Can you post your new code and your new results? Is the difference
consistent with the tolerance level?

Naively, I would expect the difference between the two implementations to
be around sqrt(N)*AbsTol -- this is what you'd expect if at each step each
implementation has an unbiased random error of size AbsTol.

Cheers,
Daniel.

On 14 June 2016 at 18:22, <[email protected]> wrote:

>
> Hi Daniel
> I did the test and the difference decreased. interesting, I had not
> noticed.
>
> Em domingo, 12 de junho de 2016 09:53:40 UTC-3, Daniel Carrera escreveu:
>>
>> I don't know what you are trying to say, but did you try running the two
>> programs with the same relative and absolute tolerances? Do the results
>> look more similar now?
>>
>> On 12 June 2016 at 02:01, <[email protected]> wrote:
>>
>>> Hi daniel
>>> in fact, there is a relationship with the tolerance between the methods.
>>>
>>> Em sábado, 11 de junho de 2016 11:51:53 UTC-3, Daniel Carrera escreveu:
>>>>
>>>> Probably because the Julia module has different default values for the
>>>> absolute and relative tolerance.  For Julia, the defaults are reltol=1e-5
>>>> and abstol=1e-8 but in Matlab the defaults are RelTol=1e-3 and AbsTol=1e-6.
>>>>
>>>> https://github.com/JuliaLang/ODE.jl
>>>> http://se.mathworks.com/help/matlab/ref/odeset.html
>>>>
>>>> That would also explain why the Julia program took so much longer. Try
>>>> again with the same tolerance and see if the results are more similar. Oh,
>>>> and like Erik said, you probably meant to have parenthesis around "1/8". If
>>>> the results still look difference, look for more options that might have
>>>> different default values.
>>>>
>>>> Cheers,
>>>> Daniel.
>>>>
>>>>
>>>>
>>>> On Saturday, 11 June 2016 04:29:47 UTC+2, [email protected]
>>>> wrote:
>>>>>
>>>>> this is the test for equation differential using runge-kutta45:
>>>>> f(x,y)= (-5*x - y/5)^1/8 + 10
>>>>>
>>>>>
>>>>> <https://lh3.googleusercontent.com/-D6U4d5pN0pw/V1t3MaV3JpI/AAAAAAAAACU/-E_ceVhrTxkT3SAgmLUy5nHDhRJTIikSgCLcB/s1600/rk45.png>
>>>>>
>>>>> why the numerical result is different? I used :
>>>>>
>>>>> function Rk_JL()
>>>>>  f(x,y)= (-5*x - y/5)^1/8 + 10
>>>>>  tspan = 0:0.001:n
>>>>>  y0 = [0.0, 1.0]
>>>>>  return ODE.ode45(f, y0,tspan);end
>>>>>
>>>>>
>>>>> and
>>>>>
>>>>>
>>>>> function [X1,Y1] = RK_M()
>>>>>  f = @(x,y) (-5*x - y/5)^1/8 + 10;
>>>>>  tspan = 0:0.001:n;
>>>>>  y0 = 1
>>>>>  [X1,Y1]= ode45(f,tspan,1);end
>>>>>
>>>>>
>>

Reply via email to