[ "casework", "combinations" ]
For 'at least k' of one kind, split into cases by the exact number of that kind and multiply selections in each case.
1. Split by the exact number of red balls chosen 2. Choose the red balls with nCr 3. Choose the rest from the non-red balls with nCr 4. Add the cases
81
TheoremQA_jianyu_xu/Ramsey_2.json
[ "ramsey numbers" ]
The smallest group guaranteeing a mutual-friend or mutual-stranger clique of given size is the corresponding Ramsey number.
1. Recognize the friend/stranger edge-colouring question 2. Recall the known small Ramsey number for that clique size
6
TheoremQA_jianyu_xu/Ramsey_3.json
[ "ramsey numbers" ]
The minimum group guaranteeing a clique of mutual friends or strangers of given size is the Ramsey number R(k, k).
1. Recognize the friends-or-strangers guarantee as a Ramsey number 2. Recall the known value of R(k, k) 3. Report it
18
TheoremQA_jianyu_xu/Ramsey_4.json
[ "ramsey numbers", "pigeonhole principle" ]
The smallest complete graph forcing a monochromatic clique of a given size under two-colouring is the Ramsey number R(k, k).
1. Recognize the question as a two-colour Ramsey question 2. Identify the clique sizes required in each colour 3. Quote the known Ramsey number R(k, k)
18
TheoremQA_jianyu_xu/Ramsey_5.json
[ "ramsey numbers" ]
The smallest complete graph forcing a red clique of one size or a blue clique of another is the Ramsey number for that pair.
1. Identify the two clique sizes. 2. Recall the known value of the corresponding Ramsey number.
25
TheoremQA_jianyu_xu/Ramsey_6.json
[ "ramsey numbers" ]
The smallest n forcing a red K_a or blue K_b in a two-colored complete graph is the Ramsey number R(a, b).
1. Identify the clique sizes required in each color. 2. Look up the known Ramsey number R(a, b).
(c) 18
TheoremQA_jianyu_xu/Stirling_number_first_kind_5.json
[ "stirling numbers", "casework" ]
Seating n people at k identical round tables is counted by the unsigned Stirling number of the first kind; by hand, split by table sizes and use (m - 1)! per table.
1. List the possible size splits between the tables. 2. For each split, choose the people with nCr and arrange each table in (m - 1)! ways. 3. Divide by 2 for equal-size splits since tables are identical. 4. Add the cases.
13068
TheoremQA_jianyu_xu/Stirling_number_first_kind_6.json
[ "stirling numbers", "permutations" ]
Seating n people around one unlabeled round table counts cyclic arrangements, (n - 1)!.
1. Recognize circular arrangements at a single table. 2. Fix one person to remove rotations. 3. Arrange the rest: (n - 1)!.
362880
TheoremQA_jianyu_xu/Stirling_number_second_kind_3.json
[ "stirling numbers", "set partition counting" ]
Partitions of n objects into k nonempty unlabeled subsets are counted by the Stirling number of the second kind S(n, k).
1. Recognize unlabeled nonempty groups 2. Compute S(n, k) by recurrence or inclusion-exclusion formula
9330
TheoremQA_jianyu_xu/Stirling_number_second_kind_5.json
[ "set partition counting", "casework", "combinations" ]
List the allowed group-size patterns, count each by choosing members, and halve when two groups have equal size since unlabeled groups would be double-counted.
1. List the group-size splits satisfying the minimum size 2. For each split choose one group's members with nCr 3. Divide by 2 for splits with equal-size unlabeled groups 4. Add the cases
25
TheoremQA_jianyu_xu/Stirling_number_second_kind_6.json
[ "set partition counting", "combinations" ]
Choose groups one after another with binomial coefficients, then divide by the factorial of the number of identical-size unlabeled groups.
1. Note the size constraint forces equal group sizes 2. Choose each group's members successively with C(n, k) 3. Divide by the number of orderings of the unlabeled groups
15
TheoremQA_jianyu_xu/combination_1.json
[ "inclusion-exclusion principle", "combinations", "complementary counting" ]
Count committees with no forbidden group by subtracting those containing each forbidden group and adding back those containing both.
1. Count all committees with nCr. 2. Count committees containing each forbidden group. 3. Count committees containing both forbidden groups. 4. Total minus each plus their overlap.
176
TheoremQA_jianyu_xu/combination_and_permutation_1.json
[ "stars and bars" ]
Distributing identical items into distinct non-empty bins: give one to each bin first, then count the rest with C(n + k - 1, k - 1).
1. Place one item in each bin 2. Count the remaining items 3. Apply C(remaining + bins - 1, bins - 1)
220
TheoremQA_jianyu_xu/derangement_1.json
[ "derangements", "inclusion-exclusion principle" ]
The number of permutations with no fixed point is D(n) = n! times the alternating sum of 1/k! for k from 0 to n, about n!/e.
1. Apply D(n) = n! * sum of (-1)^k / k! for k = 0..n 2. Evaluate the alternating sum 3. Multiply by n! and round to an integer
133496
TheoremQA_jianyu_xu/inclusion_and_exclusion_1.json
[ "inclusion-exclusion principle" ]
Count integers divisible by at least one of several numbers by adding single counts, subtracting pairwise, and adding back the triple, each as a floor.
1. Count multiples of each divisor with floor(N/d) 2. Subtract counts of multiples of each pairwise LCM 3. Add the count of multiples of the triple LCM
74
TheoremQA_jianyu_xu/integer_programming_2.json
[ "bounding with inequalities" ]
A linear objective over a polygonal feasible region attains its maximum at a vertex of the region.
1. Rewrite each constraint as a boundary line and shade the feasible side 2. Find the vertices of the feasible region by intersecting boundary lines 3. Evaluate the objective at every vertex 4. Pick the vertex giving the largest value
[2, 0]
TheoremQA_jianyu_xu/pigeonhole_1.json
[ "pigeonhole principle" ]
With k categories, k + 1 people force two into the same category.
1. Count the categories 2. Consider the worst case of all distinct 3. Add one more person
13
TheoremQA_jianyu_xu/pigeonhole_2.json
[ "pigeonhole principle" ]
Partition the region into cells that the given circle can cover; some cell holds at least ceil(points/cells).
1. Divide the square into equal small squares 2. Check the circle covers a small square (its diagonal equals the diameter) 3. Apply pigeonhole to points per cell
3
TheoremQA_jianyu_xu/pigeonhole_3.json
[ "pigeonhole principle" ]
With more objects than categories, some category holds at least ceil(objects/categories).
1. Identify objects and categories 2. Compute ceil(objects / categories)
4
TheoremQA_jianyu_xu/pigeonhole_4.json
[ "pigeonhole principle" ]
With more people than possible birthdays, some birthday is shared by at least the ceiling of people over days.
1. Identify pigeons (people) and holes (possible birthdays). 2. Divide pigeons by holes. 3. Round up to get the guaranteed minimum in one hole.
3
TheoremQA_maxku/basic-electronics-2-1.json
[ "proportional division" ]
In a series circuit the same current flows, I = V / sum of R, and each resistor's power is I^2 R, proportional to its resistance.
1. Add the series resistances 2. Current = source voltage / total resistance 3. Power for each resistor = I^2 R 4. Convert to the requested unit
[12, 8, 5]
TheoremQA_maxku/basic-electronics-6-3.json
[ "pythagorean theorem" ]
In an AC circuit, apparent power is real power divided by power factor, and real, reactive and apparent power form a right triangle.
1. Find apparent power as real power divided by the power factor. 2. Compute reactive power as sqrt(apparent^2 - real^2). 3. Report both in the requested units.
[0.8, 2]
TheoremQA_maxku/cv-cnn1.json
[ "arithmetic simplification" ]
In a convolutional layer each filter produces one output feature map, so the output depth equals the number of filters.
1. Infer the kernel size from input size - kernel + 1 = output size 2. Note the filter depth matches the input channels 3. Read the number of filters as the output depth
64
TheoremQA_maxku/cv-cnn4.json
[ "arithmetic simplification" ]
Convolution output width is (W - F + 2P)/S + 1.
1. Identify input width, filter size, padding and stride. 2. Substitute into (W - F + 2P)/S + 1.
15
TheoremQA_maxku/cv-colorsci1-rgb.json
[ "multiplication principle" ]
The number of colors is the product of the number of values each independent channel can take.
1. Find the number of values per channel. 2. Multiply across all channels.
16777216
End of preview. Expand in Data Studio

