Jump to content


Photo

Pirana 23.1.2 RSE for Omega & Sigma


  • Please log in to reply
2 replies to this topic

#1 kengwah96

kengwah96

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 16 August 2023 - 04:32 AM

Dear all,

 

I noticed in the Pirana 23.1.2 release updates, there is an update to the formula used for the calculation of RSE for sqrt(exp(OM^2) - 1) transformed Omega and Sigma.

 

Previous formula: RSE% = %RSEuntransformed/2

 

New formula: RSE% = exp(ParameterValue)/ (2*(exp(ParameterValue) – 1)) *ParameterValue*(%RSEuntransformed).

 

However, I am unable to find additional background information on the switch. Would appreciate it if someone could shed some light on this.

 

Many thanks.

Keng Wah


Edited by kengwah96, 16 August 2023 - 04:33 AM.

  • Cyharlesbob, AlexKawn and Bryanruils like this

#2 kniefort

kniefort

    Member

  • Val_Members
  • PipPip
  • 29 posts

Posted 17 August 2023 - 01:04 AM

Hi Keng,

 

The new formula is simply the correct one for calculating RSE% for that particular transform.  The previous formula was an approximation that lost accuracy as omega values increased. 

 

The background would come from error propagation theory in statistics, i.e., see https://en.wikipedia..._of_uncertainty.

 

If you have further questions, please post and we can involve some of our mathematicians in the conversation!


  • kengwah96 likes this

#3 kengwah96

kengwah96

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 18 August 2023 - 12:38 AM

Many thanks for the explanation! 

Hi Keng,

 

The new formula is simply the correct one for calculating RSE% for that particular transform.  The previous formula was an approximation that lost accuracy as omega values increased. 

 

The background would come from error propagation theory in statistics, i.e., see https://en.wikipedia..._of_uncertainty.

 

If you have further questions, please post and we can involve some of our mathematicians in the conversation!






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users