Grid Calculation of Loss Function Values

Stochastic simulations
1

Dear All,
I am trying to calculate loss function values for two different parameter values. I have a code on matlab (which is a bad modification of any other code I found in this forum) but I am not good at matlab. In loop3.m file, I don’t know how to write my loss function and store it and get the values (as a table or graph or any other way) in this loop.
My loss function will be like L=(1/2*variance.Y+variance.P+variance.b);
How can I get these loss function values for different parameters,
loop3.m (572 Bytes)
RRendiss.mod (3.1 KB)
Any help will be highly appreciated. Thanks

2

Try loop3.m (794 Bytes)
But you need to fix the parameter dependence in your mod-file first. Use model-local variables.

1 Like
3

Dear jpfeier,
Thanks a lot for reply.
I am trying to understand model-local variables now.
As I understood, these variables should be defined in the model part as
# VARIABLE_NAME = MODEL_EXPRESSION;
What should I do is still ambiguous to me. Which equations or variables should be used as model-local variables? And is it related to Steady State levels, should I change them too?
More importantly, am I finding wrong parameter values with osr command when I don’t use model local variables?
A little bit more information would be really helpful,
Thanks a lot.

4

It’s about the composite parameters set before the model-command :

RbSS=(BETAS*(1-gama*(1-RRRss))+BETAF*gama)/(BETAS*BETAF);
z1=J/(1-BETAS);                                     %qHstoCs
z2=J*RbSS/((1-BETAB)*RbSS-Kss*(1-BETAB*RbSS));     %qHbtoCb
z3=(J*Kss)/(RbSS*(1-BETAB)-Kss*(1-BETAB*RbSS));       %bSStoCbSS
z4=1-(1-RbSS)*(z3);                                %NbwbtoCb
z5=BETAS/(BETAS*gama*(1-RRRss)+BETAS*RbSS-BETAS-gama);       %bSStoCfSS
z6=((1-ALPHA)/(XSS*z4));                        %CbtoY
z7=(z3/z5)*z6;                                  %CftoY
z8=1-z6-z7;                                     %CstoY
z9=(z1*z8)/(z2*z6);                             %HsSS/HbSS
z90=z3*z6;                                     %btoY
z91=((1/BETAS)-1)*gama*z90+((ALPHA-1)/XSS)+1;  %CstoY
z92=z1*z91;                                   %qHstoY
z93=ALPHA/XSS;                                  %wsNstoY

See

1 Like