gnuplot-info — For questions and discussion about gnuplot

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[Ethan M. <eam...@gm...>]

On Tue, Aug 14, 2018 at 2:19 PM, theozh <th...@gm...> wrote:

> ok, I try again.
> I create many graphs with gnuplot. I copy these graphs e.g. to Powerpoint
> or Word, etc.
> In these programs the images will appear "too large", i.e. I need them
> half size.
> Of course, I can scale them down manually in PPT or Word, over and over
> again and again.
> The idea was to copy a graph directly from gnuplot which will appear with
> the proper size in PPT, Word, etc.
>

I am not a frequent Word user, but I think there is a preferences setting
where you can change
the default import resolution.  For instance:

https://support.office.com/en-ie/article/change-the-default-resolution-for-inserting-pictures-in-office-2016-f4aca5b4-6332-48c6-9488-bf5e0094a7d2





> Simply reducing the pixel count in gnuplot is not an option because you
> will lose details and it will look pixelated.
> I also can copy&paste the clipboard into an image viewer (e.g. GIMP,
> IrfanView, etc.) and manually set the DPIs to 192 over and over again. Copy
> the image again and paste it into PPT, Word, etc. With this, the size will
> be the desired size.
> However, I wanted to a) avoid loosing details and pixelated images, b)
> avoid scaling in PPT, Word,... and c) avoid using a third program.
>
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/
> lists/listinfo/gnuplot-info
>

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[theozh <th...@gm...>]

ok, I try again.
I create many graphs with gnuplot. I copy these graphs e.g. to Powerpoint or Word, etc.
In these programs the images will appear "too large", i.e. I need them half size.
Of course, I can scale them down manually in PPT or Word, over and over again and again.
The idea was to copy a graph directly from gnuplot which will appear with the proper size in PPT, Word, etc.
Simply reducing the pixel count in gnuplot is not an option because you will lose details and it will look pixelated.
I also can copy&paste the clipboard into an image viewer (e.g. GIMP, IrfanView, etc.) and manually set the DPIs to 192 over and over again. Copy the image again and paste it into PPT, Word, etc. With this, the size will be the desired size.
However, I wanted to a) avoid loosing details and pixelated images, b) avoid scaling in PPT, Word,... and c) avoid using a third program.

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[Ethan M. <eam...@gm...>]

On Tue, Aug 14, 2018 at 1:45 PM, Ethan Merritt <eam...@gm...> wrote:

> On Tue, Aug 14, 2018 at 1:15 PM, theozh <th...@gm...> wrote:
>
>> ... wouldn't there be any possibility with gnuplot to "save" to the
>> clipboard?
>> Kind of a "clipboard" terminal allowing for the setting of dpi (PNG,BMP)
>> or cm (PDF, EMF, SVG)?
>>
>> I don't understand the question.  Isn't that exactly what it does now?
> It is for the wxt and qt terminals anyway.  There is one widget that saves
> to the clipboard
> and another widget that saves to a filetype of your choice.
>
> Can you back up one step and explain why you can about the dpi field?
>

   ^^^^^

   care


> In my experience most if not all programs that read a bitmap file ignore
> this field.
> For instance if you import a bitmap file into Word or LaTeX or PowerPoint
> or
> whatever you can scale it up or down to fit the desired size without
> regard to
> whatever the original dpi was.
> What program are you dealing with where this is an issue?
>
>

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[Ethan M. <eam...@gm...>]

On Tue, Aug 14, 2018 at 1:15 PM, theozh <th...@gm...> wrote:

> ... wouldn't there be any possibility with gnuplot to "save" to the
> clipboard?
> Kind of a "clipboard" terminal allowing for the setting of dpi (PNG,BMP)
> or cm (PDF, EMF, SVG)?
>
> I don't understand the question.  Isn't that exactly what it does now?
It is for the wxt and qt terminals anyway.  There is one widget that saves
to the clipboard
and another widget that saves to a filetype of your choice.

