Look for performance improvements

[robynstuart]

Do some initial profiling to identify slowest parts of runtime, then create issues to improve these

[RHull-IDM]

TLDR:

• Network is the slowest component,
• sti.set_outcomes() seems slow
• updating results seems slow

No single major bottlenecks apparent.

Running this code block gives the below profile.

hiv = sti.HIV(
    beta_m2f=0.05,
    beta_m2c=0.1,
    init_prev=0.05,
)
pregnancy = ss.Pregnancy(fertility_rate=10)
death = ss.Deaths(death_rate=10)
sexual = sti.StructuredSexual(recall_prior=True)
prior = sti.PriorPartners()
maternal = ss.MaternalNet()
testing = sti.HIVTest(test_prob_data=0.2, start=2000)
art = sti.ART(coverage_data=pd.DataFrame(index=np.arange(2000, 2021), data={'p_art': np.linspace(0, 0.9, 21)}))
vmmc = sti.VMMC(coverage_data=pd.DataFrame(index=np.arange(2000, 2021), data={'p_vmmc': np.linspace(0.025, 0.125, 21)}))
sim = ss.Sim(
    dt=1 / 12,
    start=1990,
    dur=40,
    n_agents=100000,
    diseases=hiv,
    networks=[sexual, prior, maternal],
    demographics=[pregnancy, death],
    interventions=[testing, art, vmmc]
)
sim.init()
sim.run(verbose=1/12)