How to optimize parameters with dsolve?

I already have done example for optimizing parameters with 'ODE45', numerical solver. But, I think my system can be solved with dsolve. So, I want to make 'prob' wth 'dsolve'

I have five y_n(t). But I have only measured data, y_1(t) and y_2(t). I have model for my system which can be solved with 'dsolv'

here are systems,

syms LE(t) EX(t) RIP_1(t) RIP_3(t) T(t) 
A=dsolve(diff(LE) == -k_LE*LE - k_PET*LE + k_BET*EX,...
    diff(EX) == +k_PET*LE - k_EXE*EX - k_BET*EX - k_CS*EX + k_RC*RIP_1,...
    diff(RIP_1) == +k_CS*EX-k_RC*RIP_1-k_ISC_ST*RIP_1+k_ISC_TS*RIP_3,...
    diff(RIP_3) == +k_ISC_ST*RIP_1-k_ISC_TS*RIP_3-k_RC_T*RIP_3,...
    diff(T) == +k_RC_T*RIP_3+k_ISC*LE-k_Ph*T,...
    LE(0) == 100, EX(0) == 0, RIP_1(0)==0, RIP_3(0)==0, T(0)==0)

And I have measured two data set for LE and Ex with 't'

Target parameters are (k_LE k_PET k_BET k_EXE k_ISC k_CS k_RC k_ISC_ST k_ISC_TS k_RC_T k_Ph) with initial condition!

I am going to optimize 'ks' with measured data!

How can I do?

I have code with 'ODE45', but it does not work well.. :(

Answer 1

[John D'Errico]

You cannot do so.

The solution using dsolve has MANY parameters in it that have no assigned value.

As far as the use of dsolve here, I will say that after perhaps 30 minutes on a moderately fast computer, I gave up even trying to use dsolve to get the solution to that problem in a symbolic form anyway. And that means the solution would be hopelessly nasty.

There are many things in this world you may want to happen. But not all of them are achievable. I want to see world peace, but what are the odds? I'll still keep wanting that, but seriously...

[SUHYUN PARK]

Dear John D'Errico

Thank you for replying

I am so sorry about error of my question

Target parameters are ks (k_LE k_PET k_BET k_EXE k_ISC k_CS k_RC k_ISC_ST k_ISC_TS k_RC_T k_Ph) with initial condition!

With numerical ks, it is possible to sovle it.

In this case, is it impossible?