Configuration parameters¶
Biogeme can be configured using a parameter file. By default, the name is supposed to be biogeme.toml. If such a
file does not exist, Biogeme will create one containing the default values. The following table provides a description
of all parameters that can be configured.
Name |
Description |
Default |
Section |
Type |
|---|---|---|---|---|
largest_neighborhood |
int: size of the largest neighborhood considered by the Variable Neighborhood Search (VNS) algorithm. |
20 |
AssistedSpecification |
int |
maximum_attempts |
int: an attempts consists in selecting a solution in the Pareto set, and trying to improve it. The parameter imposes an upper bound on the total number of attempts, irrespectively if they are successful or not. |
100 |
AssistedSpecification |
int |
maximum_number_parameters |
int: maximum number of parameters allowed in a model. Each specification with a higher number is deemed invalid and not estimated. |
50 |
AssistedSpecification |
int |
number_of_neighbors |
int: maximum number of neighbors that are visited by the VNS algorithm. |
20 |
AssistedSpecification |
int |
bayesian_draws |
Number of draws per chain from the posterior distribution |
2000 |
Bayesian |
int |
calculate_likelihood |
Calculates likelihood-based statistics from the posterior draws |
True |
Bayesian |
bool |
calculate_loo |
Calculates the Leave-One-Out Cross-Validation (LOO) |
True |
Bayesian |
bool |
calculate_waic |
Calculates the Widely Applicable Information Criterion (WAIC) |
True |
Bayesian |
bool |
chains |
Number of independent Markov chains to run in parallel. |
4 |
Bayesian |
int |
mcmc_sampling_strategy |
Defines how MCMC sampling is performed: ‘automatic’ (selected based on hardware), ‘numpyro-parallel’ (one chain per device), ‘numpyro-vectorized’ (all chains on one device), ‘pymc’ (default PyMC sampler on CPU) |
automatic |
Bayesian |
str |
sample_from_prior |
bool: if “True”, samples from the prior distributions are generated. This may help in the diagnostic of indentification issues. |
True |
Bayesian |
bool |
target_accept |
Target acceptance probability for the No-U-Turn Sampler (NUTS) algorithm. Higher values like 0.9 or 0.95 often work better for problematic posteriors. |
0.9 |
Bayesian |
float |
warmup |
Number of warm-up / burn-in iterations per chain that are used only to adapt the sampler, not to estimate the posterior. |
2000 |
Bayesian |
int |
version |
Version of Biogeme that created the TOML file. Do not modify this value. |
3.3.2 |
Biogeme |
str |
number_of_jobs |
int: The maximum number of concurrently running jobs. If -1 is given, joblib tries to use all CPUs. |
2 |
Bootstrap |
int |
bootstrap_samples |
int: number of re-estimations for bootstrap sampling. |
100 |
Estimation |
int |
calculating_second_derivatives |
Defines how to calculate the second derivatives: analytical,finite_differences,never. |
analytical |
Estimation |
str |
large_data_set |
If the number of observations is larger than this value, the data set is deemed large, and the default estimation algorithm will not use second derivatives. |
100000 |
Estimation |
int |
max_number_parameters_to_report |
int: maximum number of parameters to report during the estimation. |
15 |
Estimation |
int |
maximum_number_catalog_expressions |
If the expression contains catalogs, the parameter sets an upper bound of the total number of possible combinations that can be estimated in the same loop. |
100 |
Estimation |
int |
optimization_algorithm |
str: optimization algorithm to be used for estimation. Valid values: [‘automatic’, ‘scipy’, ‘LS-newton’, ‘TR-newton’, ‘LS-BFGS’, ‘TR-BFGS’, ‘simple_bounds’, ‘simple_bounds_newton’, ‘simple_bounds_BFGS’] |
automatic |
Estimation |
str |
save_iterations |
bool: If True, the current iterate is saved after each iteration, in a file named |
True |
Estimation |
bool |
number_of_draws |
int: Number of draws for Monte-Carlo integration. |
10000 |
MonteCarlo |
int |
seed |
int: Seed used for the pseudo-random number generation. It is useful only when each run should generate the exact same result. If 0, a new seed is used at each run. |
0 |
MonteCarlo |
int |
number_of_threads |
int: Number of threads/processors to be used. If the parameter is 0, the number of available threads is calculated using cpu_count(). |
0 |
MultiThreading |
int |
generate_html |
bool: “True” if the HTML file with the results must be generated. |
True |
Output |
bool |
generate_netcdf |
bool: “True” if the netcdf file with the Bayesian estimation results must be generated. |
True |
Output |
bool |
generate_yaml |
bool: “True” if the yaml file with the results must be generated. |
True |
Output |
