Prev: Filter2 in cell array
Next: Fourier series of the square root of a signal
From: ShilpaSY on 7 Jan 2010 03:10
can anybody PLEASE tel me hw ode adjusts its algorithm acc. to tspan
given to it..

Am getting steady state during integration within [0 1e+8] : tspan
as am increasing tspan upto 1e+13 am getting more error n my steady
state (i knw the steady state values for my species)

beyond [0 1e+14] of tspan error is in the range of 1e+2 or
so!!!!,,,, :
(
i was thinking ki, as i increase my tspan i shd approaching closer to
accuracy bt its nt so,,,
(or suggest me a method to decide tspan from our rate constants,,)
am really confused,,
pls explain me this or suggest me a reference,,

also i wanted to know hw ode algorithm choses its steps acc to absTol
and relTol
In d case whom uses as reference? (i tried to search alot bt i didnt
get)

i'd b very much thnkful to u for ur answer,,, :),,,
From: Steven Lord on 7 Jan 2010 10:07

"ShilpaSY" <shilpa.yadahalli(a)gmail.com> wrote in message
news:c9c06f54-4750-4ff4-a5c7-9d759256672b(a)j24g2000yqa.googlegroups.com...
> can anybody PLEASE tel me hw ode adjusts its algorithm acc. to tspan
> given to it..

Read the references given in the reference page for ODE45, "doc ode45".

http://www.mathworks.com/access/helpdesk/help/techdoc/ref/ode113.html?BB=1

> Am getting steady state during integration within [0 1e+8] : tspan
> as am increasing tspan upto 1e+13 am getting more error n my steady
> state (i knw the steady state values for my species)
>
> beyond [0 1e+14] of tspan error is in the range of 1e+2 or so!!!!,,,, :
> (
> i was thinking ki, as i increase my tspan i shd approaching closer to
> accuracy bt its nt so,,,
> (or suggest me a method to decide tspan from our rate constants,,)
> am really confused,,
> pls explain me this or suggest me a reference,,

I have no idea what you're asking here. Slow down and write out your
question again, as though you were explaining it to someone who has
absolutely no knowledge of what problem you're trying to solve (which is
actually what you're doing) and using NO abbreviations whatsoever. If you
have a _SMALL_ (no more than 100 lines) section of code that would be useful
in your explanation, post that too.

--
Steve Lord
slord(a)mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ


From: ShilpaSY on 7 Jan 2010 13:26
On Jan 7, 12:10 am, ShilpaSY <shilpa.yadaha...(a)gmail.com> wrote:
> can anybody PLEASE tel me hw ode adjusts its algorithm acc. to tspan
> given to it..
>
> Am getting steady state during integration within [0 1e+8] : tspan
> as am increasing tspan upto 1e+13 am getting more error n my steady
> state (i knw the steady state values for my species)
>
> beyond [0 1e+14] of tspan error is in the range of 1e+2 or
> so!!!!,,,, :
> (
> i was thinking ki, as i increase my tspan i shd approaching closer to
> accuracy bt its nt so,,,
> (or suggest me a method to decide tspan from our rate constants,,)
> am really confused,,
> pls explain me this or suggest me a reference,,
>
> also i wanted to know hw ode algorithm choses its steps acc to absTol
> and relTol
> In d case whom uses as reference? (i tried to search  alot bt i didnt
> get)
>
> i'd b very much thnkful to u for ur answer,,, :),,,

hello sir,
I'm actually using odexx from matlab for studying a biological system
(a signaling cascade including 30 reactions and 22 species- variables)
I'm giving required input for odexx.
My aim is to study the behavior of variables (i.e.species from my
system) ,, how they changes with time (for which ode is used
generally) and also to check when- at what time they achieve steady
state (by steady state i mean here is , after some period of time in a
biological system rate of change of variables wrt time is zero)
ode returns me solution array 'y' which is concentration of variables
in my code.
Now once steady state is achieved the concentration of variables
should not change further even after increasing tspan of ode (this is
what i expect)

i'm finding that tspan of [zero to 1e+8] is sufficient to reach my
system to steady state (i've reference steady state values from which
i'm comparing)
After this even after increasing tspan above 1e+8 there is significant
error in the steady state values
(am using y (end,:) end values of all variables as steady state
values)
now, my doubt is, why this is happening?

Are we allowed to increase tspan above 1e+14 i.e. tspan [0 to 1e+14]
(i mean, logically not ,but could you please elaborate more on this)

i wanted to know how to decide tspan for your system.

secondly ,
[when tspan is let's say [100 300] then number of steps ode takes be
as 'z'.
if i'm giving tspan in installments as [100 200] and [201 300 ] then
my number of steps are 'm' and 'n' resp.
and m+n> z
can u please explain me this? ]
From: Arthur Goldsipe on 7 Jan 2010 18:49
ShilpaSY <shilpa.yadahalli(a)gmail.com> wrote in message <57bc7c55-7d1d-4dc5-a359-8de9e789bfad(a)j24g2000yqa.googlegroups.com>...
> On Jan 7, 12:10 am, ShilpaSY <shilpa.yadaha...(a)gmail.com> wrote:
> > can anybody PLEASE tel me hw ode adjusts its algorithm acc. to tspan
> > given to it..
> >
> > Am getting steady state during integration within [0 1e+8] : tspan
> > as am increasing tspan upto 1e+13 am getting more error n my steady
> > state (i knw the steady state values for my species)
> >
> > beyond [0 1e+14] of tspan error is in the range of 1e+2 or
> > so!!!!,,,, :
> > (
> > i was thinking ki, as i increase my tspan i shd approaching closer to
> > accuracy bt its nt so,,,
> > (or suggest me a method to decide tspan from our rate constants,,)
> > am really confused,,
> > pls explain me this or suggest me a reference,,
> >
> > also i wanted to know hw ode algorithm choses its steps acc to absTol
> > and relTol
> > In d case whom uses as reference? (i tried to search  alot bt i didnt
> > get)
> >
> > i'd b very much thnkful to u for ur answer,,, :),,,
>
> hello sir,
> I'm actually using odexx from matlab for studying a biological system
> (a signaling cascade including 30 reactions and 22 species- variables)
> I'm giving required input for odexx.
> My aim is to study the behavior of variables (i.e.species from my
> system) ,, how they changes with time (for which ode is used
> generally) and also to check when- at what time they achieve steady
> state (by steady state i mean here is , after some period of time in a
> biological system rate of change of variables wrt time is zero)
> ode returns me solution array 'y' which is concentration of variables
> in my code.
> Now once steady state is achieved the concentration of variables
> should not change further even after increasing tspan of ode (this is
> what i expect)
>
> i'm finding that tspan of [zero to 1e+8] is sufficient to reach my
> system to steady state (i've reference steady state values from which
> i'm comparing)
> After this even after increasing tspan above 1e+8 there is significant
> error in the steady state values
> (am using y (end,:) end values of all variables as steady state
> values)
> now, my doubt is, why this is happening?
>
> Are we allowed to increase tspan above 1e+14 i.e. tspan [0 to 1e+14]
> (i mean, logically not ,but could you please elaborate more on this)
>
> i wanted to know how to decide tspan for your system.
>
> secondly ,
> [when tspan is let's say [100 300] then number of steps ode takes be
> as 'z'.
> if i'm giving tspan in installments as [100 200] and [201 300 ] then
> my number of steps are 'm' and 'n' resp.
> and m+n> z
> can u please explain me this? ]

Hi,

This step size is controlled not only by tolerances (AbsTol and RelTol) but also by the MaxStep option. These options are documented on the odeset function reference.

If you do not specify your own MaxStep, the default value is 0.1*(timeFinal - timeInitial).

In your case, as you increase timeFinal, your MaxStep keeps increasing as well. As long as the error at each time step is within tolerance, the solver will keep using this step size.

If you want to make sure that you get the same time steps when you solve in [0 10] as in [0 20], you should explicitly set MaxStep to an appropriate value (for example, 1).

I hope this helps clarify the behavior.
-Arthur
 | 
Pages: 1
Prev: Filter2 in cell array
Next: Fourier series of the square root of a signal