0.3.1 White Shark Plus

• SequentialShareFunction used
• Handling market failure signals from RPCMarket
• Preventing RiskyShare from going out of bounds
• Improved grid for RiskyShare estimation
• Faster interpolators
• Error handling

0.3 White Shark

This release updates SHARKFin to a state where is can reproduce the original pilot whitepaper results, using the AMMPS market ABM instead of the Paddrik NetLogo model.

Some significant changes include:

• A new AgentPopulation class that enables parameterization of populations by distributions over ex ante heterogeneous parameters, as well as interpolated solutions over continuously varying beliefs.
• An exogenous dividend process that is sampled by the Market class, passed through to the financial ABM, and factored into estimations of market rates of return
• Generic simulation execution scripts that take the kind of simulation to be executed as a command line argument

Issues addressed in this release:
https://github.com/sbenthall/SHARKFin/milestone/4?closed=1

0.2 Chum

Myriad improvements in pursuit of the Chum simulation, the CalibrationSimulation designed to test the price impact of the activity of a Broker on the AMMPS market after 30 days with different buy/sell values.

This involved the creation of the ClientRPCMarket, which communicates with AMMPS using RabbitMQ.
It involved a rewrite of the startup scripts to accommodate the new way of scaling up parallel simulations, which we are doing with containers.

There has also been a great deal of refactoring of the 0.1.1 code and the deprecation of the PNL wrappers.

There has also been a lot of preemptive work towards the 0.3 White Shark simulation.
For example, the Markets now return a dividend as well as a price.

0.1.1

patch release for Sloan Whitepaper