conditional equation for evidence

The front end.
Post Reply
bahman
Posts: 21
Joined: Sun Mar 13, 2022 12:24 pm

conditional equation for evidence

Post by bahman »

HI Marek;

in this image i need to C*=dt when there is no any evidence for node Z (i mean igonore node C* ) and if node Z have evidence z_1 then node C*=exp(1)
Image
iF
evidence in node Z is set to Z_1 state
then C*=EXP(1)
ELSE
C*=node dt (Identity Matrix conditioan on dt)
END

when there is no any evidence for node Z; it should be C*=dt and i think p( C*| z_1 , dt) is true but for p( C*| z_0 , dt) it should be a Identity Matrix in discretization table but is not
Image

i tied this equation for C*

Code: Select all

Node2=If(ZD="z_1",Exponential(1),If(ZD="z_0",dt,dt))
AND

Code: Select all

Node2=If(ZD="z_1",Exponential(1),dt)
both of them have same output .i need a equation that exactly do this only when i have seted an evidence for node Z
can you suggest another equation for this ?

thank you
marek [BayesFusion]
Site Admin
Posts: 430
Joined: Tue Dec 11, 2007 4:24 pm

Re: conditional equation for evidence

Post by marek [BayesFusion] »

Hi Bahman,

I hope I understand the question correctly. My understanding is that C has two parents, dt and Z. Also, there are possibly multiple outcomes of ZD, of which one is "z_1". In that case, your definition of C should be something like:

C = If(ZD="z_1",Exponential(1),dt)

If ZD takes the value z_1, C will have a value that is a sample from Exponential, otherwise the value of the variable dt. Please note that this may be not precisely what you wrote. It is impossible to refer in the definition to whether a node is observed or not. This would be part of the result. You should be able to achieve what you want by a smart definition.

I hope this helps.

Marek
Post Reply