@laudiacay To ensure that you don't look stupid next time, let me give you an advice. When you try to plot something and the plot is empty, don't stare at your code for hours trying to figure out what's wrong.

The first thing you always want to do is evaluate the function you try to plot somewhere inside your plot-range. If you had tried to evaluate just this:

xi[10, 10]

you would have seen that your second radius call is not evaluated and there must be something wrong.

Re: moving goal posts: mathematica.stackexchange.com/questions/109257/…

It looks like Association lookup uses a more precise comparison than SameQ. Define ReallySameQ[x_, y_] := x === y && Not@MissingQ[<|x -> 0|>[y]]. Now ReallySameQ[1.\[RawBackquote]10., 1.\[RawBackquote]10.00000000001] is False. Although 1.\[RawBackquote]10. === 1.\[RawBackquote]10.00000000001 is True and their FullForms look identical.

@VladimirReshetnikov How does it compare to MatchQ?

MatchQ can distinguish these numbers. But, SameQ[Hold[Rational[1, 2]], Hold[Evaluate[1/2]]] and MatchQ[Hold[Rational[1, 2]], Hold[Evaluate[1/2]]] are both True. But ReallySameQ[Hold[Rational[1, 2]], Hold[Evaluate[1/2]]] is False. TreeForm shows the difference between them.

I wonder if there is a pair of expressions that is distinguished by MatchQ but not by ReallySameQ.

@VladimirReshetnikov Did you mean SameQ instead of MatchQ?

@R.M. No, I meant MatchQ (with patterns wrapped in Verbatim where necessary). It is more precise than SameQ on some numbers (e.g. MatchQ[1.\[RawBackquote]10., 1.\[RawBackquote]10.00000000001] is False).

@VladimirReshetnikov But this depends on what you have in mind.

BTW, can I somehow put the backquote in inline code enclosed by backquotes in this chat? 1.``10.

For your last examples there are surely a lot of use-cases where you want it to be true

@VladimirReshetnikov Yes, you can

``1.`10.``

Just wrap it in double back-ticks.

1.`10.

Nice

@VladimirReshetnikov About your DiracDelta question, I cannot say much as the mathematical theory behind such distributions is nothing I'm familiar with, but I learned that DiracDelta only makes sense in an integral expression.

I was looking for a way to check if expressions are completely indistinguishable and interchangeable in all contexts.

Why and when the simplifications you showed are legal should really be answered by a mathmatician.

@VladimirReshetnikov The problem is that some expressions are changed during reading the input.

Yes, sure. Suppose that expressions are already evaluated and stored in some variables x and y.

@VladimirReshetnikov Have you considered comparing the binary internal storage format of expressions?

How can I see this format?

@VladimirReshetnikov Hmm, I thought StringExport to something like MX might work, but it doesn't.

@VladimirReshetnikov I am sure you are aware that there already has been a lot of controverse about this. Something like this sometimes hits you hard when you do replacement and things are unexpected (like the situation with complex numbers).

For instance this paper here:

cs.berkeley.edu/~fateman/papers/better-rules.pdf

It is kind of related to what you try to do.

Just noticed another case when Uncompress@*Compress is not equivalent to Identity. MatchQ[1.`10., 1.`10.00000000001] is False, but MatchQ[1.`10., Uncompress@Compress@1.`10.00000000001] is True.

Thanks, I'll read the paper.

@VladimirReshetnikov You examples are as always very well thought of.

Thanks

@VladimirReshetnikov I'm not sure this is possible in highlevel Mathematica, but lets say you are transferring your two expression you try to compare over MathLink, then you can compare their byte-representation directionly. And it seems something like WSGetData() does exist as described in guide/LowLevelWSTPOperations.

The question is, is there an easier way to extract the real expression from within Mathematica somehow.

(I guess MatchQ is pretty close. Do you have any counter examples, where MatchQ doesn't work?)

Need to sleep. 7 am here.

Yes, MatchQ[Hold[Rational[1, 2]], Hold[Evaluate[1/2]]] - see above.

@Silvia Thanks for links. Well, can't argue with facts :) but I think you are over optimistic. The ratio and the rate of fixing FE related problems is way to low. Some that seem tiny are probably caused by fundamental problems in FE so I understand that they are not fixed. Yet they are around since I remember (V7). I don't even have hope for more than sporadic fixes.

@Silvia My new post is there to save time of people who are trying to make things neat, so they know where the limit is. Also, to have good source to refer to when one have to explain that limit to a boss graphics designer.

@MikeHoneychurch I was annoyed by this a long time. Lately I've just accepted that I don't know the market. So maybe the fact that my expectations and WRI target are not even close is something I should expect.

@MikeHoneychurch One thing I really liked about Mathematica is that it was fun. The people who made it clearly liked to just play. It had fancy graphics and sounds and cool (as opposed to purely useful) examples, and programming in it was intellectually challenging, thus exciting. I'm thinking about Mathematica many versions ago ...

