Precision of RV fit

[eloubrown]

Hi, I am doing an RV fit for a binary and have followed the relevant example, which has worked nicely. Is it possible to output the orbital parameters from the mcmc fit with greater precision (ideally at least 4 decimals)? Thank you

[mirofedurco]

Hi,

results for the least squares method are stored in a dictionary format with unchanged number of decimal points as they arrived from the optimizer function. This dictionary is an output of the lst_sqr_task.fit(x0=rv_init_params) function in demo 11. For the displaying purposes, the number of decimals is reduced to some custom number for each parameter inside the result_summary() function.

In case of MCMC function the number of decimal points for each parameter is fully governed by the confidence interval for parameter determined by the MCMC sampling. If the precision of your fitted parameters is lets say +- 0.1, it really does not make a sense to list the resulting parameter with four decimal places. The precision with which the result is displayed will increase along with the decreasing confidence interval.

I hope this answer was helpful, Miro

[eloubrown]

Hi Miro,

Thank you that helps. A final question - is there a way that after saving the chains to file, I can reload the chains to the MCMC function?

Thank you Emma

[mirofedurco]

Hi Emma,

yes you can. I would like to refer you to the usage of the my_fit_id parameter in the jupyter tutorial demo11. This parameter enables you to store and then reload the chain for further analysis.

Sincerely, Miro

[mikecokina]

Closed for inactivity.