Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Time- and space-dependent interpolation from data file?
Posted 5 avr. 2010, 12:37 UTC−4 5 Replies
Please login with a confirmed email address before reporting spam
I'm using COMSOL 3.5a. I have a 1D domain and have written my PDEs using the transient diffusion mode from the model navigator.
I wish to create an interpolation function of experimental time-course data using the (Options->Function) tool.
My input file is set up with three columns - a schematic example with a domain length 2:
% time/s x/cm conc/mM
0 0 1
0 1 1
0 2 1
1 0 1
1 1 1.5
1 2 2
.
.
.
I need to create a function conc(x,t) that I can use in my PDEs. However, the Options->Function wizard only seems to allow functions of x (or other spatial dimensions) but not of time.
Is there a way to interpolate from data in a time-dependent manner?
Any help would be appreciated - thanks!
I wish to create an interpolation function of experimental time-course data using the (Options->Function) tool.
My input file is set up with three columns - a schematic example with a domain length 2:
% time/s x/cm conc/mM
0 0 1
0 1 1
0 2 1
1 0 1
1 1 1.5
1 2 2
.
.
.
I need to create a function conc(x,t) that I can use in my PDEs. However, the Options->Function wizard only seems to allow functions of x (or other spatial dimensions) but not of time.
Is there a way to interpolate from data in a time-dependent manner?
Any help would be appreciated - thanks!
5 Replies Last Post 12 juil. 2010, 10:55 UTC−4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Bump.
Still an open question. Any helpful hints?
Still an open question. Any helpful hints?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
if I make a .txt file of this type:
% x y conc
0 0 1
0 1 1
0 2 1
1 0 1
1 1 1.5
1 2 2
2 2 2.5
2 2 3.5
2 3 4
...
I generate a function conc(x,y) where x,y are just placeholders for any variable name
I can then call "conc(t,x)" in a "Postprocessing - Domain Plot Parameter - Line" call either as multiple plots over time with "x" my 1D dimension defined as x-axis variable, or as vertical spread lines showing the min max and extent over time (expression for x-axis) for x_start <= x <= x_end of my 1D sub-domain line going from x_start to x_end
but you need to use x,y,z at most as (dummy) variable name for the function file first line and then followed by the function name that should not contain .-,/+*"&$[]...
hope this helps
Ivar
if I make a .txt file of this type:
% x y conc
0 0 1
0 1 1
0 2 1
1 0 1
1 1 1.5
1 2 2
2 2 2.5
2 2 3.5
2 3 4
...
I generate a function conc(x,y) where x,y are just placeholders for any variable name
I can then call "conc(t,x)" in a "Postprocessing - Domain Plot Parameter - Line" call either as multiple plots over time with "x" my 1D dimension defined as x-axis variable, or as vertical spread lines showing the min max and extent over time (expression for x-axis) for x_start <= x <= x_end of my 1D sub-domain line going from x_start to x_end
but you need to use x,y,z at most as (dummy) variable name for the function file first line and then followed by the function name that should not contain .-,/+*"&$[]...
hope this helps
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Thanks for the feedback Ivar.
You suggested the following approach,
da/dt = f( a(x), conc(t,x) ).
For a static distribution over the x dimension, this requires clicking the "Use space coordinates as default function arguments" in the Functions dialog box to get conc(x). This doesn't work if I use one of the other spatial dimension variables as a placeholder for time.
Also, I may need to generalize this to a full spatial model where I interpolate from data to generate a function conc(t,x,y,z). Is there a way to create such a function that can be used by the solver?
Thanks again!
You suggested the following approach,
I generate a function conc(x,y) where x,y are just placeholders for any variable nameHowever, I need to clarify something. Beyond plotting data, I need to be able to hand this function to the solver as a term in my PDEs,
I can then call "conc(t,x)" in a "Postprocessing - Domain Plot Parameter - Line" call either as multiple plots over time with "x" my 1D dimension defined as x-axis variable, or as vertical spread lines showing the min max and extent over time (expression for x-axis) for x_start <= x <= x_end of my 1D sub-domain line going from x_start to x_end
da/dt = f( a(x), conc(t,x) ).
For a static distribution over the x dimension, this requires clicking the "Use space coordinates as default function arguments" in the Functions dialog box to get conc(x). This doesn't work if I use one of the other spatial dimension variables as a placeholder for time.
Also, I may need to generalize this to a full spatial model where I interpolate from data to generate a function conc(t,x,y,z). Is there a way to create such a function that can be used by the solver?
Thanks again!
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
There must certainly be, but I must admit I have never tried, but I see the utility. In Matlab you certainly can adapt things easier, but inside the GUI's only, I cannot tell like that. I would have to tweak and try it out myself, and just now I'm stuck on missing general force/torque loads in 3D structural so I'm rather locked on my (paid) job.
Anyhow, for the methodology, I would start trying out on a very simple case with 2-4 input points, in 1,2 then 3 coordinates, and a very simple example first
Have fun Comsoling
Ivar
There must certainly be, but I must admit I have never tried, but I see the utility. In Matlab you certainly can adapt things easier, but inside the GUI's only, I cannot tell like that. I would have to tweak and try it out myself, and just now I'm stuck on missing general force/torque loads in 3D structural so I'm rather locked on my (paid) job.
Anyhow, for the methodology, I would start trying out on a very simple case with 2-4 input points, in 1,2 then 3 coordinates, and a very simple example first
Have fun Comsoling
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Some days ago, I had a problem similar to the one of James: i.e set a time dependent mechanical BC using a time-dependent interpolation of data.
In particular, in my problem I had to assing an acceleration in a certain point of a body and perform a time-domain simulation.
By default Comsol 3.5a (that's the software version I have employed), in several mechanical modules allows to set displacements and not accelerations.
I have solved such a problem in the following way:
- import the data in tabular form (one column is the physical quantity, the other is time) in Comsol
- call this function "data" (this will be, by default, a function of the type "f(x)")
- setup a global equation where the function "data" is written as "data(t)", where t is the usual time variable.
In this way the software "understands" that the independent variable is a time and not a space.
I hope this helps.
In particular, in my problem I had to assing an acceleration in a certain point of a body and perform a time-domain simulation.
By default Comsol 3.5a (that's the software version I have employed), in several mechanical modules allows to set displacements and not accelerations.
I have solved such a problem in the following way:
- import the data in tabular form (one column is the physical quantity, the other is time) in Comsol
- call this function "data" (this will be, by default, a function of the type "f(x)")
- setup a global equation where the function "data" is written as "data(t)", where t is the usual time variable.
In this way the software "understands" that the independent variable is a time and not a space.
I hope this helps.