# Dynare Sensitivity to Computing Steady State

Hi Dear All
I first computed the steady state of my model in the Maple software and then entered the extracted values in the initval block. the error I encountered is as follows:

Impossible to find the steady state. Either the model doesn’t have a steady state, there are an infinity of steady states, or the guess values are too far from the solution

Is this error due to dynare’s high sensitivity to the Steady State computation?

Although I entered the values in full. For example: C= 0.5254667235

is it better to enter the Steady state equations in dynare(Instead of computing in Maple and then entering the values in the initval block), or is there a problem with my calculations in Maple?

Of course, the method I used works on small models, but my model is medium

Have you checked the residuals using `resid`? An analytical steady state should go through easily.

This is what I did:
I defined my equations as a system of equations in Maple: and I was able to get the following answer:

But the answers obtained from the Maple do not lead to finding a steady state in Dynare. For example, if one of my equations is as follows:C=1-S-delta*K if I place the Maple output in this equation: The right side of the equation equals: 0.60 But the left is equal to: 0.60416 This causes the right and left sides of the equation not to be exactly equal and therefore residual not equal to zero.(Increasing and decreasing Digit (in Maple) does not solve the problem)
It may be said that it is better to define the steady state in the steady_state_model block in dynare (or in maple) and extract the steady state step by step and analytically. But this method is very difficult in my model . Because most of my variables are a function of each othe. that’s mean for example:
a=f(b,c,d,e,f,g,h,i,j,k)
b=f(a,c,f,h,i,j,n,m,c,x)
c=f(d,a,g,k,z,f,m,n,v,c)

What do you think is the best way to extract a steady state in my model and what solution do you suggest? Can’t Dynare solve the system of equations at a glance like Maple?

Meanwhile, one of my friends said: If the same answers obtained from Maple are defined as a parameter in the linear model, the model will be solved.for example: yhat=(c/y)*chat+(i/y)*ihat. then c/y and i/y are obtained by solving the system of equations in Maple .
In fact, this is one of the advantages of the linear to non-linear pattern. is it correct?

You need to find out why the Maple steady state does not solve the entered equations in Dynare. If you did not mistake in both programs, then the results must be identical except for numerical imprecision. The third digit after the decimal point does not really count as being accurate.
Going with the linearized model would in principle work, but given that your computed steady state in Maple may be incorrect (as it should solve the equations of the nonlinear model in Dynare), there is the risk to proceed with wrong values.