Can you back up one step and explain why you can about the dpi field?
In my experience most if not all programs that read a bitmap file ignore
this field.
For instance if you import a bitmap file into Word or LaTeX or PowerPoint or
whatever you can scale it up or down to fit the desired size without regard
to
whatever the original dpi was.
What program are you dealing with where this is an issue?

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[theozh <th...@gm...>]

... wouldn't there be any possibility with gnuplot to "save" to the clipboard?
Kind of a "clipboard" terminal allowing for the setting of dpi (PNG,BMP) or cm (PDF, EMF, SVG)?

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[theozh <th...@gm...>]

Ok, it's just a screenshot. I thought/hoped that this was some data from gnuplot which gnuplot still can somehow influence.
Well, then Windows operating system or the application apparently assumes that the clipboard contains an image with 96 dpi. Hmm, then probably there's not much which I can do? At least, I haven't found yet any settings in Windows or the application which can change this dpi value.

Re: [Gnuplot-info] How to set ppi/dpi from wxt output?

[Ethan A M. <eam...@gm...>]

On Tuesday, 14 August 2018 18:29:35 theozh wrote:
> The interactive wxt terminal offers the nice feature to copy the graph to the clipboard.
> When pasting this graph as bitmap into other applications the size will depend on the resolution (dpi or ppi).
> 
> Example: 
> A plot with 567 x 567 pixels will be 15x15cm when copied to other applications.
> Apparently, gnuplot sets the resolution to 96 dpi.

So far as I know gnuplot plays no part in this.  
This is basically a screen shot, right?  Pressing the clipboard widget in 
gnuplot just triggers a screenshot of the current window.
The dpi probably comes from the resolution of your screen, or at any rate
what the computer thinks is the resolution of your screen.
If it is possible to change this at all, it would be via some setting in
the clipboard manager program.

	Ethan

> However, this 15x15cm is to large and I have to manually scale it down in the application to e.g. 7.5x7.5cm
> 
> If it was possible to set the resolution to 192 dpi in gnuplot, this annoying scaling step for every graph would be obsolete. Changing the resolution with an additional program is also no practical solution.
> Creating a graph within gnuplot with half the size (283x283) is also no solution since many details will be missing. Copying to EMF might be an alternative, but I don't want to go via an additional file on disk.
> 
> Any ideas, alternatives?
> 
> 
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info

-- 

[Gnuplot-info] How to set ppi/dpi from wxt output?

[theozh <th...@gm...>]

The interactive wxt terminal offers the nice feature to copy the graph to the clipboard.
When pasting this graph as bitmap into other applications the size will depend on the resolution (dpi or ppi).

Example: 
A plot with 567 x 567 pixels will be 15x15cm when copied to other applications.
Apparently, gnuplot sets the resolution to 96 dpi.
However, this 15x15cm is to large and I have to manually scale it down in the application to e.g. 7.5x7.5cm

If it was possible to set the resolution to 192 dpi in gnuplot, this annoying scaling step for every graph would be obsolete. Changing the resolution with an additional program is also no practical solution.
Creating a graph within gnuplot with half the size (283x283) is also no solution since many details will be missing. Copying to EMF might be an alternative, but I don't want to go via an additional file on disk.

Any ideas, alternatives?

Re: [Gnuplot-info] How to place labels with different backgrounds within one plot?

[theozh <th...@gm...>]

Thanks again. Well, the multiplot "trick" is probably not the nicest but simplest workaround.

It looks like that the several identical plots on top of each other create, e.g. thicker axis labels and data lines. At least that's what I believe to see in the wxt terminal.
Unfortunately, interactive scaling in the wxt terminal will only work on the last (multi)plot and the other labels are gone.


### different label backgrounds in "one" plot
reset
set multiplot layout 1,1

unset label
set style textbox opaque fc rgb "red"
set label 1 "red textbox" right at graph 0.5,0.8 boxed

plot sin(x)

unset label
set style textbox opaque fc rgb "green"
set label 2 "green textbox" right at graph 0.5,0.7 boxed

replot

unset label
set style textbox opaque fc rgb "blue"
set label 3 "blue textbox" right at graph 0.5,0.6 boxed

replot

unset multiplot
### end code

Re: [Gnuplot-info] How to place labels with different backgrounds within one plot?

