how to ascribe two values to a parameter such that it will be time dependent
2 ビュー (過去 30 日間)
古いコメントを表示
In this my attached code i wanted to vary the value of gamma such that it will be dependent on the tspan in the driver.
when tspan is <=7 gamma should be equal to 0.001 as show in the attached code
but when tspan is >7 gamma should be equal to 1.
find attach the code and the driver used for runing the code.
Thanks
0 件のコメント
回答 (2 件)
Fabio Freschi
2021 年 9 月 2 日
You can pass an extra parameter to RiceG1 function using an anonmymous function. In your driver function add
% extra anonimous function to pass tspan
myfun = @(t,y)RiceG1(t,y,tspan);
% call to ode45
[B,V] = ode45(myfun,tspan,y0);
Then modify RiceG1 to accept the extra parameter and make the choice. I write below an extract of the function
function dy = RiceG1(~,y,tspan)
% ...
if tspan(2) <= 7
gamma = 0.001;
else
gamma = 1;
end
I have assumed that the check must be done on tspan(2). The modified files are attached.
0 件のコメント
Walter Roberson
2021 年 9 月 2 日
Unfortunately, Fabio's suggestion will not work. When you change the gamma during computation, you introduce a discontinuity in the derivative of the functions being integrated, and the mathematics of ode45 becomes invalid when you have discontinuities like that. You need to stop integrating and start again where you left off, but with the different gamma value.
You should, however, use the technique of passing in an additional parameter -- but it should be gamma directly, not tspan.
y0 = [1200 400 60 30 30 30 30 30 20 10 3000 30000];
tspan1 = [0 7];
tspan2 = [7 15];
gamma1 = 0.001;
gamma2 = 1;
funs = {@RiceG1, @RiceG2, @RiceG3, @RiceG4, @RiceG5};
for K = 1 : length(funs)
[t_out1, y_out1] = ode45(@(t,y)funs{K}(t,y,gamma1), tspan1, y0);
[t_out2, y_out2] = ode45(@(t,y)funs{K}(t,y,gamma2), tspan2, y_out1(end,:));
t_out{K} = [t_out1; t_out2];
y_out{K} = [y_out1; y_out2];
end
[B, C, D, E, F] = deal(t_out{:});
[V, W, X, Y, Z] = deal(y_out{:});
3 件のコメント
Bjorn Gustavsson
2021 年 9 月 2 日
@Fabio Freschi, in the matlab-release I use it is not enough to catch the event and continue integrating- that doesn't force the ode-solver to pass thrugh that point exactly. I was under the (selfmade) impression that the events would work like the waypoints work for the quadrature-functions, but they dont, so I have to stop and restart to get this done properly...
...which to me seems peculiar...
参考
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!