@MikeHoneychurch In contrast, MATLAB did not nearly look as much fun. It seemed dry and task-focused and programming was straightforward but without giving that sense of satisfaction after finding a good solution.

I still like this about Mathematica, but they seem to have lost their balance. Lately, it's only about "cool" looking features, and practical usefulness is pushed into the background. When I went to a WRI presentation, they didn't show almost a single thing which would have demonstrated how Mathematica could be practically useful for my work ... they only showed how you can use natural language input, how you can tweet with it, retrieve the population of Azerbaijan with a built-in command, etc.

A new version of MATLAB was just released, and compare their release notes with Mathematica's: they clearly care much more about practical usability. mathworks.com/help/matlab/release-notes.html Breaking changes are clearly annotated, some bug fixes are listed, and many of the changes are about improving the usability of existing stuff rather than adding completely new things.

Interestingly, they also realized that the notebook inteface is the way to go and now MATLAB has a proper notebook :) mathworks.com/help/matlab/matlab_prog/…

Anyway

Is anyone here coming to the WRI even in Grenoble tomorrow?

AlphaGo won against Lee Sedol again...

Stupid question, but say I have some array of dimensions 2 by 3, call it a. If I then use Dimensions[a] I get {2,3} as output. What do I type to just get the 3 as output?

Dimensions[a][[2]]

@user3183724 What m0nhawk said above. It is not pretty but there is no prettier way to get the column count of a matrix. If you are certain that it is a matrix and you may want to use the row count as well, {nrow, ncol} = Dimensions[a] looks nicer.

Well, I should have said is that I don't know of a prettier way ... Mathematica has so many functions, one can never know ...

@Szabolcs {nrow, ncol} = Dimensions[a] wow, this one looks awesome.

Didn't know this works in Mathematica. I like this way of unrolling tuples in Python.

@user3183724 Some would prefer Last@Dimensions[...] instead of Dimensions[...][[2]]

Thanks guys!

@Szabolcs I was wrong I though about related issue, you may consider undeleting this.

Under TrackedSymbols / Possible Issues there is a statement: "Symbols in an explicit list will not be tracked if they were not evaluated in the enclosing construct:"

Which is fundamental to Tracked symbols and the fact that it is so well hidden is at least surprising.

@Kuba u;DateString[] should have been obvious ... I feel stupid. Thanks :-)

@Szabolcs you are going to Grenoble? what a pity! I went to all the latest French events (I've been doing a nice presentation on a topic that I will soon post on the community), but this one I have to skip... Staid all week at a conference, and I have work to do. But I just booked the hotel for the WTC2016!!

We have been complaining about the same thing for on and on... with little impact. I don't think the impact has been zero. But what we see is very very little. I think that an intervention is in order. Like a closed letter or something. (I mean, not an opened letter, since it is not nice for a commercial company).

@Szabolcs I don't think so, Dynamic ref page says nothing about that, and TrackedSymbols' only mentions this in possible issues. I don't know but for me it was a surprise :)

@P.Fonseca I think I will come there too :) We should organize labels or something.

I imagine that their current target is to launch the next version (11.0) before WTC2016. This is in 7 months. And so, 5 more months of development, and 2 of stabilisation (and since they will probably launch earlier, it means even less time). This means that, most likely, there's not a lot of space for any intervention to be of success on this interval. I also have no idea of what we are talking about in terms of development time.

Are the bugs and design alignments we would typically expect to be corrected a thing of 3, 6, 9 or 12 months? I also have no idea of what is the percentage of the development team that can efficiently correct bugs. That is, a notion that everyone of the development team will be tracking bugs, looks kind of silly. And so, I have no

Nevertheless, I can align with an intervention.

This was a dream software. The dream has been fading away, but a run for the bugs, and, of the same importance, for the design inconsistencies, is still not too difficult; but it is now or never...

@P.Fonseca I'm reasonably confident WRI is aiming for one release per quarter now. Is it bug fix, minor version or major version is another thing.

And that features and fixes either make the deadline, or not. If not, they go to later version.

Don't feel like this warrants its own post, but what kind of colorfunction in mathematica can I use that looks like i.imgur.com/A3br8gx.png? I tried using a ton of them, but the contrast is usually much worse than in this blue black white scheme.

@user3183724 Use Blend to build your own.

Hi @Silvia, have been quite occupied lately. Did you send anything recently?

@user3183724, if you could obtain the actual colormap instead of the plot, it'd be really helpful.

Is there a built-in function or a simpler way to do this: fromChacterName[name_String] := ToExpression["\"\\[" <> name <> "]\""]?

Basically, I'm looking for the inverse function for CharacterName[#, "StandardName"]&.

@VladimirReshetnikov That's how I'd do it.

I'm working on a mouse painting at work.

Time to call it a day.

Curiously, it is possible that two approximate reals are considered identical by SameQ, but their corresponding intervals are not. 1`9.0 === 1`9.7 is True, but Interval[1`9.0] === Interval[1`9.7] is False.