Counting small subsets #
The number of subsets of a finite set of size at most d is the corresponding truncated
sum of binomial coefficients. This identity holds for any threshold, including those above
the size of the ground set.
Weighting a subset t of s by z ^ (#s - #t) gives total weight (z + 1) ^ #s.
When z ≥ 1, every subset of size at most d has weight at least z ^ (#s - d).
This gives a small-subset bound with a freely chosen integer weight.
Choosing z = 2 and d = #s / 3 gives the exact integral inequality
M ^ 3 * 4 ^ #s ≤ 27 ^ #s, where M counts subsets of size at most #s / 3.
Equivalently, the exponential base is (27 / 4) ^ (1 / 3), strictly below 2.
The integral form needs neither real powers nor an asymptotic convention.
Main results #
Finset.card_filter_powerset_card_le_eq_sum_choose: count bounded-size subsets of a finset.Finset.card_filter_card_le_eq_sum_choose: the specialization to a finite type.Finset.card_filter_card_le_mul_pow_le: the weighted bound for any size threshold.Finset.card_filter_card_le_third_pow_three_mul_four_pow_le: the one-third bound.
Counting subsets of s by their cardinality gives a truncated binomial sum. Terms above
the cardinality of s vanish, so the threshold D need not be bounded by #s.
Counting subsets of a finite type by cardinality gives a truncated binomial sum, including thresholds above the size of the type.
The number of subsets of s of size at most d, multiplied by the minimum weight
z ^ (#s - d) of such a subset, is at most the total weight (z + 1) ^ #s.
The subsets of size at most one third of the ground set have exponential base
(27 / 4) ^ (1 / 3). This is an exact integer inequality, including the empty ground set.