komanawa.simple_farm_model.speed_test#

created matt_dumont on: 8/30/24

Classes#

DairyModelWithSCScarcity

a version with an s-curve scarcity model requiring additional parameters s, a, b, c

Module Contents#

class DairyModelWithSCScarcity(all_months, istate, pg, ifeed, imoney, sup_feed_cost, product_price, opt_mode='optimised', monthly_input=True, s=None, a=None, b=None, c=None, peak_lact_cow_per_ha=default_peak_cow, ncore_opt=1, logging_level=logging.CRITICAL, cull_dry_step=None, cull_levels=None, dryoff_levels=None, allow_mitigation_delay=False)[source]#

Bases: SimpleDairyModel

a version with an s-curve scarcity model requiring additional parameters s, a, b, c

Parameters:

  • all_months – integer months, defines mon_len and time_len

  • istate – initial state number or np.ndarray shape (nsims,) defines number of simulations

  • pg – pasture growth kgDM/ha/day np.ndarray shape (mon_len,) or (mon_len, nsims)

  • ifeed – initial feed number float or np.ndarray shape (nsims,), MJ/ha

  • imoney – initial money number float or np.ndarray shape (nsims,) $/ha

  • sup_feed_cost – cost of supplementary feed $/MJ float or np.ndarray shape (nsims,) or (mon_len, nsims)

  • product_price – income price $/kg product float or np.ndarray shape (nsims,) or (mon_len, nsims)

  • opt_mode – one of:

  • ‘optimised’: use scipy.minimize_scalar to optimise the fraction of cows to cull dryoff at each decision point (cpu intensive, but most precise, low memory) cull_dry_step must be None

  • ‘step’: use a fixed step size for cull/dryoff decision (less cpu intensive, less precise, low memory) cull_dry_step must be set

  • ‘coarse’: pick the best of an a. priori. number of cull/dryoff decisions (low cpu, high memory, moderate precision) cull_dry_step must be None

Parameters:

  • monthly_input – if True, monthly input, if False, daily input (365 days per year)

  • s – scarcity scurve scale - the maximum value of the curve (if s=1, the maximum value is 1)

  • a – scarcity scurve steepness - smoothing parameter as a increases the curve becomes steeper and the inflection point moves to the right

  • b – scarcity scurve steepness about the inflection point

  • c – scarcity scurve inflection point (if a=1, c is the x value at which y=0.5)

  • peak_lact_cow_per_ha – peak lactating cows per ha float

  • ncore_opt – number of cores to use for optimization of the cull/dryoff decision if ncore_opt==1, do not multiprocess, if ncore_opt > 1, use ncore_opt cores, if ncore_opt =None, use all available cores

  • logging_level – logging level for the multiprocess component

  • cull_dry_step – if None, use the optimized cull/dryoff decision, if float, use the cull/dryoff decision with a step size of cull_dry_step

  • cull_levels – if opt_mode=’coarse’ the number of equally spaced cull steps to make, otherwise must be None

  • dryoff_levels – if opt_mode=’coarse’ the number of equally spaced dryoff steps to make, otherwise must be None

  • allow_mitigation_delay – if True, allow for a delay in the implementation of the cull/dryoff decision, if False, implement the cull/dryoff decision immediately. In practice this takes the “best” state and asks if delaying the application of the best state for a decision time step is more financially beneficial as compared to immediate implementation.

calc_feed_scarcity_cost(i_month, start_cum_ann_feed_import, new_feed, use_sup_feed_cost, nsims)[source]#

Calculate the feed scarcity cost based on the previous cumulative feed import and the new feed import

Parameters:

  • start_cum_ann_feed_import – previous cumulative feed import (one per sim)

  • new_feed – new feed import (ndays, nsims)

Returns:

feed scarcity cost ($ for the time period) (NOT $/MJ !) np.ndarray shape (ndays, nsims)

calc_marginal_cost_benefit(i_month, month, current_state, current_feed)[source]#

calculate the marginal cost and benefit of a potential state change

Parameters:

  • i_month – model step

  • month – integer month

  • current_state – the current states of all models

Returns:

calc_next_state_quant_1aday(month, current_state)[source]#

calculate the possible next state down (lower feed requirements) from the current state

Parameters:

current_state – the current state for all farms

Returns:

actions np.ndarray shape (nsims,) of the actions:

  • 0: no change

  • 1: cull cows

  • 2: dryoff cows

  • 3: go to once a day milking

get_annual_feed()[source]#

get the annual feed needed for each farm (does not include any feed inefficiencies)

Returns:

np.ndarray shape (nsims,)

import_feed_and_change_state(i_month, month, current_state, current_feed)[source]#

import feed and change state

Parameters:

  • i_month – month index

  • month – month

  • current_state – current state

  • current_feed – current feed

Returns:

plot_scurve(plt_dnz_fs=False, figsize=(10, 10))[source]#

plot the s-curve :param plt_dnz_fs: bool if true plot the dairy nz farm system boundaries. :return:

set_mean_pg(pg_dict, pg_raw=None, months=None)[source]#

This method can be used to set a bespoke mean pasture growth for each month to calculate marginal cost/benefit by default this is called on self.pg in the init method

one of pg_dict or pg_raw and months must be set

Parameters:

  • pg_dict – None or dict of pasture growth values for each month, keys are 1-12, values are float or np.ndarray shape (nsims,)

  • pg_raw – None or np.ndarray shape (mon_len, nsims) pasture growth values for each day

  • months – None or np.ndarray shape (nsims,) of month values for each day

Returns:

set_mean_product_price(prod_price_dict, prod_price_raw=None, months=None)[source]#

This method can be used to set a bespoke mean pasture growth for each month to calculate marginal cost/benefit by default this is called on self.prod_price in the init method

one of prod_price_dict or prod_price_raw and months must be set

Parameters:

  • prod_price_dict – None or dict of pasture growth values for each month, keys are 1-12, values are float or np.ndarray shape (nsims,)

  • prod_price_raw – None or np.ndarray shape (mon_len, nsims) pasture growth values for each day

  • months – None or np.ndarray shape (nsims,) of month values for each day

Returns:

set_mean_sup_cost(sup_dict, sup_cost_raw=None, months=None)[source]#

This method can be used to set a bespoke mean supplementary feed cost for each month to calculate marginal cost/benefit by default this is called on self.sup_feedout_cost in the init method

one of sup_dict or sup_cost_raw and months must be set

Parameters:

  • sup_dict – None or dict of supplementary feed cost values for each month, keys are 1-12, values are float or np.ndarray shape (nsims,)

  • sup_cost_raw – None or np.ndarray shape (mon_len, nsims) supplementary feed cost values for each day

  • months – None or np.ndarray shape (nsims,) of month values for each day

Returns:

supplemental_action_last(i_month, month, day, current_state, current_feed, current_money)[source]#

supplemental actions that might be needed for sub models at the end of each time step. This is a placeholder to allow rapid development without modifying the base class. it is absolutely reasonable to leave as is.

Here we use this function to save the cow buckets to output

Parameters:

  • i_month

  • month

  • day

  • current_state

  • current_feed

  • current_money

Returns:

On this page