[Ethan M. <eam...@gm...>]

On Mon, Aug 13, 2018 at 11:43 AM, theozh <th...@gm...> wrote:

> Thank you, Ethan, for the explanations. OK, I see.
> Yes, "boxstyle N" would be the feature I am looking for.
> I am mainly working with wxt, pngcairo and pdfcairo (sometimes actually
> svg).
> How would you implement different text boxes backgrounds in these
> terminals?
>

I do not know of a good way to do it. That's why I started working on the
textbox
property instead.  But that didn't turn out to be easy either.

t looks like I cannot mix/add/subtract graph coordinates and char
> dimensions. How to find this offset automatically? How to deal with
> multi-line text?
>

Those are the problems, yes.

Any suggestions are welcome.
>

About the only thing I can think of using the existing unsatisfactory
options is to
use multiplot, changing the background color of the textbox style in between
plot commands within the multiplot.

     Ethan

Re: [Gnuplot-info] How to place labels with different backgrounds within one plot?

[theozh <th...@gm...>]

Thank you, Ethan, for the explanations. OK, I see.
Yes, "boxstyle N" would be the feature I am looking for.
I am mainly working with wxt, pngcairo and pdfcairo (sometimes actually svg).
How would you implement different text boxes backgrounds in these terminals?
Find below find a cumbersome attempt of a workaround using colored rectangles.

I was hoping that "set object rectangle at..." starts from given position with the lower left corner whereas "set object rectangle center..." has the center of the rectangle at the given position.
However, gnuplot help tells me that "set object rectangle at..." and "set object rectangle center..." are identical. With this I could realize some centered boxed text, but I don't know how to "box" left or right aligned text. As a consequence, I have to set an offset for the x position depending on the text length. Furthermore, it looks like I cannot mix/add/subtract graph coordinates and char dimensions. How to find this offset automatically? How to deal with multi-line text?
Any suggestions are welcome.


### workaround for different textbox backgrounds in one plot
reset
unset label
unset object

Label_1_text = "This is a red text box"
Label_2_text = "This is a green text box"
Label_3_text = "This is a blue text box"
Label_1_color = "red"
Label_2_color = "green"
Label_3_color = "blue"
Label_1_x = 0.1
Label_1_y = 0.8
Label_2_x = 0.1
Label_2_y = 0.7
Label_3_x = 0.1
Label_3_y = 0.6
Offset_1_x = 0.10 # How to find this value automatically depending on text length???
Offset_1_y = 0.0 
Offset_2_x = 0.11 # How to find this value automatically depending on text length???
Offset_2_y = 0.0 
Offset_3_x = 0.10 # How to find this value automatically depending on text length???
Offset_3_y = 0.0 

set object 1 rect at graph Label_1_x+Offset_1_x,Label_1_y+Offset_1_y size char strlen(Label_1_text), char 1 fs solid 1.0 fc rgb Label_1_color front
set label 1 Label_1_text at graph Label_1_x,Label_1_y left front

set object 2 rect at graph Label_2_x+Offset_2_x,Label_2_y+Offset_2_y size char strlen(Label_2_text), char 1 fs solid 1.0 fc rgb Label_2_color front
set label 2 Label_2_text at graph Label_2_x,Label_2_y left front

set object 3 rect at graph Label_3_x+Offset_3_x,Label_3_y+Offset_3_y size char strlen(Label_3_text), char 1 fs solid 1.0 fc rgb Label_3_color front
set label 3 Label_3_text at graph Label_3_x,Label_3_y left front

plot sin(x)
### end code

Re: [Gnuplot-info] How to place labels with different backgrounds within one plot?

[Ethan A M. <eam...@gm...>]

On Monday, 13 August 2018 16:22:20 theozh wrote:
> How can I put several labels with different label background colors or box frames?
> I'm using gnuplot 5.2rc4
> 
> After painful placing colored rectangle objects behind label text, I found the "set style textbox" option. However, I haven't found how to place different styles within one plot.

