Wolfram Mathematica - 2013-10-22

« first day (643 days earlier) ← previous day next day → last day (3824 days later) »

12:06 AM

@rm-rf Yeah, LaTeX is great for dealing with that stuff. Although my university would make minor changes to its formatting requirement every so often, so you'd pretty much always have to do some niggling and annoying updates.

1 hour later…

1:23 AM

@rm-rf That's very useful, thanks :)

4 hours later…

5:42 AM

Is there a way to render Cells inside a Grid(Column could work also) like so

Grid@{{Cell[BoxData["empty"], "Input"]}}

let me try posting that again

Pane[Grid[
  {{Cell[BoxData["empty"], "Input"]}}
  ], ImageSize -> {210, 110}, Scrollbars -> True]

Apologizes for the post it appears DocumentNotebook gives the functionality that is needed

6:11 AM

After further investigation DocumentNotebook doesn't appear to do everything I was originally looking for. I posted another answer here which I believe covers my and potentially the questionnaires' use case.

1 hour later…

7:17 AM

I don't really agree with the closure of this question but I don't feel strongly about it to the point of overriding the community:

0

Q: Image registration: optimization of the rigid transformationn between two images

I am trying to align two images by looking for the optimal rigid transformation between them. The problem can be seen as an optimization problem. This is my code in which I get an error when I try to get the transformation using FindMinimum (*Image reference*) imref = Rasterize[Graphics3D[Spher...

image-processing mathematical-optimization

Please take a second look at it.

@Mr.Wizard I voted (in part) to close because of the inital typos

@Mr.Wizard If you were to reopen, I'd be fine w/ it, but the reason I joined in closing was 1) typo + incorrect error description which makes the question not reproducible (even after fixing the typo) and 2) the OP has been told previously (in a different question) that this is a terrible way to do image registration/set up the cost function...

and because there already were two not so stellar questions out on the same topic by the OP

Besides, the problem he described in the question no longer exists... now the problem is entirely related to his specific function and no one knows what makes a good cost function for their needs for it other than the OP.

@rm-rf my firefox looses a lot of gravatars around here - is that a known issue?

7:32 AM

I've had it happen... usually it's some caching issue, fixed either by a hard refresh or simply waiting (take a nap)

@rm-rf it´s been around for some days (and quite a few naps)

but I´ll give the cache a good thump and see.

... and I think the OPs username does not do academia a lot of credit

1 hour later…

9:02 AM

@rm-rf @YvesKlett Thanks guys; I now agree with you. :-)

Is there any NaN or similar I can put in a PackedArray?

I just want something that I can look at and know "hey that is not a number"

In hopes of avoiding having a separate list "watchOutAtPositions={3, 931, 1311, ... }"

1 hour later…

10:29 AM

@ssch Null? Mathematica sometimes uses Missing.

Ah, packed array. nvm, this might not work then.

3 hours later…

1:36 PM

@ssch nope. You will have to use the watchOutAtPositions approach

Ended up with SparseArray approach:

ind = {c,r,a,z,y};
n=Length@data;
sp=SparseArray[
Delete[Range[n], ind] -> Delete[data, ind],
{n},
Missing["NotAvailable"]];

Which at least preserves some of the PackedArray niceness

@ssch Does it?

yea, MMA is all locked up computing but check sp["Properties"] and Developer`PackedArrayQ@sp@"The properties that have to do with data"

(where sp is a SparseArray created from packedArray1 -> packedArray2 rule

And ByteCount[sp] is (more or less) same as ByteCount[Range[n]]+ByteCount[packedArray]

Hmm... ok

n = 100000;
data = RandomReal[1, n];
ByteCount[data] + ByteCount[Range[n]] (* 1200336 *)
ind = {{1, 3, 8, 5041, 99}}\[Transpose];

sp = SparseArray[
Delete[Range[n], ind] -> Delete[data, ind],
Dimensions[data],
Missing["NotAvailable"]];
ByteCount[sp] (* 1200976 *)

Developer`PackedArrayQ@sp@# & /@ {"NonzeroValues", "NonzeroPositions","AdjacencyLists"}
(* {True, True, True} *)

Would be great if it had special handling for Indeterminate as default element so things like sp + sp or sp + data didn't unpack

2:05 PM

Hi guys. Since it is crowded here... is there someone who has a 32bit windows, Mathematica 9 and an installed C Compiler?

...and who is in the mood to try something. Takes only a minute.

