people love making fun of how stupid americans are

a lot of people make that mistake, but at least the kid hasn't become stupid yet

@numbermaniac Unfortunately, you could do this experiment in a lot of countries with similar results.

agreed

does anyone know how to do arithmetic on two dynamic numbers?

like if i have two dynamic variables which are currently 1 and 2, how would i add them together

@numbermaniac Yes, I do

please tell

@numbermaniac The problem is as soon as you wrap Dynamic around it, you no longer have a number. Dynamic is sticky and you cannot work with them like you used to do. Even if you don't see it, the dynamic is still there.

Can you give a small example what you try to do?

Well essentially i have some dynamic variables being calculated

so {d, s, a, w} is {14,5,3,3} but each is dynamic

but it can't evaluate them because each is a dynamic

@numbermaniac Why don't you put the dynamic around the whole expression {d,s,a,w}?

I'm not sure I understand

d,s,a,w are being calculated using dynamic

like {d,s,a,w}=Dynamic[etc]&/@{2,3,5,7}

@numbermaniac Yes, and there is your problem. Let me give a simple example:

{Slider[Dynamic[x]], dynX = Dynamic[x]}
{Slider[Dynamic[y]], dynY = Dynamic[y]}

Now you can write dynX anywhere and it get's dynamically updated.

But the dynamic is always around it, you just don't see it.

What you could do is for instance this:

dynX + dynY /. Dynamic[x_] + Dynamic[y_] :> Dynamic[x + y]

but you don't want this.

What happens if you do

Dynamic[{d,s,a,w}={2,3,5,7}]

oh my god

i didn't realise you could do that

@numbermaniac In our example, it would be something like

Dynamic[{d, s, a, w} = {x, y, x + y, x - y}]

it works and d,s,a,w have not the Dynamic wrapper around it and can be used like normal variables

As a general rule of thumb, use Dynamic only to display things you want to be dynamically updated.

but the variable values are still updated dynamically

awesome, thank you!

@numbermaniac Not a problem.

Isn't there someone who is into deep learning who is up to write a tutorial, how we can do something like this in Mathematica? They take two images and re-color one of them regarding the colors of the other. If you don't think that this sounds impressive, then look at these amazing images:

@halirutan Without reading up further on this, i think this is similar to style transfer using neural networks. There is already a question about the latter.

@Szabolcs What problem about this?

@yode Check under Troubleshooting further down in the README.

But I still need to update for Mathematica 11.1. Certain functions have problems. I have not yet had the time to do it properly.

If anyone is willing to play a bit with the next IGraph/M on OS X / M11.1, please ping me and I will send you a paclet. Any feedback/testing would be most appreciated before the final release. Unfortunately, I cannot provide Windows/Linux/RPi binaries at this moment. (@kirma, @C.E.?)

Those who would be willing to try the paclet, please just send me an email.

Can anyone recommend me a GPU for using the Neural networks features in Mathematica? / Is there an offical list of which GPUs are currently supported?

This mean all my question is bad or I have a misunderstand what he say?

@yode I personally don't think the question was bad (he was being slightly grumpy there).

But it is certainly a difficult question. Also, as I mentioned in your sphere question, "equidistributed" seems to be the term you want, and not "equally spaced".

@J.M. I response you here.The qeustion you link is in 3-dimension.But this arbitrary just request 2D.

And the Union work well,just have two little difficulty,which have claim in my post.

And do m_goldberg will treat other user with such temper?

@yode Well, 2D is easier than 3D, but arbitrary regions are harder than well-known regions like the ball or the disk.

@yode As you say, you're doing a rejection method. It works, but as you've noticed, you don't have any control over the number of retained points.

Note another word here

@yode That comes with you wanting it for arbitrary regions. :)

@yode Maybe he had a bad day today; anyway, R. M. has said something.

Another thing to think about: are your arbitrary regions always convex, or can they be concave?

@J.M. If you can solve the case based on convex,I think that is a good start.But actually,the "arbitrary" is expected. :)

@yode I asked because my intuition tells me the concave case is harder.

@J.M. And just a ask,have you ever seen this official speaking?I thought that's what I want to express

@yode I don't have 11.1, but as I explained in your earlier question, the exact cases correspond to the vertices of the Platonic and Archimedean polyhedra.

Only approximate solutions are known in the other cases.

And I will thanks for that.

And "equally spaced" just sounds awkward in the sphere.

@J.M. If you think other term can improve my question,you can edit my original question for free.And I will thanks for that.

@J.M. - do you know much about string patterns in WL?

I have some strings, and I want to be able to match "some text here\", 1234]", "some text here\" , 1234]" and "some text here\" , 1234 ]"

I want to grab the part that matches NumberString

but I don't know how to specify an optional pattern

I suppose I can just define them via Alternatives

StringCases["some text here\", 1234]", __ ~~ "\"" ~~
   Repeated[" ", {0, 1}] ~~ "," ~~ Repeated[" ", {0, 1}] ~~
   n : NumberString ~~ Repeated[" ", {0, 1}] ~~ "]" :> n]

@JasonB

Thanks @MichaelHale! I had resorted to first stripping any white space, then doing my pattern matching, but your suggestion is probably a better solution

@JasonB It's actually easier if you want to allow 0 to infinity spaces because then you can use ~~" "...~~

@MichaelHale - I think I was overly specific in the question, I think RepeatedNull is exactly what I want (slaps forehead)

@Szabolcs Just back in the office where I have some reference material. I haven't found much of anything on real-world problems caused by "bad" random number generation but I was reminded of MacSpin on a MacPlus (where I had a copy of Mathematica on 3.5" diskettes). One of the examples that MacSpin promoted was where "structure" was observable in 3D from supposedly-random numbers: digitalassets.lib.berkeley.edu/sdtr/ucb/text/104.pdf.

I also found youtube.com/watch?v=3A5moyhfaQo.

@JasonB Michael's solution is better than what I had. :D

@halirutan someone else mentioned this in an answer to a post, but the NetTrain docs page seems to have added a demo of style transfer, under Applications > Computer Vision > Style Transfer. im giving it a shot now. reference.wolfram.com/language/ref/NetTrain.html#833168402