Standard mathematical economics studies the production, exchange, and consumption of goods “provided with units of measurement,” as in physics, in order to be enumerated, quantified, added, etc. Therefore, “baskets of goods,” which should describe subsets of goods, are mathematically represented as commodity vectors of a vector space, linear combination of units of goods, evaluated by prices, which are linear numerical functions. Therefore, in this sense, mathematical economics is a branch of physics.
However, economics, and many other domains of life sciences, investigate also what will be called entities, defining elements deprived of units of measure, which thus cannot be enumerated.
(1) Denoting by
(2) Entities can be “gathered” instead of being “added”;
(3) Entities can still be evaluated by a family of functions
(4) Subsets of entities can be evaluated by an “interval of values” between two extremal ones, the minimum and the maximum, instead of the sum of values of units of goods weighted by their quantities.
Life sciences dealing with intertwined relations among many combinations of entities, hypersets offer metaphors of “Lamarckian complexity” that keeps us away from binary relations, graphs of functions, and set-valued maps, to focus our attention on “multinary relations” between families of hypersets. Even deprived of units of measurement, these “proletarian” entities still enjoy enough properties for this pauperization to be mathematically consistent.
This is the object of this extremal manifesto: in economics and other domains of life sciences, vector spaces should yield their imperial status of “state space” to hypersets and linear prices to hypervaluato
A new Hartman–Wintner-type law of the iterated logarithm for independent random variables with mean-uncertainty under sublinear expectations is established by the martingale analogue of the Kolmogorov law of the iterated logarithm in classical probability theory.
We show that “full-bang” control is optimal in a problem which combines features of (i) sequential least-squares estimation with Bayesian updating, for a random quantity observed in a bath of white noise; (ii) bounded control of the rate at which observations are received, with a superquadratic cost per unit time; and (iii) “fast” discretionary stopping. We develop also the optimal filtering and stopping rules in this context.

