Re: Re: How can I "perturbate" a NDSolve
- To: mathgroup at smc.vnet.net
- Subject: [mg102066] Re: [mg102035] Re: [mg102000] How can I "perturbate" a NDSolve
- From: DrMajorBob <btreat1 at austin.rr.com>
- Date: Wed, 29 Jul 2009 05:04:52 -0400 (EDT)
- References: <200907260754.DAA18931@smc.vnet.net>
- Reply-to: drmajorbob at bigfoot.com
BTW, the perturbation (at that probability) makes virtually no difference
because, for instance, after running the code,
DownValues[a][[All, -1]]
Length@%
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
180
That is, a was called 180 times, and all the results were zero.
When I change the probability to 0.1, there are lots more a values and
about 10% are non-zero:
DownValues[a][[All, -1]];
Length@%
Length@DeleteCases[%%, 0]
2100
208
...but there's no visible difference in the plot! (Interpolations are
being used, and that smoothes everything.)
There's a critical point around p = 0.17 or 0.18, where NDSolve begins to
give error messages and the plot changes radically.
Bobby
On Tue, 28 Jul 2009 00:18:39 -0500, DrMajorBob <btreat1 at austin.rr.com>
wrote:
> You'll need something like
>
> Clear[a,x1,x2]
> a[t_?NumericQ]:=a[t]=If[RandomReal[]<0.001,RandomReal[],0]
> soln=NDSolve[{x1'[t]==(Sin[t]+a[t])*x2[t],x2'[t]==(Cos[t]-a[t]*x1[t]^2)*x1[t],x1[0]==1,x2[0]==0},{x1,x2},{t,0,10}];
> {x1,x2}={x1,x2}/.First@soln
>
> {InterpolatingFunction[{{0.,10.}},<>],InterpolatingFunction[{{0.,10.}},<>]}
>
> Plot[{x1@t,x2@t},{t,0,10}]
>
> NumericQ prevents symbolic evaluation of the equations, and
> a[t_?NumericQ]:=a[t]=... makes sure that a[t] is the same every time a
> specific t is used. That's what it takes to make a[t] truly a function
> of t.
>
> Bobby
>
> On Mon, 27 Jul 2009 04:55:20 -0500, Iván Lazaro <gaminster at gmail.com>
> wrote:
>
>> Thanks for the answers!
>>
>> I need a more specific perturbation. I need something like this (with
>> this I
>> expect to manipulate the frequency of the perturbation)
>>
>>
>> If[RandomReal[]<0.001,a=RandomReal[],a=0]
>>
>>
>> NDSolve[{x1'[t] == (Sin[t] + a)*x2[t],
>> x2'[t] == (Cos[t] - a*x1[t]^2)*x1[t], x1[0] == 1,
>> x2[0] == 0}, {x1[t], x2[t]}, {t, 0, 10}]
>>
>> But I need that the If condition evaluates in each time step. But I
>> haven't
>> been able to introduce the conditional in the ndsolve statment.
>>
>>
>
>
>
--
DrMajorBob at bigfoot.com
- References:
- How can I "perturbate" a NDSolve expresion?
- From: Iván Lazaro <gaminster@gmail.com>
- How can I "perturbate" a NDSolve expresion?