bool |
identification_threshold |
float: if the smallest eigenvalue of the second derivative matrix is lesser or equal to this parameter, the model is considered not identified. The corresponding eigenvector is then reported to identify the parameters involved in the issue. |
1e-05 |
Output |
float |
only_robust_stats |
bool: “True” if only the robust statistics need to be reported. If “False”, the statistics from the Rao-Cramer bound are also reported. |
True |
Output |
bool |
save_validation_results |
bool: “True” if the validation results are saved in CSV files. |
True |
Output |
bool |
enlarging_factor |
If an iteration is very successful, the radius of the trust region is multiplied by this factor |
10 |
SimpleBounds |
float |
infeasible_cg |
If True, the conjugate gradient algorithm may generate infeasible solutions until termination. The result will then be projected on the feasible domain. If False, the algorithm stops as soon as an infeasible iterate is generated |
False |
SimpleBounds |
bool |
initial_radius |
Initial radius of the trust region |
1 |
SimpleBounds |
float |
max_iterations |
int: maximum number of iterations |
1000 |
SimpleBounds |
int |
second_derivatives |
float: proportion (between 0 and 1) of iterations when the analytical Hessian is calculated |
1.0 |
SimpleBounds |
float |
steptol |
The algorithm stops when the relative change in x is below this threshold. Basically, if p significant digits of x are needed, steptol should be set to 1.0e-p. |
3.666852862501036e-11 |
SimpleBounds |
float |
tolerance |
float: the algorithm stops when this precision is reached |
6.055454452393343e-06 |
SimpleBounds |
float |
missing_data |
number: If one variable has this value, it is assumed that a data is missing and an exception will be triggered. |
99999 |
Specification |
int |
numerically_safe |
If true, Biogeme is doing its best to deal with numerical issues, such as division by a number close to zero, at the possible expense of speed. |
False |
Specification |
bool |
use_jit |
If True, the model is compiled using jit (just-in-time) to speed up the calculation. For complex models, compilation time may exceed the gain due to compilation, so that it is worth turning it off. |
True |
Specification |
bool |
dogleg |
bool: choice of the method to solve the trust region subproblem. True: dogleg. False: truncated conjugate gradient. |
True |
TrustRegion |
bool |
The structure of the biogeme.toml file is as follows.
1# Default parameter file for Biogeme 3.3.2
2# Automatically created on December 25, 2025. 21:51:51
3
4[MonteCarlo]
5number_of_draws = 10000 # int: Number of draws for Monte-Carlo integration.
6seed = 0 # int: Seed used for the pseudo-random number generation. It is useful
7 # only when each run should generate the exact same result. If 0, a new
8 # seed is used at each run.
9
10[TrustRegion]
11dogleg = "True" # bool: choice of the method to solve the trust region subproblem.
12 # True: dogleg. False: truncated conjugate gradient.
13
14[Specification]
15missing_data = 99999 # number: If one variable has this value, it is assumed that
16 # a data is missing and an exception will be triggered.
17numerically_safe = "False" # If true, Biogeme is doing its best to deal with
18 # numerical issues, such as division by a number close
19 # to zero, at the possible expense of speed.
20use_jit = "True" # If True, the model is compiled using jit (just-in-time) to speed
21 # up the calculation. For complex models, compilation time may
22 # exceed the gain due to compilation, so that it is worth
23 # turning it off.
24
25[Estimation]
26bootstrap_samples = 100 # int: number of re-estimations for bootstrap sampling.
27calculating_second_derivatives = "analytical" # Defines how to calculate the second
28 # derivatives:
29 # analytical,finite_differences,never.
30 #
31large_data_set = 100000 # If the number of observations is larger than this
32 # value, the data set is deemed large, and the default
33 # estimation algorithm will not use second derivatives.
34max_number_parameters_to_report = 15 # int: maximum number of parameters to
35 # report during the estimation.
36save_iterations = "True" # bool: If True, the current iterate is saved after each
37 # iteration, in a file named ``__[modelName].iter``,
38 # where ``[modelName]`` is the name given to the model.
39 # If such a file exists, the starting values for the
40 # estimation are replaced by the values saved in the
41 # file.
42maximum_number_catalog_expressions = 100 # If the expression contains catalogs,
43 # the parameter sets an upper bound of
44 # the total number of possible
45 # combinations that can be estimated in
46 # the same loop.
