1

I have re-parameterize Arrhenius equation of form k = kref*exp(-E/R((1/T)-(1/Tref)) and i wanted to estimate the values of parameters E & kref which i got from lmfit package and correlations between it too. However the whole idea of re-parameterization was to see if we are getting low correlation between k0 and E after re-parameterization of original Arrhenius equation i.e. k = ko*exp(-E/RT) where kref = ko*exp(-E/RTref) so to do that i got following relation

                     Cov(ko,E)/k0  = Var(E)/RTref  -  Cov(Kref,E)/kref

So my question is that is there any way we can find Var(E) and also standard deviation of kref??

1 Answers1

0

I'm not certain I understand what you are asking - your notation is not clear. It is really always better to provide a minimal and complete example of working code.

The fit result from lmfit does include the covariance matrix (covar) between the variable parameters. Is that what you are looking for?

M Newville
  • 7,486
  • 2
  • 16
  • 29
  • Thanks M Newville for the quick response. Well I'm just looking for a way to print covariance matrix instead of correlation matrix in report_fit. –  Dec 21 '20 at 07:52
  • It might seem obvious, but if you have a question, then ask that question, and don't ask a different question. You did not ask "how do I print the covariance matrix?". If you had asked that, the answer would be easy: print the `covar` attribute of the `MinimizerResult` or `ModelResult` value returned by your fit. If you need more help than that then, really no kidding, provide a minimal and complete example. – M Newville Dec 21 '20 at 17:01