Resource allocation

This repo accompanies the paper Allocation of Fungible Resources via a Fast, Scalable Price Discovery Method.

To get started with the code, clone this repo, run

python setup.py install

in a virtual environment of your choice, and try out the notebooks, which reproduce the examples from the paper.

Example

The resalloc package exports one main class representing a resource allocation problem, called AllocationProblem. It also exports a number of utility functions.

Here is a code example showing how to set up and solve a simple problem.

importtorchfromresalloc.fungibleimportAllocationProblem, utilitesn_jobs, n_resources=int(1e6), 4throughput_matrix=torch.rand((n_jobs, n_resources))
resource_limits=torch.rand(n_resoures) *n_jobs+1e3problem=AllocationProblem(
throughput_matrix=throughput_matrix,
resource_limits=resource_limits,
utility_function=utilities.Log()
)
problem.solve(verbose=True)
# X is the optimal allocationprint(problem.X)
# prices are the optimal pricesprint(problem.prices)

For more details about the available utilities, and how to customize the solve method with optional arguments, please consult the source code.