# Mathematica self-evaluation: let's get critical! [closed]

You love your site and we love your site, but there is a whole world of people out there who might not even know it exists. When they do find it, their first impression will either scare them away or keep them around. Given this, let's take a hard look at the questions and answers here and make sure newcomers see the site at its best!

Below you'll find ten questions randomly selected from this site. What do you think about each of them and their answers? Are they the best they can be or can they be improved? Would they look interesting and inviting to an outsider or are they a little embarrassing?

Upvote the corresponding post here on meta when we're awesome. Downvote when our content just isn't quite up to par.

Oh, and do comment to let everyone know your thoughts and take part in this conversation. :)

## closed as too localized by Adam Lear♦May 31 '12 at 23:45

This question is unlikely to help any future visitors; it is only relevant to a small geographic area, a specific moment in time, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet. For help making this question more broadly applicable, visit the help center. If this question can be reworded to fit the rules in the help center, please edit the question.

• I notice that all of these questions fall in the period March 29 through April 8, where three of them were asked on April 3 alone, and, excluding April 8, the period sampled is less than a week and includes a weekend. While there's arguably nothing wrong with this selection, there are some truly great questions/answers on this site, but they aren't that localized in time--and of course people arriving here from search engines typically will not confine their query to a narrow range of dates. I wonder, is the temporal locality just a coincidence, or could there be a sampling bias here? – Oleksandr R. May 25 '12 at 2:35
• Don't forget to turn off personal results in Google! It tends to rank hits from this site higher if I have them on (the default). – Szabolcs May 25 '12 at 7:17
• @OleksandrR. We pull questions from 30-40 days ago to make sure they have had a chance to be answered, edited, vetted by the community, etc. There will be more evaluations like this in the future, so it should all balance out over time. – Adam Lear May 25 '12 at 14:34
• @Anna Will there be a feedback from you after this evaluation is complete? Also, it is unavoidable that these questions/answers will be polished and fixed up simply because you posted a link to them. Is this okay? – Szabolcs May 25 '12 at 14:48
• @Szabolcs The main goal of these is for you guys to self-organize around reviewing your Q&A and identifying areas for improvement. So yes, polishing and fixing up questions posted here is very much okay. Far as feedback goes, I'll be looking at your findings, but I might not have much to say. Y'all are the experts. :) It's only been a few hours, but I'm really happy with the response and the involvement I'm seeing here so far. You guys are doing great and showing us that there's a very engaged, passionate community here. If you have any questions for me, feel free to ask. – Adam Lear May 25 '12 at 15:07
• If there's anything that stands out (like some question type that may be problematic in our format), I can offer guidance on that, but so far my impressions match the ones here so far. – Adam Lear May 25 '12 at 15:10

How to apply or map a list of functions to a list of data?

What do you think about this question and its answers? Vote and comment to let everyone know.

• Interesting question, with a wide variety of solutions. (Looks like Istvan is close to a Populist badge on this one.) – Brett Champion May 24 '12 at 21:54
• A highly voted question with 7 answers. It is the fifth hit on Google for map functions to data mathematica, and the first relevant one. The answers neatly showcase the many ways one can do things in Mathematica. It is not a newbie question: mapping one function to multiple bits of data is a standard operation; mapping p functions over q bits of data (where p<q) is not obvious. Others will find this useful. – Verbeia May 24 '12 at 22:26
• The problem is quite specialized: it's unlikely that there'd be another solution exactly to this problem on the internet. My search terms ("mathematica cyclically apply functions") return the same Mma.SE question on the first page. He was asking for alternative ('better') solutions, and got 7 different ones. Great answers, and a "classical", list-manipulation type Mathematica question that everyone loves to answer. – Szabolcs May 25 '12 at 7:30
• And this question was a perfect example that in Mathematica anything can be solved in a multitude of ways. I myself am amazed by the variety of all the different solutions. This thread shows that usually there is plenty of room for further answers, encouraging newcomers to try to contribute. – István Zachar May 25 '12 at 10:54

Numerical underflow for a scaled error function

What do you think about this question and its answers? Vote and comment to let everyone know.