@halirutan sorry, 64bit

@YvesKlett How's the conference?

@rm-rf pretty good from what I hear ;-)

Could not attend this year... conflicting dates and not a lot of budget left.

Ah... I thought you were going.

it was touch and go, I wanted to in any case.

but alas, my wanderlust was curtailed

@RHall should be the official SE ambassador, right?

2:14 PM

I think so... I hope he did actually continue discussions with SE and not leave it at the meta post

one of his comments here indicated that he would be going

ah, nice. That's good then.

Anyone read this: pixelologie.com/?page_id=165 ? I'm assuming he's interpreted the NDA correctly. If not, please delete this comment… :)

@YvesKlett Hmm darn.

@halirutan I do have a 32bit machine, but it its 1km from my current position (and there is no compiler installed there), so no luck today with me.

@cormullion very interesting! For the rest, no lawyer me.

2:25 PM

Hi all

@belisarius Hej!

Hi

a quickie: There is a function up there that takes an array and (I believe as a side effect) pads all the rows to the same max length, without having to specify how much pad is needed. I can't remember which one is ... damn

@belisarius Hehe... I know it ;-)

@halirutan a beer ?

2:28 PM

@belisarius PadRight/PadLeft?

@ssch I think not. You need to feed them the lengths

PadRight[{{1, 2}, {1, 2, 3}}] == {{1,2,0}, {1,2,3}}

@ssch Ok. So those are!

I'm always looking for them and overlooking the docs

@ssch thnks!

@belisarius Now I want to see that beer going through the wireless cable to ssch ;-)

@belisarius Let's hope that mathematica 10 has a FindMathematicaFunction[input_, ouptut_] ;)

2:31 PM

@ssch hehe ...

@ssch simply Beer[] would be nice as well

@halirutan Is that much beyond 3D-Printing now?

@belisarius I assume a 3D printed beer is extra dry..

@halirutan ted.com/talks/lee_cronin_print_your_own_medicine.html

@belisarius Good old German beers are made of magic, not of chemical inks ;-)

2:44 PM

@belisarius but who wants medicine?

@YvesKlett Shhh ... he's a cloaking the real purpose

crypto-beer?

I´m all beer, then

eh, ear

@YvesKlett I guess he is financially supported by a uni, and everyone there knows the real purpose and agree it's the most important discovery since fermentation. But you know there are things better not spoken out at the Dean's office

and the beer industry might not be amused

@YvesKlett O my! He could be killed before the contraption is finished

2:58 PM

Before he knows it, he'll be drowning in lawsuits from people that had a reaction to/died from chemicals in the delivery pipelines from the previous run

@rm-rf you are quite the optimist, are you not?

:D

I'm always a realist

Sadly, that means optimists see me as a pessimist and pessimists, the other way round

1

A: How to compute MIN$(A(x), B(x))$

Another one, possibly shorter, which works with any number of unequal degree polys: f[p_, x_] := Expand@FromDigits[Reverse[Min /@ Transpose@PadRight@CoefficientList[p, x]], x] f[{6 - 5 x + x^2 + 2 x^3 + 4 x^4 + 4 x^5 - 2 x^6, -4 + 7 x - 3 x^2 - 3 x^3 + 5 x^4}, x] -4 - 5 x - 3 x^2 - 3 x^3 +...

3:14 PM

@rm-rf as SE mod, you are an optimist in my book (idealist as well)

@YvesKlett and a toadist

:D

@VitaliyKaurov I golfed the cow before having yours at hand. I'm pretty sure it would have been impressive codegolf.stackexchange.com/q/11249/315

3:35 PM

our unanswered question count is rising... 1063 or ~10% now

@rm-rf Let's delete them all

@rm-rf 10% of all questions are unanswered??

@halirutan Yup

@belisarius There's a lot that probably needs to be...

I might have some time later this week... I'll try to sit down and answer some of them (at least, the low hanging fruits)

3:58 PM

@belisarius Do you have MATLAB?

@rm-rf nope

ok

Hello friends

I wanted to ask one question

Can anyone tell me about formulation steps of inverse problem using linear least square?

@Spider what is the "inverse problem"?

@Spider Were you, by any chance, looking for this room:

$Mathematics$ Mathematics

Associated with Math.SE; for both general discussion & math qu...

4:06 PM

@rm-rf thanks, but is it for whom? coz I am beginner..

