By default, weights are set automatically (read more here). As a general rule, the weight coefficients should quantify how certain you are about the facts. Thus, depending on your data, you have the possibility to overwrite the automatically set weight coefficients. How exactly is specified in the mydata_my_pet
template.
We provide here guidance for setting weight coefficients when using the the minimization of the loss function as criterium:
 First estimate parameters with the default setting of weights, then, if necessary, reweigh particular data points using the previous estimate as starting values.
 Change weight coefficients by applying a factor to the existing weight coefficients, to ensure the correct (vector) size: e.g.
weights.tW = 5 * weights.tW
;
 Avoid application of large factors, since the loss function then hardly depends on other data, that serve the function of determining parameters.
 Weigh pseudodata less than real data in the zerovariate data by a factor 10 (or so for lossfunction sb).
 Weigh points that you don't trust with weight coefficient zero.
A typical reason to reweigh particular data points is to avoid unrealistic parameter estimates; values of the energy conductance larger than 1 cm/d, or maintenance ratios larger than 1, are likely to be unrealistic. Use related species as reference.
The default weight coefficients for the Pseudodata are listed in the table below:
Code

Value

Unit

Description

sb

su

v

0.02

cm/d

energy conductance

0.1

0.00001

kap

0.8



allocation fraction to soma

0.1

0.1

kap_R

0.95



reproduction efficiency

0.1

0.1

p_M

18

J/d.cm^3

volumespecific somatic maintenance costs

0.1

0.00001

k_J

0.002

1/d

maturity maintenance rate coefficient

0.1

0.00001

kap_G

0.8



growth efficiency

20

20

Growth efficiency kap_G
(dimensionless) controls the specific costs for structure [E_G]
(units J/cm^3), which very much depends on the watercontent. The watercontent of structure is specified by the chemical parameter d_V
(units g/cm^3) and set to a taxonspecific default value, that can be overwritten.