This is simple project about auctions in high-science fields.

There are many types of auctions in high-science fields, but this project is focused on AUSM (Adaptive User Selection Mechanism) auction type, which is used in NASA's Jet Propulsion Laboratory.

AUSM model (with queue): -- There are n demanders and m suppliers;

-- Suppliers supplies goods (one good from each supplier) with d dimensions each
(for instance, cluster with fixed amount of CPU, RAM, HDD and so on);

-- Quantities of dimensions may differs from one good to another;

-- Each supplier may impose basic costs on one unit of each dimension (minumum costs,
at which he is ready to sell his good);

-- Auction may run as long as it need;

-- There are potential allocation of goods among demanders after each bid;

-- Everybody knows all bids and potential allocations (queue, known to all);

-- Every demander may offer bid (if his previous bid was rejected).
A bid represends requirements (list of quantities of dimensions), id of supplier,
which part of good demander wants buy, and amount of money, which he is willing to pay.
More formally, bid is < d (vector of requirements), f (id of supplier), b (amount of money) >.
If there are enough "free space" in good and all lower costs are satisfied, bid is accepted.
Otherwise, a bid replaces a contract (or group of contracts) in potential allocation if and only if
b is higher than the sum of the bids, offered by those being replaced. If so, all replaced bids rejected.

In any case, this bid and new potential allocation pushed into queue;

AUSM model (with rounds): -- There are n demanders and m suppliers;

-- Suppliers supplies goods (one good from each supplier) with d dimensions each
(for instance, cluster with fixed amount of CPU, RAM, HDD and so on);

-- Quantities of dimensions may differs from one good to another;

-- Each supplier may impose basic costs on one unit of each dimension (minumum costs,
at which he is ready to sell his good);

-- Auction runs in serveral rounds;

-- Auction stops after round in which no changes in allocation was (or after maximum number of rounds exceeded);
Of course, there may be some minimum number of rounds, which helps demanders adjust their bids;

-- On each round demander may offer bid (in any case, was or not his previous bids accepted or rejected);
Bid may contain list of preffered suppliers (if not, whole list of suppliers used as preffered);

-- Everybody knows all bids and potential allocations from previous rounds.
So, some bids may be rejected, others are accepted;

-- After each round auction model computes optimal allocation;