【发布时间】:2023-03-14 20:50:01
【问题描述】:
我在 C 中实现了一个 Matlab 脚本的等效版本。为了运行 ode45,我选择了 GNU 科学库。但是 ode45 为每个版本生成不同的输出。我已经工作了一段时间,但我无法找到问题所在。 我使用 gsl_odeiv2_driver_apply_fixed_step 进行与 Matlab 中相同的步骤。
Matlab 脚本
function ExpoGrowthEqn
% code to solve the exponential growth equation
% dN/dt = r*N, N(0)=N0
% parameter values
r=0.2;
N0=10;
% numerical parameters
step=0.25;
tspan=[0:step:10];
% EXACT solution is: N(t)=N0 * exp(r*t)
for i=1:length(tspan)
N(i,1)=N0*exp(r*tspan(1,i));
end
plot(tspan,N,'ob') %plots EXACT solution
hold on
% solve ODE using ode45
options = odeset('RelTol',1e-12,'AbsTol',1e-12);
[t y] = ode45( @growth_eqn, tspan, N0, options, r);
plot(t,y,'-r') %plot APPROXIMATE solution
solutions=[t N y]
end
function dy = growth_eqn(t, y, r)
N=y(1)
dy=r*N;
end
C 代码
void fixed_step(void) {
PARAM parameters;
parameters.N = 11500000;
parameters.beta = 0.1;
parameters.gamma = 1/2;
double I0=500.0/parameters.N;
double R0=9000000.0/parameters.N;
double S0=1-I0-R0;
double y[3] = {S0, I0, R0};
gsl_odeiv2_system sys = {func, NULL, 3, ¶meters};
gsl_odeiv2_driver * d = gsl_odeiv2_driver_alloc_y_new (&sys, gsl_odeiv2_step_rkf45, 1e-12, 1e-12, 0);
printf ("=== Initial values ===\n");
printf ("y[0]=%6.15lf y[1]=%6.15lf y[2]=%6.15lf\n", y[0], y[1], y[2]);
int length = 53;
const double step = 0.25;
double ti; double t = 0.0, tant = t;
int i;
for (i = 0; i <= length*(1/step); i++)
{
printf ("%.2f(%3.d) -> t=%.2f %6.15lf %6.15lf %6.15lf\n", (float)tant, i, (float)t, y[0], y[1], y[2]);
tant = t;
int status = gsl_odeiv2_driver_apply_fixed_step (d, &t, step/8, 8, y);
if (status != GSL_SUCCESS)
{
printf ("error, return value=%d\n", status);
}
}
gsl_odeiv2_driver_free (d);
}
Matlab 输出
S I R
------------------------------------------------------------------
0.217347826086956 0.000043478260870 0.782608695652174
0.217347603416592 0.000038578485741 0.782613818097667
0.217347405784518 0.000034229664122 0.782618364551360
0.217347230418583 0.000030370797298 0.782622398784119
0.217347074884551 0.000026948321844 0.782625976793605
...
0.217345850226451 0.000000000007068 0.782654149766481
0.217345850226413 0.000000000006234 0.782654149767354
0.217345850226380 0.000000000005510 0.782654149768110
0.217345850226351 0.000000000004888 0.782654149768761
0.217345850226327 0.000000000004356 0.782654149769316
0.217345850226307 0.000000000003903 0.782654149769790
0.217345850226289 0.000000000003514 0.782654149770197
0.217345850226273 0.000000000003172 0.782654149770554
0.217345850226259 0.000000000002859 0.782654149770881
0.217345850226245 0.000000000002556 0.782654149771198
C 输出
=== Initial values ===
y[0]=0.217347826086956 y[1]=0.000043478260870 y[2]=0.782608695652174
0.00( ) -> t=0.00 0.217347826086956 0.000043478260870 0.782608695652174
0.00( 1) -> t=0.25 0.217347589196436 0.000043715151390 0.782608695652174
0.25( 2) -> t=0.50 0.217347351015482 0.000043953332344 0.782608695652174
0.50( 3) -> t=0.75 0.217347111537070 0.000044192810757 0.782608695652174
0.75( 4) -> t=1.00 0.217346870754133 0.000044433593694 0.782608695652174
...
51.00(205) -> t=51.25 0.217258883883215 0.000132420464611 0.782608695652174
51.25(206) -> t=51.50 0.217258162689554 0.000133141658272 0.782608695652174
51.50(207) -> t=51.75 0.217257437570519 0.000133866777307 0.782608695652174
51.75(208) -> t=52.00 0.217256708504771 0.000134595843055 0.782608695652174
52.00(209) -> t=52.25 0.217255975470856 0.000135328876970 0.782608695652174
52.25(210) -> t=52.50 0.217255238447202 0.000136065900623 0.782608695652174
52.50(211) -> t=52.75 0.217254497412123 0.000136806935703 0.782608695652174
52.75(212) -> t=53.00 0.217253752343811 0.000137552004014 0.782608695652174
在第二列的末尾,数据有很大不同。在 Matlab 中几乎为 0,但对于 C 代码则不然。
Matlab 时间戳
t =
0
0.250000000000000
0.500000000000000
0.750000000000000
1.000000000000000
1.250000000000000
1.500000000000000
1.750000000000000
...
52.250000000000000
52.500000000000000
52.750000000000000
53.000000000000000
【问题讨论】:
-
你能显示你的matlab数据的时间步长吗?!在我看来,您的 matlab 脚本一直运行到 t=10,而您的 C 代码显示 t=53?!
-
还有:这可能是容忍问题吗?其他答案(第 1 列和第 3 列)实际上在 matlab 和 C 之间几乎同样不同(绝对值)
-
看起来不像这样。时间戳是一样的。