[SciPy-User] Simulations

David Goldsmith d.l.goldsmith@gmail....
Thu Oct 8 17:51:13 CDT 2009


A little problematic as no wording is guaranteed to be transparent to all
users (different disciplines have different jargon for essentially the same
thing) but possibilities include "time-domain-like" (probably the preference
of engineers and many other breeds of scientist) or "discretely-incremented"
(perhaps the most general) or "discrete dynamical system" (my preference as
a mathematician).

DG

On Thu, Oct 8, 2009 at 2:26 PM, cool-RR <cool-rr@cool-rr.com> wrote:

> How would you phrase it then, instead of "all simulations"? I wouldn't want
> to deceive people.
>
> On Thu, Oct 8, 2009 at 11:15 PM, David Goldsmith <d.l.goldsmith@gmail.com>wrote:
>
>> Probably wise, but then Cesar's comment about being careful about what you
>> advertise applies.
>>
>> DG
>>
>>
>> On Thu, Oct 8, 2009 at 12:34 AM, cool-RR <cool-rr@cool-rr.com> wrote:
>>
>>> Hey Cesar and David,
>>> I thought about this and I think I better stick to the "Do one thing
>>> well" principle for now. Thanks for the insight though.
>>>
>>> Ram.
>>>
>>>
>>> On Tue, Oct 6, 2009 at 9:23 PM, Cesar Koers <ckoers@telenet.be> wrote:
>>>
>>>> Hi Ram,
>>>>
>>>>
>>>> A specific type of time domain solver for electromagnetics is e.g. TDFD
>>>> = time domain finite difference. It is based on discretizing Maxwell
>>>> equations in time & space
>>>>
>>>> But the Maxwell equations can also be expressed in the frequency domain
>>>> (thus for every frequency instead of every time instant). This leads to
>>>> FDFD = frequency domain finite difference.
>>>>
>>>> Other kinds of models, like based on finite elements (FE) can also be
>>>> developed in the frequency domain.
>>>>
>>>> Perhaps you're now thinking that your 'step function' would still work
>>>> in the frequency domain (response at frequency f_{i+} as a function of
>>>> response at frequency f_i), but this doesn't work (to my knowledge)
>>>> because it requires that the system is modeled by differential equations
>>>> in the frequency domain (haven't encountered this before)
>>>>
>>>>
>>>> Ah and some other 'bureaucracy' features:
>>>> * tracking time spent / calculating time remaining till end
>>>> * refining/coarsening time step to improve accuracy/reduce simulation
>>>> time respectively
>>>>
>>>> best regards
>>>>
>>>> C
>>>>
>>>>
>>>> cool-RR wrote:
>>>> > Hey Cesar,
>>>> >
>>>> > Your comments are interesting.
>>>> >
>>>> > Can you explain to me a bit about frequency domain simulations? Can
>>>> you
>>>> > give an example of a simulation simulating a real world process?
>>>> >
>>>> > I agree that GarlicSim must handle the bureaucracy well, as its job is
>>>> > to let the user write a simulation with as little bureaucracy as
>>>> possible.
>>>> >
>>>> > P.S. I registered garlicsim.org <http://garlicsim.org> and it is now
>>>> the
>>>> > main domain.
>>>> >
>>>> > Ram.
>>>> >
>>>> > On Mon, Oct 5, 2009 at 9:11 PM, Cesar Koers <ckoers@telenet.be
>>>> > <mailto:ckoers@telenet.be>> wrote:
>>>> >
>>>> >     Hi Ram,
>>>> >
>>>> >
>>>> >     I quickly read through your intro doc, I think you've explained
>>>> your
>>>> >     idea quite well.
>>>> >
>>>> >     One remarks though: I think your framework would fit well to
>>>> time-domain
>>>> >     (transient) models. But at this moment I don't see how you could
>>>> cast a
>>>> >     frequency domain simulation (commonly used in EM solvers) in it.
>>>> I'd be
>>>> >     careful with the idea that 'all simulations' fit into this.
>>>> >
>>>> >     What I think is key to success of this kind of framework is how
>>>> well it
>>>> >     handles the 'bureaucracy' of performing simulations (and speed,
>>>> but
>>>> >     you've already mentioned that the actual number crunching is up to
>>>> the
>>>> >     user of the GarlicSim). With this, I mean the boring stuff, like
>>>> e.g.:
>>>> >
>>>> >     * keeping track of which parameters vary between simulations
>>>> >     * extracting data from a set of simulations as a function of one
>>>> of
>>>> >     these parameters
>>>> >     * storing (and backing up) simulation results without taking up
>>>> too much