Development on the textbox style options kind of stalled because it is very
hard to produce equivalent behaviour across all the various terminal types.
That is why it is documented as an "experimental" option.

textbox properties that either exist only conceptually or for only a few terminals:
- control over the margin sizes
- rotated box around rotated text
- fill color
- multiple textbox styles, to support "set label ... boxed [boxstyle N]"

You are asking about that last item, but it does not exist yet.
As I say, the main roadblock is that even the existing options have not
been implemented on terminals
	gd lua pslatex x11 canvas svg

It is particularly annoying that so far as I can tell there is nothing in 
the svg standard that would allow drawing boxed text.  So we may never be
able to support svg output of boxed labels, except maybe indirectly via
conversion from the qt or cairo terminals to a dumped svg file.

	Ethan



> The below example should create a red and green box but creates two green boxes. Apparently, the last style is applied to all textboxes.
> Can several textbox styles 1,2,3,... be defined? Like linetype, linecolor, etc.?
> 
> ### different label background styles within one plot???
> reset
> unset label
> unset object
> 
> set label 1 "This is some text" at graph 0.1, 0.7
> 
> set style textbox opaque border lc rgb "black" lw 1 fc rgb "red"
> set label 2 "Some text with red background" at graph 0.1, 0.6 boxed front
> 
> set style textbox opaque noborder lw 1 fc rgb "green"
> set label 3 "Some text with green background" at graph 0.1, 0.5 boxed front
> 
> plot 10*sin(x), 10*cos(x)
> ### end code

[Gnuplot-info] How to place labels with different backgrounds within one plot?

[theozh <th...@gm...>]

How can I put several labels with different label background colors or box frames?
I'm using gnuplot 5.2rc4

After painful placing colored rectangle objects behind label text, I found the "set style textbox" option. However, I haven't found how to place different styles within one plot.
The below example should create a red and green box but creates two green boxes. Apparently, the last style is applied to all textboxes.
Can several textbox styles 1,2,3,... be defined? Like linetype, linecolor, etc.?

### different label background styles within one plot???
reset
unset label
unset object

set label 1 "This is some text" at graph 0.1, 0.7

set style textbox opaque border lc rgb "black" lw 1 fc rgb "red"
set label 2 "Some text with red background" at graph 0.1, 0.6 boxed front

set style textbox opaque noborder lw 1 fc rgb "green"
set label 3 "Some text with green background" at graph 0.1, 0.5 boxed front

plot 10*sin(x), 10*cos(x)
### end code

[Gnuplot-info] restart session

[Patrick D. <pd...@gm...>]

Hello,

Sometimes it would be helpful to restart a gnuplot session which has been 
previously interrupted.

Is it possible?

Thank.

===========================================================================
 Patrick DUPRÉ                                 | | email: pd...@gm...
 Laboratoire de Physico-Chimie de l'Atmosphère | |
 Université du Littoral-Côte d'Opale           | |
 Tel.  (33)-(0)3 28 23 76 12                   | | Fax: 03 28 65 82 44
 189A, avenue Maurice Schumann                 | | 59140 Dunkerque, France
===========================================================================

Re: [Gnuplot-info] sum of squares of residuals

[Patrick D. <pd...@gm...>]

===========================================================================
 Patrick DUPRÉ                                 | | email: pd...@gm...
 Laboratoire de Physico-Chimie de l'Atmosphère | |
 Université du Littoral-Côte d'Opale           | |
 Tel.  (33)-(0)3 28 23 76 12                   | | Fax: 03 28 65 82 44
 189A, avenue Maurice Schumann                 | | 59140 Dunkerque, France
===========================================================================


> Sent: Thursday, August 02, 2018 at 11:25 PM
> From: "Hans-Bernhard Bröker" <HBB...@t-...>
> To: gnuplot <gnu...@li...>
> Subject: Re: [Gnuplot-info] sum of squares of residuals
>
> Am 02.08.2018 um 22:17 schrieb Patrick Dupre:
> 
> > When I do not weight my data, I get the following
> > final sum of squares of residuals : 1.85937
> > 
> > but, when I weight my data, I get:
> > final sum of squares of residuals : 54.6063
> > 
> > The big difference between the 2 values of the Chi2 indicates me that the
> > weighting of the data is not entirely correct.
> 
> That differences can indicate so many different things, all by itself, 
> that it effectively indicates nothing at all.
> 
> > When the data are weighted, if \Sum w(i) = 1, I must get a
> > similar Chi2.
> 
> Even that's not true, because the solution can still be a different one, 
> yielding a wildly different WSSR.
> 
> > How can I get the "weighted final sum of squares of residuals"?
> 
> You're already getting it.
Yes and no.

