Is there any way to extract the name of the parameter from the ODE function

General Usage
1

Hi!
I want to extract the name of the parameter from the ODE function to make the tuple out of given parameters.
For example, we have an ODE function:

function f(du,u,p,t)
  a, b = p
  x = u
  du[1] = a * x+b * x^2
end

and I want to create a tuple for the parameters automatically according to the function, like:

para_tuple = (a = 1,b = 2)

How can I do that? (may use @ntuple function in DrWatson.jl.

Many thanks!

2

I don’t quite get the question. Try again with a bit more detail? You can use a NamedTuple for p though.

3

Hi Chirs!
Thanks so much for your time!
Usually, we define an ODEproblem with a vector para = [1.0,2.0] like:

function f!(du,u,p,t)
  a, b = p
  x = u
  du[1] = a * x+b * x^2
end
u0 = [0.0]
para = [1.0,2.0]
prob = ODEProblem(f,u0,(0.0,100.0),para)

My question is, how can I create a namedtuple like para_tuple = (a = 1.0,b = 2.0) after the definition of prob?

I know NamedTuple, but how can I use it for p? Sorry I tried several times inside functionf! or like NamedTuple(prob.p) but it doesn’t work.

Many thanks man.

4

Just pass it in to ODEProblem

5

Sorry… still unclear. Could you be more specific?
NamedTuple should be key-value pairs if I am not wrong, but there is no value inside the function f!

also, how can I output it in ODEProblem? I think there is no field in prob.f or prob.p.

6

prob = ODEProblem(f,u0,(0.0,100.0),para_tuple) and then you can use it as p.

7

Sorry for the unclear message.
of course I know I can use a tuple (or dict) like para_tuple = (a = 1,b = 2) when defining an ODEProblem.

My question is, can I create a parameter tuple by an existing ODEProblem? I want to do this because the BifurcationKit.jl only accept parameters in tuples but I want to do the bifurcation analysis automatically after plotting the trajectory.

Until now that package can’t support the direct use of ODEProblem, so I want to ask about a possible transition between vector to tuple for parameters.

8

I just don’t understand what you mean by this sentence. Can you show an example of the code you would like to work? Maybe that would explain the idea.

9

Okey…
What I have right now:

  • a defined ODE problem prob:
function f!(du,u,p,t)
  k1, k2, k3 = p
  x = u
  du[1] = k1 * x+ k2 * x^2 +k3
end
u0 = [0.0]
tspan = (0.0,100.0)
para = [1.0, 2.0, 1.0]   #parameters in vector form
prob = ODEProblem(f!,u0,tspan,para)

What I want then:

  • generate a tuple para_tuple from the prob which saves the keys as well as values of the parameter p :

para_tuple = (k1 = 1.0, k2 = 2.0, k3 = 1.0)

So that I can use it to define a BifurcationProblem (which requires the parameter tuple:

prob_bifur = BifurcationProblem(f!, u0, para_tuple, (@lens _.k1) )

10

If you just always use para_tuple, then your f code would still work and prob.p would be what you want. That seems like the most straightforward way?

11

Yes, you are right. But I am thinking of building a general function or module that can create parameter tuples for different models. I think I would try DrWatson.jl to find the possibility.

Still thank you Chris, for your time and your extraordinary work!