logairthmic contour plot

Community matplotlib-users
1

Hi,

im actually trying to make a countour plot Z=f(X,Y) from two variables X,Y .
My Problem is that i have to use a logarithmic scale for the Z values.
If i plot the data with the logarithmic scale it gets pretty ugly, because i
have a lot of values which are zero,
which means on the log scale the value goes to -inf.
Here is an example what i mean

http://www.imagebanana.com/view/qh1khpxp/example.png

I acutally have no idea how to make the plot look better,
maybe somebody has an idea ?

thank you

···

--
View this message in context: http://old.nabble.com/logairthmic-contour-plot-tp34007155p34007155.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

2

Hi,

im actually trying to make a countour plot Z=f(X,Y) from two variables X,Y .
My Problem is that i have to use a logarithmic scale for the Z values.
If i plot the data with the logarithmic scale it gets pretty ugly, because i
have a lot of values which are zero,
which means on the log scale the value goes to -inf.
Here is an example what i mean

http://www.imagebanana.com/view/qh1khpxp/example.png

I acutally have no idea how to make the plot look better,
maybe somebody has an idea ?

Use np.ma.masked_less to mask out values below some threshold before taking the log.

e.g.,

import matplotlib.pyplot as plt
import numpy as np
x = np.arange(0, 1, 0.01)
y = np.arange(0, 8, 0.05)
X, Y = np.meshgrid(x, y)
Z = 10 ** (-5 + 11 * X * np.sin(Y))
Z = np.ma.masked_less(Z, 1e-4)
Zlog = np.ma.log10(Z)
CS = plt.contourf(X, Y, Zlog, levels=np.arange(-3, 5.01, 1.0), extend='both')
plt.colorbar()

Eric

···

On 06/13/2012 07:31 AM, jonasr wrote:

thank you

3

List,

I'm making a scatter plot using a for loop. Here's a simple example..

for i in range(10):
     x=rand()
     y=rand()
     scatter(x,y,label='point')

legend()
show()

When you do this, you get a legend entry for every single point. In this case, I get 9 entries in my legend.

Is there a way to only get a single entry? I have looked into creating the legends by hand, but I'm not having much luck. Googling, only turned up a single example of someone else with the same problem.

Help me list, you're my only hope.

Steven

4

Whoops, I forgot to change the subject. Sorry list.

List,

I'm making a scatter plot using a for loop. Here's a simple example..

for i in range(10):
     x=rand()
     y=rand()
     scatter(x,y,label='point')

legend()
show()

When you do this, you get a legend entry for every single point. In this case, I get 9 entries in my legend.

Is there a way to only get a single entry? I have looked into creating the legends by hand, but I'm not having much luck. Googling, only turned up a single example of someone else with the same problem.

Help me list, you're my only hope.

Steven

···

On 06/13/2012 01:33 PM, Eric Firing wrote:

On 06/13/2012 07:31 AM, jonasr wrote:

Hi,

im actually trying to make a countour plot Z=f(X,Y) from two variables X,Y .
My Problem is that i have to use a logarithmic scale for the Z values.
If i plot the data with the logarithmic scale it gets pretty ugly, because i
have a lot of values which are zero,
which means on the log scale the value goes to -inf.
Here is an example what i mean

http://www.imagebanana.com/view/qh1khpxp/example.png

I acutally have no idea how to make the plot look better,
maybe somebody has an idea ?

Use np.ma.masked_less to mask out values below some threshold before
taking the log.

e.g.,

import matplotlib.pyplot as plt
import numpy as np
x = np.arange(0, 1, 0.01)
y = np.arange(0, 8, 0.05)
X, Y = np.meshgrid(x, y)
Z = 10 ** (-5 + 11 * X * np.sin(Y))
Z = np.ma.masked_less(Z, 1e-4)
Zlog = np.ma.log10(Z)
CS = plt.contourf(X, Y, Zlog, levels=np.arange(-3, 5.01, 1.0),
extend='both')
plt.colorbar()

Eric

thank you

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
matplotlib-users List Signup and Options

--
Steven Boada
Dept. Physics and Astronomy
Texas A&M University
boada@...3847...

