Why is it so silent in here? Am I the only one working and everyone else is drinking or on a date?

Interpreter does not recongise part of a definition when simplifying Sum: twitter.com/ejenk/status/545761138944659456 pic: t.co/BMZBtLY9SH

I'm sorry but I don't have time to post it right now. I haven't checked thoroughly if the issue is known already, too.

@Akater There is no issue. It does exactly what it should. Maybe this guy just doesn't understand what the semantics of the code he wrote.

I wouldn't even be able to imagine a Sum function intelligent enough to do what ejenk wants if I hadn't already been spoiled by Mathematica. It's certainly possible a future version could break down the summation range based on the available definitions of user-defined functions in the summation, in some cases.

For now though, documentation does clearly state: "If a sum cannot be carried out explicitly by adding up a finite number of terms, Sum will attempt to find a symbolic result. In this case, f is first evaluated symbolically."

So it essentially evaluates f[x] and gets x, and then does Sum[x,{x,m}]

General notice, I'm testing some stuff for a new LinkSnooper which has very basic highlighting and uses the new JavaFX framework. I hope that GUI will run more smoothly than the current one, but I'm not sure whether it really turns out to work as I expected it. First screenshot

At the moment, I'm only testing ListView, which has to to work extremely well with all the traffic that comes of the KernelLink. I would not have come that far without the help of eckig who was patient enough to give as much tips as possible:

He already has my upvote (and one downvote of someone for no obvious reason) and he will get the bounty. Please upvote his answer if you come by!

Apparently Framed is font...

Style[Framed[1], FontSize -> 25, FontColor -> Red]

@halirutan very nice! Is it possible to replace long frequently used names with abbreviations there too?

Strange questions that pop up in my mind on a slow day: if you peel an ideal orange (==sphere) along a rhumb line and flatten the peel on flat surface (as well as you can), what this shape does look like?

@Kuba I have to see what operations on the packets don't need too much time. Some of the packets (eve if you don't create large graphics or other stuff) are very long, e.g. when the FE requests an updated list of kernel symbol contexts, then the answer from the kernel had in my case about 23000 characters.. It is this long

StringTake[ExampleData[{"Text", "FaustI"}], 23000]

With those packets, the SyntaxTokenizer in JLink is too slow and a fluent scrolling is no longer possible...

@Kuba I'll try to use a JFlex scanner and hope that it is faster when I do only a basic tokenization.

@Kuba When you say you want to abbreviate long names, what exactly do you have in mind?

@Kuba On a simple replacement basis this would work, but it would be possible that it replaces this at positions where it shouldn't because we cannot check the syntax without parser.

@Kuba What I often do is copying packet-code back to Mathematica to try it. I think I will make this easier. Otherwise I'm open for suggestions.

@halirutan Ok, I will let you know if anything come to my mind :P

@MichaelHale If asked nicely, MMA will give you correct symbolic sum for this function: Sum[Piecewise[{{1, n == 0}}, n], {n, 0, m}]

@jkuczm Sure, the symbolic functionality tends to work best when used with symbolic functions. I just think that Twitter post is an example of a common symptom from people exposed to Mathematica design (myself included back when I was first learning) to extrapolate the functionality to unrealistic expectations.

Like, what if the condition on your custom definition was to apply the definition only if the input raised to the billionth power had a prime number of 1's in the digit sequence. If they thought about their question like that, then they probably wouldn't expect a symbolic sum to work in that case.

@MichaelHale I agree, it can only work in some special cases, set of those "special cases" covered in MMA is already huge.

Sure, so I guess the proper response to ejenk is to tell him to try to use symbolic math functions to define his input constraints when he's expecting symbolic output.

@MichaelHale I think problem from Twitter post comes from problems in basics: how DownValues work and what's the difference between expression that is zero and expression that equals zero.

I think he was confused because he expected the basic, very flexible constructs like arbitrary downvalues with pattern matching and potential conditions, to work with the symbolic functionality. I think that's a reaction from the design of Mathematica being more consistent and unified than other languages and libraries they've used, so they extrapolate to unrealistic expectations.

Like, anything you can express with Piecewise you can express with conditions/PatternTest but not vice versa. People like how flexible pattern matching is, so they forget that Piecewise limits you to some constraints that play well with symbolic functionality.

@MichaelHale I don't think problem there really is related to Sum, result of f[n]/.n->0 might be as "unexpected".

Perhaps. Then you are right that it would be an issue of skipping over understanding the basics. A refresher on evaluation procedure, FullForm, and Attributes should be enough.

More comments about that tweet here: plus.google.com/u/0/+DanPiponi/posts/fF9xHcxo9Ze

Quoting. Please drop a star below. :D

I finally have the function I've wanted ever since I began using Mathematica:

The first documented time I wanted to do this was back in 2012 :)

@Pickett your own, or a package?