>>>> >     space and needing to invent unique and descriptive file names
>>>> >     * being able to redo a simulation (storing simulation parameters
>>>> with
>>>> >     results)
>>>> >     * making simulation reports
>>>> >     * comparing results with real-world data
>>>> >     * for long simulations, being able to continue simulation after a
>>>> crash
>>>> >
>>>> >     Just my 2 cents
>>>> >
>>>> >     Best regards
>>>> >
>>>> >     C
>>>> >
>>>> >
>>>> >     cool-RR wrote:
>>>> >      > Hello,
>>>> >      >
>>>> >      > This is not directly related to SciPy; I'm posting it here
>>>> because I
>>>> >      > figure that there may be people here who know the scientific
>>>> >     computing
>>>> >      > world enough to help me with my question.
>>>> >      >
>>>> >      > I've been working on an open-source scientific computing
>>>> project for
>>>> >      > about 6 months now, and I've come to the conclusion that it's
>>>> >     about time
>>>> >      > to find other users except myself for it, so I may get valuable
>>>> >     feedback
>>>> >      > about which direction I should be taking this project.
>>>> >      >
>>>> >      > The project is called GarlicSim (http://garlicsim.com
>>>> >      > <http://garlicsim.com/>). It's a Pythonic platform for working
>>>> with
>>>> >      > simulations. You may read more about it on the webpage. In
>>>> short,
>>>> >     it's a
>>>> >      > very general framework for creating, running and analyzing
>>>> >     simulations.
>>>> >      > It's not specific to any scientific field; Its role is to
>>>> provide a
>>>> >      > general mold into which all simulations can be cast. If you
>>>> want
>>>> >     to know
>>>> >      > more about it you can also read a (yet-incomplete) introduction
>>>> >      >
>>>> >     <
>>>> http://dl.getdropbox.com/u/1927707/Introduction%20to%20GarlicSim.doc>
>>>> >     to
>>>> >      > it.
>>>> >      >
>>>> >      > So what I want to know is, who would be good potential first
>>>> >     users for
>>>> >      > this, and how could I reach them?
>>>> >      > I'm not even sure which scientific field I would like to
>>>> target, so
>>>> >      > please suggest.
>>>> >      >
>>>> >      >
>>>> >      > Thanks,
>>>> >      > Ram Rachum
>>>> >      >
>>>> >      >
>>>> >      >
>>>> >
>>>> ------------------------------------------------------------------------
>>>> >      >
>>>> >      > _______________________________________________
>>>> >      > SciPy-User mailing list
>>>> >      > SciPy-User@scipy.org <mailto:SciPy-User@scipy.org>
>>>> >      > http://mail.scipy.org/mailman/listinfo/scipy-user
>>>> >
>>>> >     --
>>>> >     Gaetan Cesar Koers
>>>> >     Kerkveldweg 82
>>>> >     1851 Humbeek
>>>> >     +32(0)486 20 11 16
>>>> >     _______________________________________________
>>>> >     SciPy-User mailing list
>>>> >     SciPy-User@scipy.org <mailto:SciPy-User@scipy.org>
>>>> >     http://mail.scipy.org/mailman/listinfo/scipy-user
>>>> >
>>>> >
>>>> >
>>>> >
>>>> > --
>>>> > Sincerely,
>>>> > Ram Rachum
>>>> >
>>>> >
>>>> >
>>>> ------------------------------------------------------------------------
>>>> >
>>>> > _______________________________________________
>>>> > SciPy-User mailing list
>>>> > SciPy-User@scipy.org
>>>> > http://mail.scipy.org/mailman/listinfo/scipy-user
>>>>
>>>> --
>>>> Gaetan Cesar Koers
>>>> Kerkveldweg 82
>>>> 1851 Humbeek
>>>> +32(0)486 20 11 16
>>>> _______________________________________________
>>>> SciPy-User mailing list
>>>> SciPy-User@scipy.org
>>>> http://mail.scipy.org/mailman/listinfo/scipy-user
>>>>
>>>
>>>
>>>
>>> --
>>> Sincerely,
>>> Ram Rachum
>>>
>>> _______________________________________________
>>> SciPy-User mailing list
>>> SciPy-User@scipy.org
>>> http://mail.scipy.org/mailman/listinfo/scipy-user
>>>
>>>
>>
>> _______________________________________________
>> SciPy-User mailing list
>> SciPy-User@scipy.org
>> http://mail.scipy.org/mailman/listinfo/scipy-user
>>
>>
>
>
> --
> Sincerely,
> Ram Rachum
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User@scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/scipy-user/attachments/20091008/5bb49a4d/attachment-0001.html 


More information about the SciPy-User mailing list