If \Sum w(i) = 1, I would get the same WSSR if I multiply all the weights (errors) by a constant.
This is not the case with gnuplot.
Actually, by definition the weights should such as \Sum w(i) = 1 (unitless)
This is not the case with gnuplot, strictly speaking the naming "weight"
is inappropriate.
The WSSR provided by gnuplot is unitless (\Sum (y(x) - y_i)^2 / (dy_i)^2) when weighting,
while it has the unit of y^2 when not weighting (uniform weighting) is used, which is not coherent.
Actually both WSSR (WSSR, and normalized WSSR) could be provided.

> 
> 
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>

Re: [Gnuplot-info] sum of squares of residuals

[Hans-Bernhard B. <HBB...@t-...>]

Am 02.08.2018 um 22:17 schrieb Patrick Dupre:

> When I do not weight my data, I get the following
> final sum of squares of residuals : 1.85937
> 
> but, when I weight my data, I get:
> final sum of squares of residuals : 54.6063
> 
> The big difference between the 2 values of the Chi2 indicates me that the
> weighting of the data is not entirely correct.

That differences can indicate so many different things, all by itself, 
that it effectively indicates nothing at all.

> When the data are weighted, if \Sum w(i) = 1, I must get a
> similar Chi2.

Even that's not true, because the solution can still be a different one, 
yielding a wildly different WSSR.

> How can I get the "weighted final sum of squares of residuals"?

You're already getting it.

[Gnuplot-info] sum of squares of residuals

[Patrick D. <pd...@gm...>]

Hello,

When I do not weight my data, I get the following
final sum of squares of residuals : 1.85937

but, when I weight my data, I get:
final sum of squares of residuals : 54.6063

The big difference between the 2 values of the Chi2 indicates me that the
weighting of the data is not entirely correct.
When the data are weighted, if \Sum w(i) = 1, I must get a 
similar Chi2.
How can I get the "weighted final sum of squares of residuals"?

Thank.

===========================================================================
 Patrick DUPRÉ                                 | | email: pd...@gm...
 Laboratoire de Physico-Chimie de l'Atmosphère | |
 Université du Littoral-Côte d'Opale           | |
 Tel.  (33)-(0)3 28 23 76 12                   | | Fax: 03 28 65 82 44
 189A, avenue Maurice Schumann                 | | 59140 Dunkerque, France
===========================================================================

Re: [Gnuplot-info] Fitting

[Hans-Bernhard B. <HBB...@t-...>]

Am 02.08.2018 um 12:58 schrieb Patrick Dupre:
> Hello,
> 
> I have data with error bars on x and y.
> To fit them properly, i.e., by weighting the data, I need to convert
> the x error bar in y error bar.

Not really.  If your gnuplot is not very ancient, it can do all that for 
you.  If you do not see "{y|xy|z}error" in your version's "help fit", 
get a newer one.

Re: [Gnuplot-info] Fitting

[Ethan A M. <eam...@gm...>]

On Thursday, 02 August 2018 12:58:24 Patrick Dupre wrote:
> Hello,
> 
> I have data with error bars on x and y.
> To fit them properly, i.e., by weighting the data, I need to convert
> the x error bar in y error bar.
> To do this I need to calculate the derivative of the fitting function.
> To make it right I need to calculate this derivative at each iteration and 
> then to calculate the new error bars.
> It possible to do it with gnuplot?

Gnuplot can handle data with separate error estimates on x and y.

>From the documentation in "help fit"
 
 A few shorthands for the `errors` qualifier are available:
 `yerrors` (for fits with 1 column of independent variable), and
 `zerrors` (for the general case) are all equivalent to `errors z`,
 indicating that there is a single extra column with errors of the
 dependent variable.

