Multi-backend annealing schedule for solving QUBO problems.

[GaboFGuerra]

User Story

As a user of the OptimizationSolver, I want to chose a linear or exponetial annealing schedule (user only provides decay parameter not time series) to control the exploration-explotation trade-off of the search algorithms across runtime.

Conditions of Satisfaction

• User can pass a time series to the solver which defines the temperature schedule.
• The annealing schedule affects the temperature parameter value on the NEBM neuron model during runtime at the timesteps specified by the time series.
• The annealing schedule affects the noise amplitude for the SCIF neuron model during runtime at the timesteps specified by the time series.
• The temperature schedule works for both Loihi 2 and CPU backends

Acceptance Tests

• Solver offers an API to receive input timeseries as annealing schedule.
• Temperature value of the NEBM neuron model follows the annealing schedule through time.
• Noise amplitude of the SCIF neuron model follows the annealing schedule through time.

Tasks

• Solve issue CPU OptimizationSolver with annealing schedule for solving QUBO problems  #173
• Solve issue Loihi 2 OptimizationSolver with annealing schedule for solving QUBO problems #174

Lava version:

• 0.3.0 (feature release)

I'm submitting a ...

• feature request

Other information:

insert the output from lava debug here