5

Perhaps not exactly what you want, but an answer is don't use a loop:

x = rand(10)
y = rand(10)
scatter(x,y, label='points')
legend()
draw()
show()

Of course if you want to color each point differently, then this won't work.

M

···

On 6/13/12 3:23 PM, Steven Boada wrote:

Whoops, I forgot to change the subject. Sorry list.

List,

I'm making a scatter plot using a for loop. Here's a simple example..

for i in range(10):
      x=rand()
      y=rand()
      scatter(x,y,label='point')

legend()
show()

When you do this, you get a legend entry for every single point. In this
case, I get 9 entries in my legend.

Is there a way to only get a single entry? I have looked into creating
the legends by hand, but I'm not having much luck. Googling, only turned
up a single example of someone else with the same problem.

Help me list, you're my only hope.

6

Are you trying to make 9 scatter plots? In your for loop, if you are trying to make a set of x values and a set of y values, then I think this is wrong. Since you didn’t provide import statements I don’t know which rand() function you are using. Assuming it is scipy.rand(), you will only have one x value and one y value, not much of scatter chart with just one point :slight_smile:

Otherwise, Mike’s suggestion is valid.

Regards,
Daniel

···

On Jun 13, 2012 3:35 PM, “Steven Boada” <boada@…4110…> wrote:

Whoops, I forgot to change the subject. Sorry list.

List,

I’m making a scatter plot using a for loop. Here’s a simple example…

for i in range(10):

 x=rand()

 y=rand()

 scatter(x,y,label='point')

legend()

show()

When you do this, you get a legend entry for every single point. In this

case, I get 9 entries in my legend.

Is there a way to only get a single entry? I have looked into creating

the legends by hand, but I’m not having much luck. Googling, only turned

up a single example of someone else with the same problem.

Help me list, you’re my only hope.

Steven

On 06/13/2012 01:33 PM, Eric Firing wrote:

On 06/13/2012 07:31 AM, jonasr wrote:

Hi,

im actually trying to make a countour plot Z=f(X,Y) from two variables X,Y .

My Problem is that i have to use a logarithmic scale for the Z values.

If i plot the data with the logarithmic scale it gets pretty ugly, because i

have a lot of values which are zero,

which means on the log scale the value goes to -inf.

Here is an example what i mean

http://www.imagebanana.com/view/qh1khpxp/example.png

I acutally have no idea how to make the plot look better,

maybe somebody has an idea ?

Use np.ma.masked_less to mask out values below some threshold before

taking the log.

e.g.,

import matplotlib.pyplot as plt

import numpy as np

x = np.arange(0, 1, 0.01)

y = np.arange(0, 8, 0.05)

X, Y = np.meshgrid(x, y)

Z = 10 ** (-5 + 11 * X * np.sin(Y))

Z = np.ma.masked_less(Z, 1e-4)

Zlog = np.ma.log10(Z)

CS = plt.contourf(X, Y, Zlog, levels=np.arange(-3, 5.01, 1.0),

extend=‘both’)

plt.colorbar()

Eric

thank you

Live Security Virtual Conference

Exclusive live event will cover all the ways today’s security and

threat landscape has changed and how IT managers can respond. Discussions

will include endpoint security, mobile security and the latest in malware

threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Matplotlib-users mailing list

Matplotlib-users@lists.sourceforge.net

https://lists.sourceforge.net/lists/listinfo/matplotlib-users

–

Steven Boada

Dept. Physics and Astronomy

Texas A&M University

boada@…3847…

Live Security Virtual Conference

Exclusive live event will cover all the ways today’s security and

threat landscape has changed and how IT managers can respond. Discussions

will include endpoint security, mobile security and the latest in malware

threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Matplotlib-users mailing list

Matplotlib-users@lists.sourceforge.net

https://lists.sourceforge.net/lists/listinfo/matplotlib-users

7

Well I am doing a lot more than this simple example shows. Point is
that there are nine different points each with their own legend
entry.