• This is a good example of a less-than-stellar question inspiring multiple exemplary answers. The answers involved a fair bit of mathematical sophistication and clearly explain some numerical precision issues that have more general implications. It is the sixth Google hit for underflow mathematica, which is again remarkable. – Verbeia May 24 '12 at 22:33
• I'm quite proud of the two answers I gave there, even if they're not that popular (with respect to votes). Many people are ignorant of techniques for numerically computing special functions, and I got to show off two different methods there. – J. M. is away May 24 '12 at 23:10
• +1 because there are advices both on handling large cancelling terms generally, and approximation methods specifically for this function. What I learned from Google that's not explicitly mentioned in the thread is that this is a commonly occurring special function and there are several known approximation methods for it and for related functions: one, two (My search terms: erfc exp.) – Szabolcs May 25 '12 at 8:46
• MathGroup mention with no explicit solution, I didn't manage to find it in Abramowitz and Stegun, though there are hints online that it should be there. – Szabolcs May 25 '12 at 8:49
• @Szabolcs: the information is indeed in the Handbook, albeit somewhat concealed. 7.1.13 is the inequality Artes exploited in his answer, 7.1.23 is the asymptotic series used in F'x's answer, 7.1.14 is the extension of the CF in one of my answers (conversely put, the CF in my answer is the even part of the CF in 7.1.14; the virtue of the CF in my answer is that it converges twice as fast as the CF in the Handbook.) Additionally, 7.1.25 and 7.1.26 are simple approximations that can be rearranged into an approximation for the function needed in the question. – J. M. is away May 25 '12 at 9:09
• P.S. I replaced the JStor link in Szabolcs's comment with an AMS link; back issues of Mathematics of Computation are freely available from the AMS site, and there is almost no need to link to the JStor versions. – J. M. is away May 25 '12 at 9:12
• For my small part, the responses to this question showcase the best of this community and the uniqueness of this site. Rather than just a programming "how to..." at its best this site brings together broad expertise across multiple analytic disciplines. The participants on the site supply an almost encyclopedic set of ideas, many entirely original, that provide solutions and insights for attacking a very broad range of problems. Those of us with lots to learn get to witness real science and mathematics getting done. Very few forums anywhere give such a glimpse into work of this calibre. – Jagra May 31 '12 at 14:38

Finding real roots of negative numbers (for example, $\sqrt[3]{-8}$)

What do you think about this question and its answers? Vote and comment to let everyone know.

• This is a question that comes up fairly often, especially by beginning users of systems like Mathematica. And Artes' answer is quite comprehensive. – Brett Champion May 24 '12 at 21:46

Memory Leak in Frontend - anyone know a workaround?

What do you think about this question and its answers? Vote and comment to let everyone know.

• A sophisticated question involving a subtle memory leak. The question is clear. There are three good answers, two from WRI staff, providing non-obvious workarounds. It is the sort of problem that people might come across again and spend a long time trying find a solution. This question is the top Google hit for mathematica memory leak graphics and mathematica memory leak rasterize, and the sixth hit for just mathematica memory leak. – Verbeia May 24 '12 at 22:12

How to generate a real-time stream of data?

What do you think about this question and its answers? Vote and comment to let everyone know.

• The question was clear and dealt with some sophisticated functionality. The answer is likewise clear, and while it is mainly a discussion of what is in the documentation, the functionality described is not well understood by many users and would not have been obvious to most users as the thing to look for in order to solve the problem described in the question. – Verbeia May 24 '12 at 22:39
• This is a perfect example of introducing new functionality that the OP did not know about and others, like myself, may have missed between versions. – rcollyer May 25 '12 at 2:41

How to determine the convex hull of some text?

What do you think about this question and its answers? Vote and comment to let everyone know.

• I think the question is a bit weak; some background of what they're going to do with the convex hull would have been nice. I'm a bit surprised nobody turned the text into filled curves and tried to work with that. (Actually, that might be a bit trickier than I thought at first.) – Brett Champion May 24 '12 at 21:39
• The question is a bit terse; I agree with @BrettChampion. But the answers are clear and show two different ways to do it (Using ComputationalGeometry or not). The question is currently seventh Google hit for convex hull mathematica, which is surprising given how general that search term could be. – Verbeia May 24 '12 at 22:30
• @BrettChampion: there are a few questions dealing with the vectorization of text (not that I can find any of these ATM...), and I think the general consensus in the community is that everyone trustingly waits for Mathematica 9 to include an integrated text-to-polygon solution. – István Zachar May 25 '12 at 11:11

Autorotating 3D plots

What do you think about this question and its answers? Vote and comment to let everyone know.

• This is a cool graphics question, which are very popular on this site. It is the eighth hit for rotating 3d plots mathematica on Google. Both the question and answer are well explained. The answerer mentioned in comments that they would have liked something even more automated, but the answer does explain a relatively obscure feature (the difference between ViewPoint and ViewVector) very well. – Verbeia May 24 '12 at 22:21

Facegrids at ticks

What do you think about this question and its answers? Vote and comment to let everyone know.

• This question is the sixth hit and top non-Wolfram hit for mathematica facegrids. The question is clear. The answer is mainly drawn from the documentation, but is a bit clearer. The answerer found it useful. (@Sjoerd, perhaps it could be expanded on?) – Verbeia May 24 '12 at 22:17
• @verbeia The description in the documentation was rather unclear, so I originally added a full working example which I thought was self-explaining. On your request I elaborated a bit on that. My opinion about the question: it's almost a RTFM type, but, given the phrasing in the FM it's understandable. People facing the same problem should be able to find this answer easily. – Sjoerd C. de Vries May 25 '12 at 7:41
• When I search for "mathematica custom facegrid", I get the StackOverflow version of the same question by the same OP only, which is closed and unanswered. For these keywords I don't get out version. I flagged the SO version to be moved here, to provide another Google entry point. – Szabolcs May 25 '12 at 8:08
• Otherwise the answer is very clear and beginner-friendly, on par with other answers one can find online. – Szabolcs May 25 '12 at 8:11

Unsaved notebook is crashing: is there any way I can save the state of the MathKernel before I kill it?

What do you think about this question and its answers? Vote and comment to let everyone know.

• This is a real problem, concisely described. The answers aren't perfect solutions but they helped the asker. As a side bonus, one answer included the TextRecognize function, which was news to me. +1 for the joyous hackery of that. – Verbeia May 24 '12 at 22:36
• The question is: can I disconnect the kernel from the frozen front end, reconnect it to something else (another kernel?), and extract the data. None of the answers solve the problem, but I can't Google up a better answer either. I have some doubts if it's possible to do this at all. I'm not voting up or down. – Szabolcs May 25 '12 at 7:40
• I second @Szabolcs opinion that I thought a real solution wasn't possible, but I really hoped to be proven wrong here. So, good question, but unfortunately impossible to answer. Overall: small + – Sjoerd C. de Vries May 25 '12 at 7:46

What are the differences between the “Home Edition” and the regular Mathematica?

What do you think about this question and its answers? Vote and comment to let everyone know.

• Googling difference between mathematica home and professional editions brings up this answer as the fifth hit (an older question on SuperUser is first). The answer is a link to the fourth Google hit. So while useful, it's probably not the most exemplary question. The answer is, however, useful and concise, going beyond the link to the documentation and extracting the relevant information. Some of the comments are useful clarifications. – Verbeia May 24 '12 at 21:58
• In addition to what @Verbeia said, the comments do add useful real-world information and clarify whether the kernel is 32-bit or 64-bit. in SuperUser, there is an incorrect comment on this. – acl May 24 '12 at 23:50
• @acl I think at some point the home edition did use to be 32-bit only. In the revision history for 8.0.1 you'll find "Improved functionality and stability of Home Edition by including more 64-bit components" (also a hint on why they might have lifted the 32-bit only restriction). It's a fine answer, but the whole thread (the comments) create the impression that people are a bit unsure about the exact differences. – Szabolcs May 25 '12 at 7:58
• @Szabolcs Yes, it does create that impression but at least what is stated isn't false... – acl May 25 '12 at 8:31
• @acl I forgot to mention that I consider our version a good resource and I upvoted this. I am simply trying to research all these questions using Google and comment about what I find (in this case I found something more about the 32/64-bit question) – Szabolcs May 25 '12 at 8:35
• @Szabolcs You are correct about the home edition being 32 bit in the past (I remember mine was at some point), by the way. I think this demonstrates one advantage of this site over google: if the information was out of date, there's ways to fix it. – acl May 25 '12 at 8:58
• I really did not care for this question. The accepted answer is a quote directly from the FAQ, so I don't think it adds much. I did not downvote it, as I only rarely do so, but I think it was a poor fit, overall. – rcollyer May 27 '12 at 0:42
• I downvoted because I think this question could have a much better answer. Although Home Edition is somewhat fluid from version to version (witness the change from 32- to 64-bit.) – Brett Champion May 28 '12 at 1:28
• Looking at this question, the extra information that's not easily found elsewhere (e.g. that the HE is indeed 64-bit), is in the comments, not the answer itself. So what we can learn from it is perhaps: make sure all important information is included in the answers. More generally: on this site we tend to comment a lot. If all comments were to disappear, the site would lose a lot. – Szabolcs May 28 '12 at 16:34