This room is attached to the Mathematica site, dealing with the CAS software Mathematica. The other room is for general mathematics

Inverse problem is you have got measured data and inputs values, but

u dont know about the parameters of object.

e.g., u know the distance between two object (measuring the temperature of the inner side of Earth) and you send signal to measure temperature...

@Spider yes, we know what "inverse problems" are... belisarius was asking about your specific problem.

I am calculating the tomography in body (human)

It is tomography problem

I have attenuation data {y} through the body

and I have distance between each body cell like a sparse matrix

y=A*x

A- sparse matirx, x - unknown values in each cell

y - total attenuation in one measurement?

For this purpose I am using non-linear least square technique

But tomography is a well studied problem ... you have tons of books about it

4:17 PM

@ belisarius, yes you are right, even though, well stuied, I could not do it by myself, I have a lot of formula ))), could not used the least square method for it!!

that's why I asked steps (simple one ;) )

@Spider Step 1: Get a good book :)

The problem is my misunderstanding in one formula

Well, then why not start with that instead of leaving us in the dark?

@rm-rf : In least square, I will put the formula to the solution

room topic changed to Wolfram Mathematica: BeginPackage["Chat`", {" mathematica.stackexchange.com `"}] (no tags)

4:23 PM

but, I could not understand that formula, could you explain it to me. or this place is wrong for it?

@Spider Hehe, I told you long ago that this is the wrong place for it and also pointed you to the right place... :)

@rm-rf C'mon !

@ rm - rf : Thanks alot )) sorry

@belisarius Ok, ok... @Spider we'll help you this time (but in future, the math room is what you're looking for) — what's your specific trouble with the formula (and, btw, which formula)?

y = A*t $ or, $ f_i(t) = y_i - A_{ij}*t_j, so, i=1..m, j=1..n, (m<n)

I have formula but could not understand that:

F(t)=f'(t)

just a moment

t'_{j+1} = t'_{j} + C_{t}*F(t)^{T}[ F(t)^{T}*C_{t}*F(t)]^{-1}(y - F(t))

more or less

C_{t} - covariance matrix; {T} - transpose

t'_{j} - solution value

4:48 PM

It doesn't look like your dimensions are consistent...

F is a vector, so if you do C*F^T in the numerator, it implies that C is NxN and F is 1xN. Then you can't do F^T*C*F in the denominator...

yeap, it is under-determined problem

not underdetermined... you can't do this matrix multiplication

I have m measurements and have n unknowns ?

m<n?

but your F is a vector

I think I gave somethiing wrong by mistake

just a moment

I should check it out

I write again:

let's say not x instead of t, okay?

x'_{j+1} = x0 + C(0)*F_{j}^(T)*[ F_{j}*C(0)*F_{j}^{T}]^{-1}(F_{j}*(x'_{j} - x0 - - f(x'_{j}))

x0 - a priori point

C0 - a priori ccovariance matrix

for iteration, x0 can be simply "zero"

but, not necessarily

@ rm-rf : What I didnt understand in formula is that C(0) and multiplication process

5:10 PM

@Spider Well, I don't know what the individual terms mean physically in your problem, but you can consider C*F^T as a series of linear transformations and rotations by C on your vector F. The product F*C*F^T is a real and positive scalar quantity if your covariance matrix is positive definite, thus the multiplication by its inverse is some form of scaling/normalization of the aforementioned transformation.

What it appears to be is an iterative algorithm, where the state/value at the j+1th step is given by some constant + a modification of your state at step j. What those actually translate to in real terms depends on your specific problem.

Is this some form of MAP estimator for your quantity?

1 hour later…

6:33 PM

Hi friends, somebody can help me to recall how to use Pade Approximation for calculating coefficients of the cubic term in the polynomial ?

6:55 PM

@Irina Perhaps PadeApproximant[f, {x, x0, 3}]?

@Michael, thanks very much. Which kind of function can be f beside exponent ?

Is there some more natural Mathematica way to find out if a large list has more than k distinct elements? Length@DeleteDuplicates is too slow.

atLeastKDistinct = Compile[{
{list, _Real, 1}, {k, _Integer} (*, {tolerance, _Real} *)},
Block[{n = 0, known = Table[0., {k}], newValue},
Do[
newValue = True;
(* Pretend there's a better search algo here *)
Do[If[list[[i]] == known[[j]], newValue = False; Break[]], {j, n}];
If[newValue,
If[++n >= k, Break[]];
known[[n]] = list[[i]]
];
, {i, Length[list]}];

7:11 PM

@halirutan. Your javascript formatting aid plugin for the site editor has ceased to function on my system (Safari running on OS X 10.6.8). Are you aware of any recent changes in the site that might have caused this? Or should I be looking into my side of things for the cause? I miss having it a lot.

