def generate_jump_process(pp: Sequence[float], duration: int, jump_mean: float, rng: np.random.Generator, discount: float = 0.0):
jp = []
current_jump = 0
for k in range(duration):
jumps = [int(t) for t in pp if int(t) == k]
for jump in jumps:
current_jump += rng.exponential(jump_mean)
jp += [current_jump]
current_jump *= (1-discount)
return np.array(jp)
def generate_poisson_process(l: float, duration: float, rng: np.random.Generator) -> Sequence[float]:
"""
Generates a Poisson arrival process of rate `l` and duration `duration`.
"""
t = 0
ia_times = []
while t < duration:
ia_time = rng.exponential(1.0 / l)
t += ia_time
ia_times += [t]
return ia_times
def apply_block_time_variance(demand_process: Sequence[float], blocks: int, mean_ia_time: float = 13, rng: np.random.Generator = rng) -> Sequence[int]:
# Block time differences are distributed along a Poisson(13)
ia_times = rng.exponential(13, blocks)
demand_per_block = []
current_time = 0
for block_index, ia_time in enumerate(ia_times):
ia_time = int(ia_time)
new_demand = 0
for t in range(current_time, current_time + ia_time):
new_demand += demand_process[t]
demand_per_block += [int(new_demand)]
current_time += ia_time
return demand_per_block