 `xyerrors`, for the case of 1 independent variable, indicates that there
 are two extra columns, with errors of both the independent and the
 dependent variable.  In this case the errors on x and y are treated by
 Orear's effective variance method.

There is a worked example of use in the fit demo.

	Ethan


> Thank.
> 
> ===========================================================================
>  Patrick DUPRÉ                                 | | email: pd...@gm...
>  Laboratoire de Physico-Chimie de l'Atmosphère | |
>  Université du Littoral-Côte d'Opale           | |
>  Tel.  (33)-(0)3 28 23 76 12                   | | Fax: 03 28 65 82 44
>  189A, avenue Maurice Schumann                 | | 59140 Dunkerque, France
> ===========================================================================
> 
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info

[Gnuplot-info] Fitting

[Patrick D. <pd...@gm...>]

Hello,

I have data with error bars on x and y.
To fit them properly, i.e., by weighting the data, I need to convert
the x error bar in y error bar.
To do this I need to calculate the derivative of the fitting function.
To make it right I need to calculate this derivative at each iteration and 
then to calculate the new error bars.
It possible to do it with gnuplot?

Thank.

===========================================================================
 Patrick DUPRÉ                                 | | email: pd...@gm...
 Laboratoire de Physico-Chimie de l'Atmosphère | |
 Université du Littoral-Côte d'Opale           | |
 Tel.  (33)-(0)3 28 23 76 12                   | | Fax: 03 28 65 82 44
 189A, avenue Maurice Schumann                 | | 59140 Dunkerque, France
===========================================================================

Re: [Gnuplot-info] sin(x)/x smoothing option?

[Reginald B. <pul...@ya...>]

For regularly sampled data this is trivial.  There is an excellent sinc(x) interpolator in the CWP/SU corpus which is BSD licensed that Dave Hale wrote.  I supported both CWP/SU and SLT/SU for many years, so I know the code base very well.    I'll extract that and remove any gratuitous dependencies later this week. I *wasn't * busy when I posted the question and was working on a project when suddenly I had higher priority tasks confront me.

Hale's interpolator is zero phase as that is what is used in seismic processing.  However, for many applications such as electrical engineering I think one needs a minimum phase interpolator.  I need to study the subject a bit and verify that is what is being done in digital storage oscilloscopes.  However, that's a relatively simple change to Hale's code.

Irregularly sampled data requires a different approach.  There is a relatively new (2004-2008) development in applied mathematics called "compressive sensing" developed by Emmanuel Candes and David Donoho.  The mathematical justification is *very* complex.  However, it's quite easy to use and is extremely general.  I refer to the field as "sparse L1 pursuits".  Applications include the "Netflix prize problem", predict what someone would like based on previous selections, inverse problems where you know the form of an equation that fits the data, but don't know the coefficients, "blind source separation" in which from a small number of recordings you select a single conversation in a room full of people all talking at the same time and a slew of other things.  It's a general solution to Ax=y under the constraint that most of the terms in x are zero and the columns of A are incoherent.  Mathematically the constraint is called the "restricted isometry property", any two combinations of columns has near zero cross correlation.  That is NP hard to solve, so in practice one attempts to solve Ax=y and if you get an answer it's been proven to be the optimal L0 solution, but obtained in L1 time.

The Marquardt-Levenberger L2 solver in gnuplot was how I always did things until I stumbled across sparse L1 pursuits.  I did a lot of data analysis with gnuplot using that.  Addling sparse L1 pursuits would make gnuplot a *very* powerful data analysis tool.

In "Numerical Methods that Work" Foreman Acton has sections on problems not to try to solve because they are illposed.  One of these is solving d = a*exp(A*t) + b*exp(B*t) which is the form of the radioactive decay of two isotopes.  Acton's advice is to send the problem to a chemistry lab to get A & B.  Sparse L1 pursuit will solve for a, A, b and B without any problems.

