On 12/11/2012 12:36 PM, daniel rupis wrote:
To balance my (a little trolling post), I must say that racket community is
very nice. Thanks to all that respond to the post.
For my part, you're welcome. :)
I am interested in math, and I follow for example maxima, julia and R.
It should be
On Tue, Dec 11, 2012 at 05:56:03PM +0100, Pierpaolo Bernardi wrote:
> On Tue, Dec 11, 2012 at 5:11 PM, Stephen Bloch wrote:
>
>
> > How many primes are below ten million? A hundred million? At some
> > point storing the primes will take less memory than storing
> > primality flags, but that
2012-12-12 Boston Lisp Meeting: Marc Battyani, Alex Plotnick
http://fare.livejournal.com/170783.html
When? TOMORROW, Wednesday December 12th 2012 at 6pm.
Where? MIT 32-D463 (Star conference room at the Stata Center).
Who Speaks? Marc Battyani will showcase his web framework written in Common
L
To balance my (a little trolling post), I must say that racket community is
very nice. Thanks to all that respond to the post.
I am interested in math, and I follow for example maxima, julia and R.
It should be wonderful if the new math library could be integrated in any way
with the symboli
2012/12/11 Jens Axel Søgaard :
I have just pushed an improvement for prime? on
small numbers.
Before:
>> (time (for/sum ([n (in-range 100)]
>> #:when (prime? n))
>> n))
>>
>> ; cpu time: 8944 real time: 9079 gc time: 257
>> ; 37550402023
Now:
cpu time: 1511 real time:
On 12/11/2012 08:03 AM, daniel rupis wrote:
I should say that I like racket, but I find macros in racket rather difficult.
I can use macros in common-lisp but I still can't use racket macros. (I am
trying to say that perhaps macros in racket are something difficult to grasp).
Yeah, dealing wi
On 12/11/2012 09:59 AM, Sam Tobin-Hochstadt wrote:
On Tue, Dec 11, 2012 at 11:43 AM, daniel rupis
I am using racket from the console, not using DrRacket. I just copy the
code
with control-c and paste with control-v then wait a seconds for the definitions
to be loaded in memory and then run
At Tue, 11 Dec 2012 16:43:25 + (UTC),
daniel rupis wrote:
> Anyway, my point was that I was expecting something more from typed racket.
> Since typed racket use types (like declaring type in sbcl) I was expected
> better
> perfomance, that's all.
There's a big difference between Typed Racket
2012/12/11 Phil Bewig :
> I can't type. The correct count of primes less than a million is 78498. And
> there is something wrong if (nth-prime 78498) returns 13, as the 78498th
> prime is 83.
That's because nth-prime counts from 0 and Mathematica counts from 1.
--
Jens Axel Søgaard
On Tue, Dec 11, 2012 at 11:43 AM, daniel rupis
wrote:
> Sam Tobin-Hochstadt writes:
>
>>
>> While you're certainly right about DrRacket introducing noise in
>> performance measurement, I don't think you need to generate an
>> executable to eliminate that overhead. Simply running `racket` from
>>
On Tue, Dec 11, 2012 at 5:11 PM, Stephen Bloch wrote:
> How many primes are below ten million? A hundred million? At some point
> storing the primes will take less memory than storing primality flags, but
> that point may be above the size of tables we can realistically store today.
> (for
I can't type. The correct count of primes less than a million is 78498. And
there is something wrong if (nth-prime 78498) returns 13, as the
78498th prime is 83.
On Tue, Dec 11, 2012 at 10:46 AM, Martin Neal wrote:
> I clicked the link, and the result shows 78498 which jives with nth-pri
I clicked the link, and the result shows 78498 which jives with nth-prime
On Tue, Dec 11, 2012 at 8:31 AM, Phil Bewig wrote:
> There must be an error in the prime-counting function. According to href="
> http://www.wolframalpha.com/input/?i=how+many+primes+less+than+a+million";>Wolfram|Alpha,
>
Sam Tobin-Hochstadt writes:
>
> While you're certainly right about DrRacket introducing noise in
> performance measurement, I don't think you need to generate an
> executable to eliminate that overhead. Simply running `racket` from
> the command line on a file in a module ought to be sufficient
On Tue, Dec 11, 2012 at 5:28 PM, Sam Tobin-Hochstadt wrote:
>> This looks like you are testing from inside DrRacket. As I already
>> wrote you should generate an executable and measure that.
>>
>> If you run the test from inside DrRacket, you are measuring DrRacket
>> overhead, which is far from
Sorry my mistake , it is 6.8 seconds. :) .
Veer
On Tue, Dec 11, 2012 at 9:52 PM, daniel rupis
wrote:
> Veer Singh writes:
>
>>
>> I am getting 6.8 ms without modifying the code.
>> When I change modulo to remainder I get 6.3 ms consistently.
>
>
> I think you mean 6.8 seconds not milliseconds
There must be an error in the prime-counting function. According to http://www.wolframalpha.com/input/?i=how+many+primes+less+than+a+million";>Wolfram|Alpha,
there are 79486 primes less than a million, not 78497.
I don't use Racket, but I do have lots of Scheme code that computes with
prime number
On Tue, Dec 11, 2012 at 11:23 AM, Pierpaolo Bernardi
wrote:
> On Tue, Dec 11, 2012 at 5:08 PM, daniel rupis
> wrote:
>> Pierpaolo Bernardi writes:
>
>>> Remember to generate an executable, to obtain the maximum speed.
>
>
>> Welcome to Racket v5.3.1.
>>
>>
>> (define (test)
>> (time (displayln
On Tue, Dec 11, 2012 at 5:23 PM, Pierpaolo Bernardi wrote:
> If you run the test from inside DrRacket, you are measuring DrRacket
> overhead, which is far from negligible.
