Regarding reviewing of proposal in physics.control

[ABHISHEK KUMAR]

Hi,

I'm Abhishek, and I'm currently a third-year student at NIT Delhi. I'm eager to participate in the upcoming GSoC 2024. Due to my examinations, I couldn't contribute for about three weeks. Now, I'm keen to work on the control system module of SymPy. I've included a link to my draft proposal below. While it's not perfect, I'm committed to refining it as much as possible.

I kindly request potential mentors Jason and Anurag to review my proposal. Any suggestions or feedback would be greatly appreciated. 

Link to Proposal

https://docs.google.com/document/d/1dmm7goYEyVcVkXrpwJm84yp6BmwfKrLB9rntZDA7R8g/edit?usp=sharing

Thank you.

Abhishek Kumar

Github: abhiphile

[ABHISHEK KUMAR]

--
You received this message because you are subscribed to the Google Groups "sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sympy+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/9553738b-f42f-4d9d-99c9-9031c3276c22n%40googlegroups.com.

[Jason Moore]

Hi Abhishek,

The proposal looks good and thorough. My only concern is that the proposed contributions are mostly focused on mimicking behavior from python control or matlab, which are purely numerical. We should focus on adding things to and improving sympy control that leverages symbolics. Please think about each of your proposed features and what they look like from a symbolic perspective and whether you are adding functionality that python control already has or actually bringing some interesting new symbolic features to a user's "control toolbox".

Jason

moorepants.info
+01 530-601-9791

To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/CAHZvv1b6zY-ysyv%3DOz4LOeCjnscg5%2BL7ZbPh1M64%2B4yU8wM_VQ%40mail.gmail.com.

[Anurag Surendra Bhat (B20CS097)]

Sorry I pressed post message by mistake. Continuing the reply here - 

Hi Abhishek,
I went over all the ideas and they look well researched to me. Personally, I would prefer if the idea - 

"Add a symbolic solver (and a numeric solver if required) with the help of the existing ODE module to solve x' = Ax + Bu form that to return the symbolic evaluation of the system for input u and known state x0 at time t0" is given greater priority over adding more features, let's say "Adding Laub's and Horner's Methods". We can work to improve what we already have rather than trying to make the Transfer Function class more feature rich.

Regards,

Anurag

On Wednesday, March 27, 2024 at 12:59:45 PM UTC+5:30 Anurag Surendra Bhat (B20CS097) wrote:

Hi Abhishek,
I went over all the ideas and they look well researched to me. Personally, I would prefer if the idea - 

"Add a symbolic solver (and a numeric solver if required) with the help of the existing ODE module to solve x' = Ax + Bu form that to return the symbolic evaluation of the system for input u and known state x0 at time t0" is given greater priority over adding more features, let's say 

[ABHISHEK KUMAR]

Hi Anurag, Jason

I think using ODE module isn't compulsory in for finding solution of the equation x' = Ax +  Bu , I've tried it by using sympy.solvers.ode.systems.linodesolve but it was giving weird result like this , I don't understand what is Dummy terms here.

I've another idea to find the solution of this differential equation by using Inverse Laplace Transform and by using it the process becomes more simpler. This approach will be more simple and I think inverse_laplace can be used in many places inside of control.

The final solution in this case for a Forced response with input will be

x(t) = inv_laplace([SI - A]**-1)X(0) + inv_laplace([SI-A]**-1 * B * U(S))

X(0) is the initial response it can be also denoted as X(t_0) here.

Here we just need to find the inverse of SI - A once and put it where ever needed.

To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/eeeb3932-3a81-4ad3-8b97-8cc01952a7f4n%40googlegroups.com.

[ABHISHEK KUMAR]

This method is also available in this lecture slide https://www.collegesidekick.com/study-docs/4719732

It also has examples for this method.

[Jason Moore]

The ode module should solve x' = A*x + B*u. If it doesn't then the best approach would be to fix it there instead of introducing duplicate solving code in control. Always try to use the existing functionality instead of reinventing the wheel.

Jason

moorepants.info
+01 530-601-9791

To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/CAHZvv1bVuyCd%2B0ifEXWWi-DSAOx_%3DMt%2BgxB8OyTPcRs6Gm3Ofg%40mail.gmail.com.

[Anurag Surendra Bhat (B20CS097)]

Hi Abhishek,

I agree with Jason here. You should try to find an elegant solution using the existing ODE module. 

I had researched if the ODE module would be able to solve given differential equation, while writing my proposal last year. So I would suggest you to look into it.

Regards,

Anurag Bhat.

To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/CAP7f1AiBQHL40gRT5Y5xC41PgbcKt21jntKH_HGihSgaGzxo9A%40mail.gmail.com.

[ABHISHEK KUMAR]

Hi Jason, Anurag,

I've added the calculation of the state vector and output vector by using the ODE module and a function to calculate the bandwidth of the TransferFunction using the sympy solve function.

https://docs.google.com/document/d/1dmm7goYEyVcVkXrpwJm84yp6BmwfKrLB9rntZDA7R8g/edit?usp=sharing

To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/CAOqN0PMTyC8V%3D%3Dzzh%2BfWX5_SNBmS_49RSCCeeAPHJ33-RWTASg%40mail.gmail.com.