keyword-rag builds

Build outputs for rag-qa collections: a vocabulary of technique terms, and each document's tags from it with its abstracted form (principle, method, answer). A rag-qa load of the same dataset, limit, recipe and build settings applies them with no AI call; anyone can read them without rag-qa.

Entry Dataset Limit Recipe Documents Authored by
bright_aops-2000-worked_examples bright_aops 2000 worked_examples 2000 Jonathan Chow; vocabulary reviewed

manifest.json lists every entry with its build settings and the sha256 of its files.

Columns

vocabulary.parquet: position (the term's place in the list), term, definition (may be empty).

outputs.parquet: doc_id (the source dataset's own document id), tags (1 to 3 terms of the vocabulary), principle, method, answer (empty for a document with no worked solution).

import pandas as pd

base = "hf://datasets/jonathanmiro97/keyword-rag-builds/builds/bright_aops-2000-worked_examples"
outputs = pd.read_parquet(f"{base}/outputs.parquet")
vocabulary = pd.read_parquet(f"{base}/vocabulary.parquet")

Credit and licence

The documents these outputs describe are not copied here; they come from BRIGHT: Su et al. 2024, BRIGHT: A Realistic and Challenging Benchmark for Reasoning-Intensive Retrieval (xlangai/BRIGHT, CC BY 4.0), which collects AoPS, MATH, AQuA-RAT and TheoremQA problems (as collected in BRIGHT). The outputs are released under CC BY 4.0.

Downloads last month
70