As an example, the test I just sent, with the limit changed to 60999,
on my machine runs in 20083 ms from inside DrRacket, w
On Tue, Dec 11, 2012 at 5:08 PM, daniel rupis
wrote:
> Pierpaolo Bernardi writes:
>> Remember to generate an executable, to obtain the maximum speed.
> Welcome to Racket v5.3.1.
>
>
> (define (test)
> (time (displayln (total-primes 60999> > > >
>> (test)
> 6145
> cpu time: 39170 real tim
Veer Singh writes:
>
> I am getting 6.8 ms without modifying the code.
> When I change modulo to remainder I get 6.3 ms consistently.
I think you mean 6.8 seconds not milliseconds.
Try with the bigger n to remove transient behaviour.
Racket Users list:
http://lists
Yesterday, Matthew Flatt wrote:
> I've run into this, too, and I don't have a better solution right now.
>
> It's not clear to me what happens with fancy argument specifications
> when the function description is used for a callback, but I've never
> gotten around to sorting it out.
Sounds like w
Pierpaolo Bernardi writes:
>
> On Tue, Dec 11, 2012 at 4:03 PM, daniel rupis
> wrote:
> >
> > I was comparing some code in Qi with that of sbcl, I posted a question in
> > comp.lang.lisp asking for a way to improve the perfomance, WJ gave a typed
> > racket version that was slower than sbcl an
On Dec 11, 2012, at 9:03 AM, Jens Axel Søgaard wrote:
> 2012/12/11 Stephen Bloch :
>
>> Would it perhaps make more sense for small-primes to contain primes
>> themselves, in increasing order so one can be found by binary search, rather
>> than booleans? The O(1) behavior would be replaced by O(
I am getting 6.8 ms without modifying the code.
When I change modulo to remainder I get 6.3 ms consistently.
Veer.
On Tue, Dec 11, 2012 at 8:33 PM, daniel rupis
wrote:
>
> I was comparing some code in Qi with that of sbcl, I posted a question in
> comp.lang.lisp asking for a way to improve the
On Tue, Dec 11, 2012 at 4:03 PM, daniel rupis
wrote:
>
> I was comparing some code in Qi with that of sbcl, I posted a question in
> comp.lang.lisp asking for a way to improve the perfomance, WJ gave a typed
> racket version that was slower than sbcl and also much slower than cpp.
The sbcl versi
I was comparing some code in Qi with that of sbcl, I posted a question in
comp.lang.lisp asking for a way to improve the perfomance, WJ gave a typed
racket version that was slower than sbcl and also much slower than cpp.
Daniel Rupis wrote:
Note: The code compute the number of primes below 3
2012/12/11 Stephen Bloch :
> Would it perhaps make more sense for small-primes to contain primes
> themselves, in increasing order so one can be found by binary search, rather
> than booleans? The O(1) behavior would be replaced by O(log(limit)), but
> perhaps you would save enough memory to put
Unfortunately, Maxima seems to be distributed under the GPL, which
means that we can't easily just take their code and add it to Racket.
Sam
On Tue, Dec 11, 2012 at 8:52 AM, Jens Axel Søgaard
wrote:
> 2012/12/11 Jens Axel Søgaard :
>> 2012/12/11 Jens Axel Søgaard :
>>> Here is what Maxima does (
2012/12/11 Jens Axel Søgaard :
> 2012/12/11 Jens Axel Søgaard :
>> Here is what Maxima does (line 628 and below):
>>
>> https://github.com/andrejv/maxima/blob/master/src/ifactor.lisp
>
> I have attached a port in the attached file. The new primality
> tester is called new-prime? and the old one pri
On Tue, Dec 11, 2012 at 6:01 AM, Jens Axel Søgaard
wrote:
>
> The second will be to experiment with "medium numbers"
> and see whether trial division is faster than the pseudo
> prime test, when the small prime limit is raised to larger number
> again.
One thing that might be worth considering, e
2012/12/11 Stephen Bloch :
> On Dec 11, 2012, at 6:01 AM, Jens Axel Søgaard wrote:
> Would it perhaps make more sense for small-primes to contain primes
> themselves, in increasing order so one can be found by binary search, rather
> than booleans? The O(1) behavior would be replaced by O(log(lim
On Dec 11, 2012, at 6:01 AM, Jens Axel Søgaard wrote:
> For a small (less than 1) number n prime? looks up primality of
> with a simple (vector-ref small-primes n).
>
> For a large number (greater than 1) a pseudo prime test is used.
>
> Originally the limit for small primes was a mil
2012/12/11 Pierpaolo Bernardi :
> On Tue, Dec 11, 2012 at 12:01 PM, Jens Axel Søgaard
> wrote:
>
>> Below 60 there are only 16 numbers not divisible by 2, 3 or 5.
>> The number of bytes needed to store primality results for the
>> first million numbers will therefore be
>> 100 * 16/60 *
On Tue, Dec 11, 2012 at 12:01 PM, Jens Axel Søgaard
wrote:
> Below 60 there are only 16 numbers not divisible by 2, 3 or 5.
> The number of bytes needed to store primality results for the
> first million numbers will therefore be
> 100 * 16/60 * 1/8 ~ 3.3
>
> How much space would b
FWIW, I think these questions are best answered with an app in hand,
but my apriori guess would be to say that 32k isn't too bad, but that
it also doesn't seem necessary in this case. Martin Neal's micro
benchmark already saw lots of improvements with just a different
algorithm so perhaps that's en
2012/12/11 Martin Neal :
> First - I love the library. There's so much functionality that it's going
> to take quite some time to explore it all
>
> I'd love to see an `in-primes` function added to math/number-theory.
Excellent idea. I have added it to the todo-list.
> Using a pseudo prime test
37 matches
Mail list logo