I had been solving inverse problems in fluid flow in porous media with gnuplot which struggled as it is a more complex variant of  the problem referred to by Acton.  Following Mallat's 3rd ed of "A Signal Processing Tour of Wavelets" I decided to try basis pursuit.  I was quite stunned by the results.  And then I remembered Acton's advice to the reader.  So I got a copy of "A Mathematical Introduction to Compressive Sensing" by Foucart and Rauhut and began a 3 year odyssey into the most complex and bizarre mathematical logic I have ever encountered.

So to summarize:

I am going to implement both sin(x)/x interpolation and sparse L1 pursuit inversions for gnuplot.  I've used gnuplot for about 25 years and consider it one of the best pieces of software I have ever used.  So I think it high time I made a significant contribution.

I don't know the code base for gnuplot at all and am only marginally familiar with the GLPK code which is needed for sparse L1 pursuits.  Normally I use the GMPL languagewhen using GPLK.  However, I've maintained several million lines of other people's code and know all the relevant background material very well.  So it's just a matter of putting in the time.  Sinc(x) interpolation is a couple of days work, so I'll get that done fairly quickly.  Sparse L1 pursuits are quite a bit more work, probably several weeks full time, so that will take a couple of months to complete as being "retired" has made me busier than I ever was before.

 

Re: [Gnuplot-info] sin(x)/x smoothing option?

[Ethan A M. <eam...@gm...>]

On Tuesday, 31 July 2018 20:49:48 Kevin Oberman wrote:
> On Tue, Jul 31, 2018 at 12:38 PM, Ethan Merritt <eam...@gm...> wrote:
> 
> > On Mon, Jul 30, 2018 at 5:10 PM, Reginald Beardsley via gnuplot-info <
> > gnu...@li...> wrote:
> >
> > > Is there an option to apply a sin(x)/x aka Whittaker interpolator to the
> > > points when plotting?  I'm working on a digital signal processing
> > tutorial
> > > and it would be helpful to be able to apply a sinc interpolator to the
> > data
> > > points.
> > >
> > > If there is not, I'd like to suggest considering that as an enhancement
> > as
> > > it is widely used in DSP applications.
> > >
> >
> > Not my field, so I'm working with the assistance of Wikipedia.
> > Do I understand correctly that the idea would be to plot the function
> >
> >       f(x) = Summation_over_all_data { yn * sinc( (x - nT)/T )}
> >
> > where the original data points were [xn, yn] and T = 1/(2*bandwidth) is a
> > constant specified by the user?
> > Analogous to the kdensity smoothing option, the plot command would be
> >
> >      plot $DATA smooth Whittaker {bandwidth <value>}
> >
> > In the case of kernel density smoothing there is an obvious default for the
> > bandwidth.
> > Is there something similar for this formula or would the bandwidth always
> > be required?
> > Would it be more natural to provide T or to provide bandwidth?
> >
> >     Ethan
> >
> 
> The 5.2 manual does not seem to mention "smooth Whittaker".

Well yeah, that's the point.  There is no such smoothing option now but
Reginald is suggesting that it would be useful to add one.
I was trying to understand what such an option would look like to the
user, and how well it would fit with the existing mechanisms for smoothing
in gnuplot.

	Ethan

> Any idea where
> I might find some information on it, especially as used in gnuplot?

Re: [Gnuplot-info] sin(x)/x smoothing option?

[Kevin O. <rko...@gm...>]

On Tue, Jul 31, 2018 at 8:49 PM, Kevin Oberman <rko...@gm...> wrote:

> On Tue, Jul 31, 2018 at 12:38 PM, Ethan Merritt <eam...@gm...>
> wrote:
>
>> On Mon, Jul 30, 2018 at 5:10 PM, Reginald Beardsley via gnuplot-info <
>> gnu...@li...> wrote:
>>
>> > Is there an option to apply a sin(x)/x aka Whittaker interpolator to the
>> > points when plotting?  I'm working on a digital signal processing
>> tutorial
>> > and it would be helpful to be able to apply a sinc interpolator to the
>> data
>> > points.
>> >
>> > If there is not, I'd like to suggest considering that as an enhancement
>> as
>> > it is widely used in DSP applications.
>> >
>>
>> Not my field, so I'm working with the assistance of Wikipedia.
>> Do I understand correctly that the idea would be to plot the function
>>
>>       f(x) = Summation_over_all_data { yn * sinc( (x - nT)/T )}
>>
>> where the original data points were [xn, yn] and T = 1/(2*bandwidth) is a
>> constant specified by the user?
>> Analogous to the kdensity smoothing option, the plot command would be
>>
>>      plot $DATA smooth Whittaker {bandwidth <value>}
>>
>> In the case of kernel density smoothing there is an obvious default for
>> the
>> bandwidth.
>> Is there something similar for this formula or would the bandwidth always
>> be required?
>> Would it be more natural to provide T or to provide bandwidth?
>>
>>     Ethan
>>
>
> The 5.2 manual does not seem to mention "smooth Whittaker". Any idea where
> I might find some information on it, especially as used in gnuplot?
>

Never mind. DuckDuckGo found lots!
--
Kevin Oberman, Part time kid herder and retired Network Engineer
E-mail: rko...@gm...
PGP Fingerprint: D03FB98AFA78E3B78C1694B318AB39EF1B055683

Re: [Gnuplot-info] sin(x)/x smoothing option?

[Kevin O. <rko...@gm...>]

On Tue, Jul 31, 2018 at 12:38 PM, Ethan Merritt <eam...@gm...> wrote:

> On Mon, Jul 30, 2018 at 5:10 PM, Reginald Beardsley via gnuplot-info <
> gnu...@li...> wrote:
>
> > Is there an option to apply a sin(x)/x aka Whittaker interpolator to the
> > points when plotting?  I'm working on a digital signal processing
> tutorial
> > and it would be helpful to be able to apply a sinc interpolator to the
> data
> > points.
> >
> > If there is not, I'd like to suggest considering that as an enhancement
> as
> > it is widely used in DSP applications.
> >
>
> Not my field, so I'm working with the assistance of Wikipedia.
> Do I understand correctly that the idea would be to plot the function
>
>       f(x) = Summation_over_all_data { yn * sinc( (x - nT)/T )}
>
> where the original data points were [xn, yn] and T = 1/(2*bandwidth) is a
> constant specified by the user?
> Analogous to the kdensity smoothing option, the plot command would be
>
>      plot $DATA smooth Whittaker {bandwidth <value>}
>
> In the case of kernel density smoothing there is an obvious default for the
> bandwidth.
> Is there something similar for this formula or would the bandwidth always
> be required?
> Would it be more natural to provide T or to provide bandwidth?
>
>     Ethan
>

The 5.2 manual does not seem to mention "smooth Whittaker". Any idea where
I might find some information on it, especially as used in gnuplot?
--
Kevin Oberman, Part time kid herder and retired Network Engineer
E-mail: rko...@gm...
PGP Fingerprint: D03FB98AFA78E3B78C1694B318AB39EF1B055683

Re: [Gnuplot-info] sin(x)/x smoothing option?

[Ethan M. <eam...@gm...>]

On Mon, Jul 30, 2018 at 5:10 PM, Reginald Beardsley via gnuplot-info <
gnu...@li...> wrote:

> Is there an option to apply a sin(x)/x aka Whittaker interpolator to the
> points when plotting?  I'm working on a digital signal processing tutorial
> and it would be helpful to be able to apply a sinc interpolator to the data
> points.
>
> If there is not, I'd like to suggest considering that as an enhancement as
> it is widely used in DSP applications.
>

Not my field, so I'm working with the assistance of Wikipedia.
Do I understand correctly that the idea would be to plot the function

      f(x) = Summation_over_all_data { yn * sinc( (x - nT)/T )}

where the original data points were [xn, yn] and T = 1/(2*bandwidth) is a
constant specified by the user?
Analogous to the kdensity smoothing option, the plot command would be

     plot $DATA smooth Whittaker {bandwidth <value>}

In the case of kernel density smoothing there is an obvious default for the
bandwidth.
Is there something similar for this formula or would the bandwidth always
be required?
Would it be more natural to provide T or to provide bandwidth?

    Ethan