@halirutan. By "ceased to function", I mean the buttons don't even appear in my browser when I edit a question or answer.

@m_goldberg chat.stackexchange.com/… You are not alone, still works for me though (firefox)

@Irina Probably just about any f that has a power series expansion about x0. See the manual. (Also linked to the word PadeApproximant in my first reply.)

@Michael, the statement approximations works for linear terms, but not cubic;

7:29 PM

@m_goldberg The same here

@belisarius. Thanks for the confirmation. I hope this turns out to be repairable.

@m_goldberg The strange thing is that it's working for some people

@belisarius I think I must be misunderstanding. It seems to work for me. Example?

@MichaelE2 The icons aren't there anymore

@belisarius Oops! I thought I had clicked a different message. Yeah, it doesn't work for me, either

7:37 PM

@MichaelE2. What browser are you using on what system?

@MichaelE2 :)

@m_goldberg Safari. Mac 0S 10.8.5

@Irina I think I must be misunderstanding. It seems to work for me. Example?

@MichaelE2 And you see the plug-in editing buttons?

@Michael, shall I give my example ?

@m_goldberg I don't see them, not the "ref" etc. ones of halirutan's plug in

@Irina If it's not too long, sure. Somehow I think you're trying to do something different that I think you're doing.

7:41 PM

@Michael, let me tell you my problem here:

uh, it's quite big file actually Michael

@Irina Is there anyway to boil it down to a problem concerning a simple function?

@Michael, I am solving physics problem and have an exact function;

Linear terms give beautiful coefficients, but cubic term is in error

The linear terms and cubic terms -- are they the terms of the function or of the PadeApproximant? -- which do you mean?

huh, @Michael, done :) juhuuuu...happy happy

math is wonderful:)

@Irina Awesome. Have a great day! :D

7:52 PM

@Michael, dancke ;) thanks for your time and help.

@Irina You're welcome. Bitte, or something like that. I'm not sure what you native language is. :)

@Michael, awesome, you guessed it ;)

8:38 PM

@belisarius haha - that's a nice cow punchline

@VitaliyKaurov :)

user image

@VitaliyKaurov And I enjoyed your article a lot :)

More colors for all
Blending color gradients to expand color schemes available for plotting
http://community.wolfram.com/groups/-/m/t/142819

@belisarius yep that guy can write ;)

@VitaliyKaurov You're quite an artist :)

8:44 PM

@belisarius thx ;) not sure how to speed up that blender

@VitaliyKaurov post a question ;)

@belisarius haha i just might

@VitaliyKaurov Have you tried interpolating List @@ ColorData[ColorData["Gradients"][[2]]][x] and then blending by hand?

9:07 PM

@m_goldberg It's not working because the script depends on jquery. But it used to load it from files.quickmediasolutions.com/js/jquery.livequery.js and that URL isn't working anymore. It's easily fixable, just need to change the URL.

Perhaps also ping @halirutan

@Calle And why it's working for some people? any idea?

Cache probably

yup

@belisarius Yeah, I was going to say the same thing as ssch.

stopped working for me now that I clicked the link, damn you Calle :p

@ssch :)

9:17 PM

haha :D

I changed the url to this one and now it works again

@ssch you here?

9:32 PM

@belisarius Yea

@ssch Try this one and let me know the compared perfomance

h = System`Utilities`HashTable[]; i = 0;
k = 8;
l = RandomInteger[10, 20];
If[i <= k && ! System`Utilities`HashTableContainsQ[h, #], i++;
System`Utilities`HashTableAdd[h, #]] & /@ l;
If[i > k, Print["Exceeded"]]

 :11816041 For a 20k long packed array with 101 distinct reals:

    k       HashTable    DeleteDuplicates      atLeastKDistinct
    8          .085                     .00007                            .0001
    20        .081                        -"-                                  .00009
    100      .05                          -"-                                  .0005
    150      .09                          -"-                                  .036

fixed width in this chat isn't really too great :D

@ssch mmm .. oh well, another frustration in my list :)

