Wolfram Mathematica

12:21 AM

@Verbeia removed all but the "pertinent" ones of mine.

1:14 AM

@MrWizard Incidentally, DensityPlot has the Evaluated option, also, but it is set to SystemPrivate`$Evaluated`, which of course is set to Automatic.

belisarius

1:25 AM

@Verbeia I think I'll leave mine in place. I always considered it a good welcome reminder ... if you don't mind.

Verbeia

1:42 AM

@belisarius I didn't mean yours - those welcomes are really useful and I think they have been one of the things that have helped expand the community. But the stuff about his registration is all superseded now.

Oleksandr R.

2:00 AM

Anyone here at the moment?

Found an odd behaviour and wondering what others think of it.

Given:

f1 = Function[{x}, x];
With[{cf = Compile[{{x, _Real, 0}}, x]},
  f2 = cf;
  f3 = Function[{x}, cf[x]];
  ];

and

ClearAll[g];
g[f : _Function | _CompiledFunction] := (Print["memoized"]; g[f] = f)

we have:

g[f1]; g[f1]

memoized
Function[{x}, x]

as you'd expect. However,

g[f2]; g[f2]

gives:

memoized
memoized
CompiledFunction[{x},x,-CompiledCode-]

Memoization didn't work!

We can fix it using:

ClearAll[g];
g[f : _Function | _CompiledFunction] := (Print["memoized"]; g[Verbatim[f]] = f)

(Note the added Verbatim.)

I take this to mean that CompiledFunction is another object with internal state not represented by its FullForm. Any other views on this?

(Incidentally behaviour is the same for f2 and f3.)

Hello, @Vitaliy; @MrWizard -- any thoughts on that? Are there any other functions like this? LibraryFunction; maybe Graph?

belisarius

2:28 AM

@yoda I LOVE your link to the medical paper! I spent some time helping physicians to publish, but that one is a real perl

@yoda BTW It took me almost a week to convince a research group that doing statistics out of three cases is rather unconvincing

Mr.Wizard

@Oleksandr sorry, I'm a bit tired and hungry at the moment. I'll take a look at this later.

Oleksandr R.

@MrWizard not to worry; don't feel obliged. I just found this behaviour unexpected and had wondered whether others felt similarly or had run across it before.

yoda

@belisarius ha, very true!

I was at this conference and talking to a chap about his poster... he was telling me about his product (sensory aids) that worked much better than his competitors. He told me that his had a 75% success rate, whereas his competitors had only 33%...

Those nice numbers made me pause. Poking further, the guy explained that his "statistical results" were from a sample size of 4 and he took the numbers for his competitor's from another study with a sample size of 3!

rcollyer

3:05 AM

@yoda I'm impressed.

A friend of mine took over her surgeon's statistics because he did not have a clue.

Leonid Shifrin

11:42 AM

@OleksandrR This is an expected behavior (at least I expected it). The thing is that, when you look at the FullForm of your compiled function, you will see `CompiledFunction[{8, 8., 5468}, {_Real},
{{3, 0, 0}, {3, 0, 0}}, {}, {0, 0, 1, 0, 0},
{{1}}, Function[{x}, x], Evaluate]`. This contains a type specification _Real, which happens to be a pattern. Therefore, your (memoized) pattern does not match,

@OleksandrR just as this does not match: MatchQ[_Real, _Real]. Thus, you need Verbatim, which is useful in exactly such cases. I used memoization of compiled functions before, and was also puzzled at the time.

Stack Exchange

11:58 AM

1

Q: Extending the Mathematica highlighter to gist and github

I hope this is not too off topic, but does somebody know, how hard it would be to extend the current Mathematica syntax highlighter to work on gists and github in general? Or perhaps we should just contact github and ask them to add yet another language syntax highlighter? They have Matlab, so w...

Transcript for