Skip to content

Latest commit

 

History

History
20 lines (17 loc) · 1.26 KB

dev.md

File metadata and controls

20 lines (17 loc) · 1.26 KB

Dev log

02/11/2024

  • Creating UserDefined functions for boundary and path costs without the data struct for now, to test the gradient evaluations
  • alternative to PerturbationSelectionVector : so idea is only one variable of the decision variable is perturbed at one time so like if you the diff wrt a single x_0 i.e. f_{x_0(i)} then you would choose i so have a vector = {i, j, k...} where i, j, k etc are the idices to choose the value to perturb for x_0, x_f, u_0... etc
  • dont forget link all your cpp files g++ -std=c++11 main.cpp gradientFD.cpp UserDefinedFunctions.cpp -o main

04/11/2024

  • Changed everything to double precision cause IPOPT uses double as default. Also noticed improvment in gradient approximation single precision ans - 0,3.99971,0,0,0,0,7.9999,0,0,0,0,0,7.9999,0 double precision ans - 0,4,0,0,0,0,8,0,0,0,0,0,8,0
  • Done with cost grad calc of J_E and J_L but need to write verification code

06/12/2024

  • Need to refactor the code base so theres less boilerplate to write down the line.
    • Introduce Problem, Jacobian and Hessian classes to not have pass in a bunch of variables
    • better way of perturbation
  • More robust type system, create custom type with operator overloading
  • Move to Cmake and add unit testing via Catch2