Timeline for Can Mathematica's random number generation be improved?
Current License: CC BY-SA 4.0
18 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 10, 2022 at 11:58 | history | edited | Greg Hurst | CC BY-SA 4.0 |
added 131 characters in body
|
| Feb 20, 2015 at 23:50 | comment | added | ciao | @Mr.Wizard: I get your point, but a "That car is crap and needs fixin' 'cause I crashed" post when it turns out the driver didn't know how to drive is... well, I'll leave it to you to judge. Now the question appears to have changed - nowhere in the OP is the fact that the distribution of the parameters themselves is some undefinable construct. My vote to close stands - this feels more like a marketing rant from a competitor. I shall refrain from commenting further on this post, the hackles are at attention. | |
| Feb 20, 2015 at 23:47 | comment | added | JEP | Mr Wizard, I didn't really want you to go through and read all these links! That is a waste of time. The "brain dead" was rasher's response the first time I raised this issue. Yes. I thought people came here to to help and be helped. I don't know how to make it clearer that I was looking for a solution than posting the solution as I did. blochwave posted a fine solution which I couldn't get to work on my Macs. I imagine my solution will work on any OSX Mac. It received two up votes whereas rasher's solution which does not solve the general problem received 17. Go figure. | |
| Feb 20, 2015 at 23:40 | comment | added | JEP | I'm glad to see that you've added code. THis is not a solution to the general problem in which the input parameters are not parametrically distributed. | |
| Feb 20, 2015 at 23:37 | comment | added | Mr.Wizard | To @rasher and others I ask you to take a step back and consider how you might assist a user to better understand the software, accept its limitations, and participate constructively. Rejoinders like "What's your point? Of course..." sound confrontational even when they are true. Issues that are "of course" understood by experienced users may be new to others. | |
| Feb 20, 2015 at 23:32 | comment | added | Mr.Wizard | @JEP There are plenty of things in any system to complain about; I've certainly pointed out my share of problems. However this community should be about helping each other. We have had people come to this site who seem to be here to do nothing but tear down Mathematica and its users, and I think you struck a raw nerve. It is entirely valid to publicly wonder why Mathematica has certain failings but you must find a way to make it apparent that you are working toward solutions not simply complaining. | |
| Feb 20, 2015 at 23:28 | comment | added | Mr.Wizard | @JEP I fail to see where "brain dead" originated. However calling people stupid, even by implication, is not appropriate. rasher: it should be possible to argue a position without stooping to ad hominem attacks, which you are flirting with above; please do not cross that line. | |
| Feb 20, 2015 at 23:03 | comment | added | JEP | Mr Wizard, I am not confused. If you are going to start chastising me for my tone you ought to start at the beginning to this thread and read the links. I was told I was brain dead, then lazy, then stupid. All for expecting MMA to be able to generate random variates with performance similar to that of other languages. | |
| Feb 20, 2015 at 22:59 | comment | added | ciao | @blochwave: Since the distribution is defined for the generation, that cost is amortized- there's no extra cost generating 10 or 10 Billion variates. Only if the distribution's parameter RVs distributions themselves changed would that come into play, and then there's other ways to skin that cat... so I think it more than "fair" as is, not to mention shows techniques that cannot even be done with purely numeric s/w. Thanks for timing on your box! | |
| Feb 20, 2015 at 22:22 | history | edited | dr.blochwave | CC BY-SA 3.0 |
deleted 2 characters in body
|
| Feb 20, 2015 at 22:21 | comment | added | dr.blochwave | @rasher - Cheers for adding the code, it's very quick - 0.03 seconds on my machine, compared to the 0.15 seconds for C++, although a fairer comparison including the generation of the mixture distribution bumps that up to 0.13 seconds. | |
| Feb 20, 2015 at 22:12 | history | edited | ciao | CC BY-SA 3.0 |
Added example per request
|
| Feb 20, 2015 at 19:44 | comment | added | Mr.Wizard | @JEP You seem to be confused. This site is not run by or directly affiliated with the developers or publishers of the software Mathematica. This is not the place for such complaints, at least not in that tone. Please desist from making them. Constructive participation is expected here. | |
| Feb 20, 2015 at 19:23 | comment | added | JEP | It would be stupid not to complain about a slow buggy $1500 software package that claims about itself: "built to provide industrial-strength capabilities—with robust, efficient algorithms across all areas" & "draws on its algorithmic power—as well as the careful design of the Wolfram Language—to create a system that's uniquely easy to use" & this "With its intuitive English-like function names and coherent design, the Wolfram Language is uniquely easy to read, write, and learn" and which in addition has been unresponsive to bug reports. It is stupid to defend it. | |
| Feb 19, 2015 at 23:32 | comment | added | ciao | @jep Again, what's your point? Mathematica has no automatic detection of stupid coding and coders. If you prefer some other system, use it, complain on their site... | |
| Feb 19, 2015 at 23:23 | comment | added | JEP | The 300x speedup of weibull is impressive. On this problem Mathematica is only 123 times slower than Octave. cat wbl.m t=cputime;num=10^5;n=50*rand(num,1);p=100*rand(num,1);out=wblrnd(n,p);cputime-t $ octave wbl.m ans = 0.018561 | |
| Feb 18, 2015 at 13:07 | comment | added | JEP | An example would have been helpful. | |
| Feb 18, 2015 at 7:43 | history | answered | ciao | CC BY-SA 3.0 |