I could put it all out of the for loops, but it is all already

written, and I’d rather just fix the legend at the end than move
sections of the code around. I’m willing to do it, if that is the
only choice, but I wanted to ask before I commit my time.

Wouldn't it be a good (smart) thing for the code to lump all the

points with the same label together? This would be a great feature
to be added IMO.

S
··· 
-- Steven Boada
Dept. Physics and Astronomy
Texas A&M University

boada@…3847…

8

I have a routine where I start outside the loop with and empty sequence

leg =

then add at each iteration the label

leg.append('mylabel')

then call legend with the sequence at the end

legend(leg)

I think I once truncated the list and got part only

legend(leg[:2])

It depends on what you want to do.

   Paul

···

On Wed, Jun 13, 2012 at 8:56 PM, Mike Kaufman <mckauf@...287...> wrote:

On 6/13/12 3:23 PM, Steven Boada wrote:

Whoops, I forgot to change the subject. Sorry list.

List,

I'm making a scatter plot using a for loop. Here's a simple example..

for i in range(10):
x=rand()
y=rand()
scatter(x,y,label='point')

legend()
show()

When you do this, you get a legend entry for every single point. In this
case, I get 9 entries in my legend.

Is there a way to only get a single entry? I have looked into creating
the legends by hand, but I'm not having much luck. Googling, only turned
up a single example of someone else with the same problem.

Help me list, you're my only hope.

Perhaps not exactly what you want, but an answer is don't use a loop:

x = rand(10)
y = rand(10)
scatter(x,y, label='points')
legend()
draw()
show()

Of course if you want to color each point differently, then this won't work.

M

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
matplotlib-users List Signup and Options

--

* * * * * * * * Paul Kuin, Mullard Space Science Laboratory, UCL * * * *
Dr. N.P.M. Kuin (npmk@...3511...)
phone +44-(0)1483 (prefix) -204256 (work) -276110 (home)
mobile +44(0)7806985366 skype ID: npkuin
Mullard Space Science Laboratory – University College London –
Holmbury St Mary – Dorking – Surrey RH5 6NT– U.K.

9

2012/6/13 Mike Kaufman <mckauf@...287...>:

Whoops, I forgot to change the subject. Sorry list.

List,

I'm making a scatter plot using a for loop. Here's a simple example..

for i in range(10):
x=rand()
y=rand()
scatter(x,y,label='point')

legend()
show()

When you do this, you get a legend entry for every single point. In this
case, I get 9 entries in my legend.

Is there a way to only get a single entry?

Maybe you can adapt this to your use case:

for i in range(10):
     x=rand()
     y=rand()
     collection = scatter(x,y,label='point')

legend((collection,), ('Label',))
show()

Goyo

···

On 6/13/12 3:23 PM, Steven Boada wrote:

10

Assuming that you already have ten scatter plots, change the labels on the ones you don't want in the legend to '_nolegend_' (see help(legend))

for i in range(10):
   gca().collections[i].set_label('_nolegend_')
gca().collections[0].set_label('the one label I want')
legend()
draw()

M

···

On 6/13/12 4:06 PM, Steven Boada wrote:

Well I am doing a lot more than this simple example shows. Point is that
there are nine different points each with their own legend entry.

I could put it all out of the for loops, but it is all already written,
and I'd rather just fix the legend at the end than move sections of the
code around. I'm willing to do it, if that is the only choice, but I
wanted to ask before I commit my time.

Wouldn't it be a good (smart) thing for the code to lump all the points
with the same label together? This would be a great feature to be added IMO.

11

The

gca().collections.set_label('label') works great.

Admittedly, I'm not sure why it works. I'm not that great with the collections stuff. But thanks!

S

···

On 06/13/2012 03:22 PM, Mike Kaufman wrote:

On 6/13/12 4:06 PM, Steven Boada wrote:

Well I am doing a lot more than this simple example shows. Point is that
there are nine different points each with their own legend entry.

