Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.soft-sys.math.mathematica > #15821 > unrolled thread
| Started by | Sebastian <sebhofer@gmail.com> |
|---|---|
| First post | 2011-02-04 06:42 +0000 |
| Last post | 2011-02-04 11:58 +0000 |
| Articles | 2 — 2 participants |
Back to article view | Back to comp.soft-sys.math.mathematica
Filtering data from numerical minimization Sebastian <sebhofer@gmail.com> - 2011-02-04 06:42 +0000
Re: Filtering data from numerical minimization Ray Koopman <koopman@sfu.ca> - 2011-02-04 11:58 +0000
| From | Sebastian <sebhofer@gmail.com> |
|---|---|
| Date | 2011-02-04 06:42 +0000 |
| Subject | Filtering data from numerical minimization |
| Message-ID | <iig75a$rkb$1@smc.vnet.net> |
I have a function f(n,x,y,z) that I want to numerically minimize with respect to x,y,z for a list of n=1...N. The function is pretty complicated and also contains other parameters, which are fixed as far as optimization is concerned. I use NMinimize to create a table for different values of n. This effectively leaves me with 4 lists (f_min, x_min,...), where f_min = f(x_min, y_min, z_min). I then plot these lists against the vector n. Depending on the other parameters the minimization sometimes works well, and sometimes it doesn't. The problem is the following: while the overall form of the curves is often easily recognizable by eye, there is some noise on top of it, i.e. some of the points (every 5th say) are just way off. As I don't think that f behaves that way, but rather NMinimize fails to find the correct value, I'd like to filter that noise. Filtering the points by hand, however, doesn't seem like a feasible solution. Does anyone know a good way to tackle this problem? I already tried running MovingMedian which works reasonably well but I'd like to keep the rest of the data untouched if possible. I'm happy for all suggestions! TIA, Sebastian
[toc] | [next] | [standalone]
| From | Ray Koopman <koopman@sfu.ca> |
|---|---|
| Date | 2011-02-04 11:58 +0000 |
| Message-ID | <iigpl3$21k$1@smc.vnet.net> |
| In reply to | #15821 |
On Feb 3, 10:42 pm, Sebastian <sebho...@gmail.com> wrote:
> I have a function f(n,x,y,z) that I want to numerically minimize with
> respect to x,y,z for a list of n=1...N. The function is pretty
> complicated and also contains other parameters, which are fixed as far
> as optimization is concerned. I use NMinimize to create a table for
> different values of n. This effectively leaves me with 4 lists (f_min,
> x_min,...), where f_min = f(x_min, y_min, z_min). I then plot these
> lists against the vector n.
>
> Depending on the other parameters the minimization sometimes works
> well, and sometimes it doesn't. The problem is the following: while
> the overall form of the curves is often easily recognizable by eye,
> there is some noise on top of it, i.e. some of the points (every 5th
> say) are just way off. As I don't think that f behaves that way, but
> rather NMinimize fails to find the correct value, I'd like to filter
> that noise. Filtering the points by hand, however, doesn't seem like a
> feasible solution. Does anyone know a good way to tackle this problem?
> I already tried running MovingMedian which works reasonably well but
> I'd like to keep the rest of the data untouched if possible.
>
> I'm happy for all suggestions!
> TIA, Sebastian
Try Tukey's "4253h,twice" -- t4253h below. I've wrapped the multiline
functions in parentheses so that I wouldn't have to worry about where
the CRs come.
med42[y_] := ( (* running medians of 4, recentered by
running medians of 2; endpoints copied *)
Join[{y[[1]]},
ListCorrelate[{.5,.5}, Join[{{.5,.5}.Take[y,2]},
Median/@Partition[y,4,1],
{{.5,.5}.Take[y,-2]}]],
{y[[-1]]}] );
med5[y_] := ( (* running medians of 5, endpoints copied *)
Join[{y[[1]], Median@Take[y,3]},
Median/@Partition[y,5,1],
{Median@Take[y,-3], y[[-1]]}] );
med3e[y_] := ( (* running medians of 3, endpoints adjusted *)
Join[{Median@{{3.,-2.}.Take[#, 2], y[[ 1]], #[[ 1]]}}, #,
{Median@{{-2.,3.}.Take[#,-2], y[[-1]], #[[-1]]}}]& [
Median /@ Partition[y,3,1] ] );
hann[y_] := ( (* 1:2:1-weighted running means; endpoints copied *)
Join[{y[[1]]}, ListCorrelate[{.25,.5,.25},y], {y[[-1]]}] );
s4253h[y_] := hann @ med3e @ med5 @ med42 @ y;
t4253h[y_] := # + s4253h[y - #] & [s4253h @ y];
y = {60,70,54,56,70,66,53,95,70,69,56,70,70,60,60,
60,50,50,48,59,50,60,70,54,46,57,57,51,51,59};
t4253h[y]
{60., 60.3594, 60.5781, 60.9375, 62.2109, 65.0078,
67.8438, 69.2813, 69.8438, 70.0508, 69.5313, 67.8711,
65.5625, 63.0078, 59.9336, 55.793, 51.8281, 50.2344,
50.2031, 52.0352, 55.6055, 57.375, 57.1328, 56.4063,
55.4141, 54.5781, 54.1875, 54.2148, 54.5742, 55.2031}
[toc] | [prev] | [standalone]
Back to top | Article view | comp.soft-sys.math.mathematica
csiph-web