use POM to estimate some parameters

[achubaty]

with #67

Use an objective function that minimizes difference between proportions of each species at t = 0 and t = 250:

min[Σ(p_i,t=250 - p_i,t=0)^2], where i = species

adjust following parameters:

• establishProbAdjFacResprout: [0,1]
• establishProbAdjFacNonResprout: [1,2]
• growthcurve decid: [0,1]
• growthcurve non-decid: [0,1]
• mortalityshape decid: [12,27]
• mortalityshape non-decid: [12,27]

Since we are adjusting 6 params total, and we need 10 'populations' per parameter, we need to run 60 'populations'. At ~4 GB per run on tolko_SK, that's ~240GB RAM.

[SteveCumming]

will be very interested to learn how this works out, and whether the new parameters give better results.

sc

On Mon, Jan 14, 2019 at 2:18 PM Alex Chubaty notifications@github.com wrote:

with #67 https://github.com/eliotmcintire/LandWeb/issues/67

Use an objective function that minimizes difference between proportions of each species at t = 0 and t = 250:

min[Σ(p_i,t=250 - p_i,t=0)^2], where i = species

adjust following parameters:

• establishProbAdjFacResprout: (0,1)
• establishProbAdjFacNonResprout: (1,2)
• growthcurve decid: [0,1]
• growthcurve non-decid: [0,1]
• mortalityshape decid: [12,27]
• mortalityshape non-decid: [12,27]

Since we are adjusting 6 params total, and we need 10 'populations' per parameter, we need to run 60 'populations'. At ~4 GB per run on tolko_SK, that's ~240GB RAM.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/eliotmcintire/LandWeb/issues/111, or mute the thread https://github.com/notifications/unsubscribe-auth/AG5_O4jPG93roMnz7ljADBKAhHRgo2Baks5vDNf4gaJpZM4Z_kL1 .

-- Steve Cumming Professeur agrégé Département des sciences du bois et de la forêt Université Laval (418) 656-2131 poste 2593

[achubaty]

closing. although POM potentially useful, these params are being better estimated from data by @ianmseddy which will be pulled into LandWeb eventually.