47optimization_algorithm = "automatic" # str: optimization algorithm to be used for
48 # estimation. Valid values: ['automatic',
49 # 'scipy', 'LS-newton', 'TR-newton',
50 # 'LS-BFGS', 'TR-BFGS', 'simple_bounds',
51 # 'simple_bounds_newton',
52 # 'simple_bounds_BFGS']
53
54[Bayesian]
55mcmc_sampling_strategy = "automatic" # Defines how MCMC sampling is performed:
56 # 'automatic' (selected based on hardware),
57 # 'numpyro-parallel' (one chain per device),
58 # 'numpyro-vectorized' (all chains on one
59 # device), 'pymc' (default PyMC sampler on
60 # CPU)
61sample_from_prior = "True" # bool: if "True", samples from the prior distributions
62 # are generated. This may help in the diagnostic of
63 # indentification issues.
64bayesian_draws = 2000 # Number of draws per chain from the posterior distribution
65warmup = 2000 # Number of warm-up / burn-in iterations per chain that are used
66 # only to adapt the sampler, not to estimate the posterior.
67chains = 4 # Number of independent Markov chains to run in parallel.
68target_accept = 0.9 # Target acceptance probability for the No-U-Turn Sampler
69 # (NUTS) algorithm. Higher values like 0.9 or 0.95 often work
70 # better for problematic posteriors.
71calculate_waic = "True" # Calculates the Widely Applicable Information Criterion
72 # (WAIC)
73calculate_loo = "True" # Calculates the Leave-One-Out Cross-Validation (LOO)
74calculate_likelihood = "True" # Calculates likelihood-based statistics from the
75 # posterior draws
76
77[Output]
78identification_threshold = 1e-05 # float: if the smallest eigenvalue of the
79 # second derivative matrix is lesser or equal to
80 # this parameter, the model is considered not
81 # identified. The corresponding eigenvector is
82 # then reported to identify the parameters
83 # involved in the issue.
84only_robust_stats = "True" # bool: "True" if only the robust statistics need to be
85 # reported. If "False", the statistics from the
86 # Rao-Cramer bound are also reported.
87generate_html = "True" # bool: "True" if the HTML file with the results must be
88 # generated.
89generate_yaml = "True" # bool: "True" if the yaml file with the results must be
90 # generated.
91generate_netcdf = "True" # bool: "True" if the netcdf file with the Bayesian
92 # estimation results must be generated.
93save_validation_results = "True" # bool: "True" if the validation results are saved
94 # in CSV files.
95
96[MultiThreading]
97number_of_threads = 0 # int: Number of threads/processors to be used. If the
98 # parameter is 0, the number of available threads is
99 # calculated using cpu_count().
100
101[Bootstrap]
102number_of_jobs = 2 # int: The maximum number of concurrently running jobs. If -1
103 # is given, joblib tries to use all CPUs.
104
105[Biogeme]
106version = "3.3.2" # Version of Biogeme that created the TOML file. Do not modify
107 # this value.
108
109[AssistedSpecification]
110maximum_number_parameters = 50 # int: maximum number of parameters allowed in a
111 # model. Each specification with a higher number
112 # is deemed invalid and not estimated.
113number_of_neighbors = 20 # int: maximum number of neighbors that are visited by
114 # the VNS algorithm.
115largest_neighborhood = 20 # int: size of the largest neighborhood considered by
116 # the Variable Neighborhood Search (VNS) algorithm.
117maximum_attempts = 100 # int: an attempts consists in selecting a solution in the
118 # Pareto set, and trying to improve it. The parameter
119 # imposes an upper bound on the total number of attempts,
120 # irrespectively if they are successful or not.
121
122[SimpleBounds]
123second_derivatives = 1.0 # float: proportion (between 0 and 1) of iterations when
124 # the analytical Hessian is calculated
125tolerance = 6.055454452393343e-06 # float: the algorithm stops when this
126 # precision is reached
127max_iterations = 1000 # int: maximum number of iterations
128infeasible_cg = "False" # If True, the conjugate gradient algorithm may generate
129 # infeasible solutions until termination. The result
130 # will then be projected on the feasible domain. If
131 # False, the algorithm stops as soon as an infeasible
132 # iterate is generated
133initial_radius = 1 # Initial radius of the trust region
134steptol = 3.666852862501036e-11 # The algorithm stops when the relative change in
135 # x is below this threshold. Basically, if p
136 # significant digits of x are needed, steptol
137 # should be set to 1.0e-p.
138enlarging_factor = 10 # If an iteration is very successful, the radius of the
139 # trust region is multiplied by this factor
140