{{"k", HashTable, DeleteDuplicates, "atLeastKDistinct"}, {8,
0.050402`5.153992763507569, 0.000049`2.141741073524482,
0.000057`2.2074198491684647}, {20, 0.047262`5.126057089348931,
0.000056`2.199733020502175, 0.000062`2.2439366829942307}, {100,
0.048016`5.132930971576746, 0.000049`2.141741073524482,
0.000386`3.0381322981677212}, {150, 0.062666`5.248576967861354,
0.000051`2.1591151695939024,
0.030271`4.932571761572084}} // MatrixForm

like that anyway

@ssch But I think the mapping Op consuming too much time

It isn't the way to go

9:49 PM

Map[(Print[#]; Return[True, Map]) &, {1, 2, 3}] kind of thing might do the trick

Scan[
If[
i <= k && ! System`Utilities`HashTableContainsQ[h, #],
If[++i >= k, Return[True, Scan]];
System`Utilities`HashTableAdd[h, #]
] &,
l] === True

Got it down quite a bit, interesting

will have to look at it further tomorrow, time for bed :)

@ssch bye!

10:12 PM

I'd totally forgotten about Scan...

10:31 PM

@belisarius are you here?

@Mr.Wizard yep

I'm playing a game -- I want to hit 77,777 rep points, and I'm going to be five off. (Yes, I know I'm weird.) If you could find an old question of mine that is worth voting for I'd appreciate it. :^)

@Mr.Wizard Mmmm .. that's worth a beer

@belisarius (trying to find a video clip....)

:o)

@Mr.Wizard ok, now you need me upvoting a question :)

there you are

10:41 PM

Yay! :D

hehehe

thx for humoring me

user image

Hello @rm-rf

Hey, I'm playing the same game! I need to hit 55,555 rep points, so please upvote uhh... about 510 answers of mine :D

2

@rm-rf Or we could downvote you to 44,44

lol

@rm-rf Did you catch 50,000 btw?

10:47 PM

@Mr.Wizard nope, was asleep...

It would've been 50,001 anyway. I was totally prepared to downvote you to round it, had I chanced upon it

When I realized I was going to be off by five I considered bumping a popular question of mine, but I thought asking belisarius for a vote was less disruptive.

@rm-rf I'd eat the downvote for the cause. :^)

@Mr.Wizard :D

@Mr.Wizard This is more controlled... you don't want to end up getting 2 upvotes rapidly, thus missing the window

@rm-rf You could place a +50 bounty and cast the downvote for a second chance...

Might be taking things just a bit too far though.

@Mr.Wizard You can easily fudge the numbers anytime... have you seen my addendum to that post?

user image

@rm-rf Yes :-p But there's no fun in editing a page. Playing with the actual numbers is a valid distraction.

10:53 PM

@Mr.Wizard Heh, just don't say you did it :P

@rm-rf You disappoint me. :-p

btw, why didn't you apply to go to the tech conference?

(this yr and last yr)

Lots of reasons, but honestly I wouldn't belong.

I don't see how you'd be any more out of place there than the average mma.se joe...

In fact if anything, you would've been a good choice to represent the community

(not that R Hall isn't, but just saying, given your participation here)

@rm-rf Thanks for the support, but not really. I merely cast the illusion of seeming like I know anything. Among a bunch of academics I'd be totally lost.

11:05 PM

@Mr.Wizard Don't feel bad about that. I've quite a few friends that are completely lost among humans in general

@belisarius I can always count on you to find some humor in things. Thanks. :)

:)

well with that rimshot I'm off to other things. See you both later.

@Mr.Wizard bye!

11:20 PM

@belisarius @halirutan and others: I asked the devs and apparently, they made some changes to the javascript, which is why the editor toolbar broke. Could one of you please open a bug report on meta, so that the devs can investigate?

@rm-rf I think @hali is the one ...

@rm-rf But, have you seen this chat.stackexchange.com/transcript/message/11815521#11815521

@belisarius Oh... the problem is that his server is down. Hmmm.

@belisarius Yup... ssch is right. Changing the link to jquery-utils.googlecode.com/svn-history/r307/trunk/src/… fixed it. Perhaps @halirutan can include the change.

11:35 PM

@rm-rf Yep, tried the fix & it works here too

« first day (643 days earlier) ← previous day next day → last day (3824 days later) »

Transcript for

Oct '1322

Wolfram Mathematica

Welcome! This is the main Mathematica chat room for mathematic...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile