Getting started¶
After installation, the TAS server can be started with a default or manual configuration.
Currently, an instance of GPRTASServer is used which implements the methodology described in the reference.
Run server¶
Run ariane-tas-server.
ariane-tas-server [-h] [-c CONFIG_FILE] [-s RANDOM_SEED] [-p PORT] [--version]
Start ARIANE server
optional arguments:
-h, --help show this help message and exit
-c CONFIG_FILE, --config_file CONFIG_FILE path to configuration file
-s RANDOM_SEED, --random_seed RANDOM_SEED seed for pseudo-random numbers
-p PORT, --port PORT port number
--version show program's version number and exit
The following output should be visible:
(venv) user@host:path/to/folder$ ariane-tas-server
[server] Start GPRTASServer...
[server] GPRTASServer started.
[server] Wait for requests...
The server is now ready to receive requests from the client.
For every experiment that is started by the client, the server creates .json files under logs/ to store acquired experimental data.
In a typical workflow, three different experiment files are stored per experiment, each containing the same types of information:
*_0_init.json(after initialization)*_1_interm.json(during the experiment after each new measurement point to store most current data)*_2_final.json(after the experiment)
These files are used to diagnose an experiment (see Diagnostics).
Configuration¶
The GPRTASServer can be configured by a .cfg configuration file with a corresponding section ariane-tas-server-gpr.
For a list of available parameters, see Tab. 1 in the reference.
The following table shows all parameters and brief descriptions.
Any, except the first, parameter in the table is used for the server and thus can be set in the configuration file.
The first parameter, \(N_{\textnormal{row}}\), needs to be set in the client.
Parameter |
Notation |
Description |
|---|---|---|
– |
\(N_{\textnormal{row}}\) |
Number of rows in initialization grid, needs to be set in the client
(default value is |
|
\(\Delta^{\textnormal{rel}}_{\max}\) |
Parameter for estimation of background level, see Eq. (45) |
|
\(\Delta^{\textnormal{abs}}_{\min}\) |
Parameter for estimation of background level, see Eq. (45) |
|
\(l_{\max}\) |
Maximum decile in estimation for backgroud level, see Eq. (45) |
|
\(\beta\) |
Parameter for estimation of intensity threshold, see Eq. (47) |
|
\(N_{\mathcal{H}}\) |
Number of restarts of local kernel optimizations for a single
global kernel optimization with |
|
– |
Minimum value of kernel variance parameter |
|
– |
Maximum value of kernel variance parameter |
|
– |
Minimum value of kernel length scale parameters |
|
– |
Maximum value of kernel length scale parameters |
|
\(N_{\textnormal{KO}}^{\min}\) |
Minimum number of kernel optimizations |
|
\(N_{\textnormal{KO}}^{\max}\) |
Maximum number of kernel optimizations |
|
\(k_{\textnormal{KO}}\) |
Parameter for stopping criterion for kernel optimizations, see Eq. (57) |
|
\(\varepsilon_{\textnormal{KO}}\) |
Parameter for stopping criterion for kernel optimizations, see Eq. (57) |
|
\(\delta^{-}\) |
Parameter for detection of degenerate cases, see Eq. (64) |
|
\(\delta^{+}\) |
Parameter for detection of degenerate cases, see Eq. (64) |
By default, ARIANE uses the configuration file $VENV_ROOT/config/ariane-tas-server-gpr.cfg in the root directory of your Python environment.
The default parameter values set in this configuration file correspond to the values specified in Tab. 1 in the reference.
[ariane-tas-server-gpr]
level_backgr_diffs_rel_max = 0.5
level_backgr_diffs_abs_min = 15
level_backgr_decile_max = 6
thresh_intens_fact = 0.5
num_kernel_optim_restarts = 100
kernel_variance_bounds_min = 1e-3
kernel_variance_bounds_max = 1e2
kernel_length_scales_bounds_min = 1e-3
kernel_length_scales_bounds_max = 1e2
num_kernel_optims_min = 25
num_kernel_optims_max = 75
kernel_optims_stop_num_last = 9
kernel_optims_stop_eps = 0.025
length_scales_min_fact = 1e-3
length_scales_max_fact = 1.
