Wolfram Mathematica

« first day (2440 days earlier) ← previous day next day → last day (2034 days later) »

12:23 AM

The length of x=1 and y=1 is not same, how to make it same?

Plot[1/(1 - x), {x, -1, 1}, AspectRatio -> Automatic]

3 hours later…

zhk

2:53 AM

Hello @xzczd
These PDEs are giving me too much trouble,

Clear[Ra, Pr, n, pde1, pde2, pde3]

pde2[Ra_, Pr_, n_] =
u[x, y]*(D[v[x, y], x]) + v[x, y]*D[v[x, y], y] ==
Pr*D[D[v[x, y], x]^(n - 1)*D[v[x, y], x], x] + Ra*Pr*T[x, y];

pde3 = u[x, y]*D[T[x, y], x] + v[x, y]*D[T[x, y], y] ==
D[T[x, y], x, x];

With[{eps = 10^-3},
pde1 = D[u[x, y], x] + D[v[x, y], y] == D[v[x, y], y]*eps];

bcs = {{u[x, 0] == 0, v[x, 0] == 0, T[x, 0] == 0}, {u[-1/2, y] == 0,
v[-1/2, y] == 0, T[-1/2, y] == 1/2}, {u[1/2, y] == 0,

4 hours later…

Carl Woll

7:16 AM

@yode Add AxesOrigin->{0,0}

xzczd

@zhk Currently NDSolve can't solve static nonlinear PDE… FDM might help.

2 hours later…

yode

8:50 AM

@CarlWoll Any information about AspectRatio -> Automatic, AxesOrigin -> {0, 0} will make the length of x=1 and y=1 same?

Szabolcs

9:09 AM

Does anyone (@xzczd?) know where NDSolve's "MethodMonitor" option is documented?

Never mind

I misread "This feature is not officially documented" as "This feature is now officially documented"

@b3m2a1 The problem is that I still can't get it working :((( I'd really love to use this though

yode

@Szabolcs Any information can give about IGraphM 0.4?

3 hours later…

zhk

11:46 AM

@xzczd Thx I will try that too.

4 hours later…

b3m2a1

3:37 PM

@Szabolcs does the compilation work? If not could I bundle a .so file for Mac users like you to use (with things set up appropriately so it doesn’t try to load off of a Mac)? I’ll sit down with a 3.6 virtualenv and see if I can get it to work there and figure out what’s wrong if not. Does InstallPython feed you any error message?

If not there’s a debugging feature where I write to a log file and I’ll write some logic for customizing that via command-line in 1.0.2. I want to make sure this is working for you.

vasili111

3:57 PM

If Mathematica does not know of any symbol it colors it blue.
If Mathematica sees any syntactic error it colors it red.
1. Is the all above true?
2. Any other coloring that is useful to know?

Henrik Schumacher

4:40 PM

@vasili111 Green and turquoise for are used for scoped variables.

1 hour later…

vasili111

5:51 PM

@HenrikSchumacher thanks.

Szabolcs

6:23 PM

@b3m2a1 Still keeps complaining about <cstdint> not found, but I just realize that this header is C++11 only, and the default on mac is still C++98. Should add the -std=c++11 option.

Let me try now ...

@b3m2a1 Maybe this stackoverflow.com/a/23261706/695132

I just don't get it

running the compiler command directly works

running setup.py doesn't

maybe last comment here: github.com/vrasneur/pyfasttext/issues/1

doesn't look like it has a gcc though

@b3m2a1 I think it's not the compiler that's different. I think anaconda has installed C++ headers, and those are the ones being used when compiling with distutils. They lack cstdio. If I compile from the command line, the system headers are used.

That's just a guess though

Another guess is that it might set some environment variables to make this compatible with old macOS from before C++11 support

Kuba

7:14 PM

@vasili111 check menu Edit > Preferences then Appearance > Syntax Coloring

vasili111

@Kuba Nice link. Thank you :)

Szabolcs

7:35 PM

@b3m2a1 I give up ... I cannot figure out the difference between running a compiler command directly and running it through setup.py

One works, the other doesn't

Szabolcs

7:55 PM

@b3m2a1 Breakthrough! evernote.com/client/…

What's happening is that distutils sets MACOSX_DEPLOYMENT_TARGET=10.7 for some reason. This disables C++11 support. For that we need 10.9 at least.

But why does it set it?

FINALLY! Setting MACOSX_DEPLOYMENT_TARGET=10.9 manually gets me through this error, but now there's other problems

1 hour later…

b3m2a1

9:02 PM

@Szabolcs wonderful to hear this one is resolved. I’ll handle all this as soon as I can which will probably be tonight for me (Monday morning for you). I know some people suggest distribution not through distutils, too, and perhaps I can try something like that.

What other problems are you hitting now? I think I didn’t end up using cstdint in then end, so I can remove it, but I think I may eventually want it if issues crop up so I’d prefer to have it compilable that. I think the only proper C++ feature I use is templates, so if need be I can also rewrite things to be pure C instead and just use stdint.h and things.

The JLink client library is pure C

vasili111

Why this does not returns 11?
In[35]:= g[x_]:=x+1
In[36]:= NestWhile[g, 1, 10]