I could put it all out of the for loops, but it is all already written,
and I'd rather just fix the legend at the end than move sections of the
code around. I'm willing to do it, if that is the only choice, but I
wanted to ask before I commit my time.

Wouldn't it be a good (smart) thing for the code to lump all the points
with the same label together? This would be a great feature to be added IMO.

Assuming that you already have ten scatter plots, change the labels on
the ones you don't want in the legend to '_nolegend_' (see help(legend))

for i in range(10):
    gca().collections[i].set_label('_nolegend_')
gca().collections[0].set_label('the one label I want')
legend()
draw()

M

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
matplotlib-users List Signup and Options

--
Steven Boada
Dept. Physics and Astronomy
Texas A&M University
boada@...3847...

12

If all your values are positive (and you are sure of it), you could
use the SymmetricalLogScale It uses log scale for large values (both
positive and negative), and linear for small ones.

http://matplotlib.sourceforge.net/devel/add_new_projection.html

I believe there is a way to do it without going symetrical (ie, only
defined for positive values), but I am unable to find it.

···

On Wed, Jun 13, 2012 at 7:31 PM, jonasr <jonas.ruebsam@...273...> wrote:

Hi,

im actually trying to make a countour plot Z=f(X,Y) from two variables X,Y .
My Problem is that i have to use a logarithmic scale for the Z values.
If i plot the data with the logarithmic scale it gets pretty ugly, because i
have a lot of values which are zero,
which means on the log scale the value goes to -inf.
Here is an example what i mean

http://www.imagebanana.com/view/qh1khpxp/example.png

I acutally have no idea how to make the plot look better,
maybe somebody has an idea ?

thank you
--
View this message in context: http://old.nabble.com/logairthmic-contour-plot-tp34007155p34007155.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
matplotlib-users List Signup and Options

13

First, this is another topic, so please, change the subject of the
message so it doesn't get messed up with others (and possible help
lost in the process).

Now, you are indeed plotting one dot at the time and generating a
label for it. If you don't want that, you have to plot the whole list
at the time:

x=[rand() for i in xrange(10)]
y=[rand() for i in xrange(10)]

scatter(x,y, label='points')
legend()
show()

where the definition of x includes a list comprehension (equivalent at
"for i in xrange(10): x.append(rand())" ).

On another topic, people are not usually fan of using from MODULEX
import *, as it can turn into poor code and name collisions. It is
nicer if you write "import pylab as plt", and refer to the functions
as plt.scatter and so on.

Regards.

···

On Wed, Jun 13, 2012 at 9:21 PM, Steven Boada <boada@...3847...> wrote:

List,

I'm making a scatter plot using a for loop. Here's a simple example..

for i in range(10):
x=rand()
y=rand()
scatter(x,y,label='point')

legend()
show()

When you do this, you get a legend entry for every single point. In this
case, I get 9 entries in my legend.

Is there a way to only get a single entry? I have looked into creating
the legends by hand, but I'm not having much luck. Googling, only turned
up a single example of someone else with the same problem.

Help me list, you're my only hope.

Steven

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
matplotlib-users List Signup and Options

14

Point of style: in general, yes, but pylab was intended for that to help transition matlab users. Pylab really shouldn’t be loaded as plt, because that is what pyplot is usually imported as. Of course, this is all just a matter of style and preference.

Cheers
!
Ben Root

···

On Thursday, June 14, 2012, Daπid wrote:

First, this is another topic, so please, change the subject of the

message so it doesn’t get messed up with others (and possible help

lost in the process).

Now, you are indeed plotting one dot at the time and generating a

label for it. If you don’t want that, you have to plot the whole list

at the time:

x=[rand() for i in xrange(10)]

y=[rand() for i in xrange(10)]

scatter(x,y, label=‘points’)

legend()

show()

where the definition of x includes a list comprehension (equivalent at

“for i in xrange(10): x.append(rand())” ).

On another topic, people are not usually fan of using from MODULEX

import *, as it can turn into poor code and name collisions. It is

nicer if you write “import pylab as plt”, and refer to the functions

as plt.scatter and so on.

Regards.