Skip to content

Latest commit

 

History

History
123 lines (90 loc) · 2.98 KB

README.md

File metadata and controls

123 lines (90 loc) · 2.98 KB

OED_Solver for C++

How to use

This program is header only. You can just use all of integral solver scheme by just including.

    #include "OED_Solver.hpp"

All of N-order OED return to N simultaneous OED equations,so this program only focus on soloving N simultaneous OED equations.

Let the independent variable be $t$,and let the N simultaneous OED equations like this below.

$$ \begin{eqnarray*} \frac{dy_1(t)}{dt} &=& f_1(t,y_1,y_2,\cdot\cdot\cdot,y_N) \\ \frac{dy_2(t)}{dt} &=& f_2(t,y_1,y_2,\cdot\cdot\cdot,y_N) \\ &\cdot& \\ &\cdot& \\ &\cdot& \\ \frac{dy_N(t)}{dt} &=& f_N(t,y_1,y_2,\cdot\cdot\cdot,y_N) \end{eqnarray*} $$

In this program, you can define N simultaneous OED equations like this below

    const int arg_num = N+1 //N is number of equations, and the number of independent variable is 1 so arg num is
    auto f1 = [](std::array<double, arg_num> args) { return ;};
    auto f2 = [](std::array<double, arg_num> args) { return ;};
    .
    .
    .
    auto fN = [](std::array<double, arg_num> args) { return ;};

Next, you have to set initial value of $y_1,y_2,\cdot\cdot\cdot,y_N$ at $t$

In this program, you can define initial value at $t$ like this below

    std::array<double, arg_num> init_values = {t,y1,y2, ... ,yN};

Let $dt&gt;0$ be step size for solving OED

Finally, you can solve OED like this

    double dt = 1e^-3;
    auto ans = OED_Sover::Euler<arg_num>({f1,f2,...,fN},init_values,dt);

Content of ans is like this

    ans = {t+dt,y1(t+dt),y2(t+dt),...,yN(t+dt)}

If you want to solve OED equations until $t = goal$,you can do like this

    std::array<double, arg_num> init_values = {t,y1,y2, ... ,yN}
    while(t < goal)
    {
        init_values = OED_Sover::Euler<arg_num>({f1,f2,...,fN},init_values,dt);
        t           = init_values[0];
    }

you can define array of {f1,f2,...,fN} like this below

    std::array<std::function<double(std::array<double, arg_num>)>, arg_num-1> equations = {f1,f2,...,fN}

Example

Lets solve this

$$ \begin{eqnarray*} \frac{dx}{dt} &=& x+2y \\ \frac{dy}{dt} &=& 2x+y \\ \end{eqnarray*} $$

In c++,

    const int arg_num = 3;

    auto func1 = [](std::array<double, arg_num> args) { return args[1] + 2*args[2];};
    auto func2 = [](std::array<double, arg_num> args) { return 2*args[1] + args[2];};

Let initial values be like this below

    std::array<double, arg_num> init_values = {0,1,2};

and, lets solve this problem in c++ !!

    while(t < 2)
    {
        init_values = OED_Solver::Euler<arg_num>({func1,func2},init_values,dt);
        t           = init_values[0];
    }

OED_Solver::Runge_Kutta_Fehlberg is the scheme that you can decide accuracy of answer you can use it like this below

    real error_ratio = 1e-5;
    real ans = OED_Solver::Runge_Kutta_Fehlberg<arg_num>({f1,f2,...,fN},init_values,dt,error_ratio);

You can define error_ratio of ans comparing exact ans. dt is going to be updated into another value.