addinit(v=a)

[tvonbrasch]

If you first solve a model, then use addinit(v=a), and then solve the model again, why can the solver then terminate?

addinit_terminated.png (13.95 KiB) Viewed 672 times

I thought that the residuals would be set to the values they had in the first simulation when using the addinit(v=a) command. What could cause the solver to terminate when the addinit-command is used in this way?

Thomas

[random_user]

It seems that this is specific to the model you're using. Because I tried it on my end, and the model I was using solved successfully both times. Strange!

[tvonbrasch]

Hi,

Thanks for looking into this, random_user! I still don't understand the cause, but I think it is related to the actexist-option being activated.
Thomas

[EViews Matt]

Hello,

It's possible that if solution values (or expressions they appear in) are very close to numeric singularities, then the adjustment of add factors combined with the solution rounding and significant digits setting could produce such failures. Here's a small program that demonstrates this effect:

Code: Select all

create u 10
series x = 3e-8
series x_a = 3e-8
model m
m.append x = x(-1)
m.append y = @log(3.3e-7 - x)
m.addassign x
m.solve
m.addinit(v=a) @all
m.scenario "Scenario 1"
m.solve


[tvonbrasch]

Thanks, Matt, for looking into this!