Résultats du concours TI-Planet PalPrem !

[Adriweb]

Indeed, Weregoose, your program respected the rules - I'm not sure wher you see a problem ?
The ones not respecting the rules got penalty points off [for some, severely, depending on what they did].

I believe the difference with Xeda, who got first in your category, was about speed than respecting the rules, as you can see in the graph
You can download her program anyway and look at her code

(BTW, congrats to you too especially for making them very small, while Xeda is a few bytes lighter, it's quite an achievement !)

[Weregoose]

You can download her program anyway and look at her code

I did that before making an account. What you're telling me is that I could have deliberately incurred a penalty for the sake of a faster program and gotten a better overall score. Do you confirm this?

That bit about keeping a small list of primes was not addressed in the rules for judging, so toss that out of the argument. That point is moot; "it destroys the value of algorithmic thinking" to use such a list as a crutch.

As an aside, I find it odd how you mention my program's size when it's actually smaller than Xeda's after you replace the system variables (which are inherently faster) with standard letter variables. Size was not the goal here.

[maestroarte]

J'avais eu aussi l'idée d'un crible, mais trop long pour le début... après c'est sur que ça réduit considérablement le temps de l'algo pour les grands nombres
- je connaissais pas par contre l'algo de miller probabiliste : bonne idée
- c'est qui le p008?


merci!

[Lionel Debroux]

Weregoose: if you had used a lengthy static list of primes (or equivalent) aimed at returning the result immediately for low inputs, you could indeed have incurred a small judgement penalty. But we duly indicated that small static lists of primes (say, the 4 primes < 10), which includes equivalent small if chains in my mind, were tolerated, at least an initialization. For instance, I don't think I would have given a penalty to the TI-BASIC equivalents of, say,

Code: Select all

if (n == 1) { return 2; }
if (n < 5) { return n/2+1; }

or

Code: Select all

if (n < 5) { n = n/2+1; if (n < 2) { n++; } return n; }

The raw size of Xeda's program and your program is mentioned as an aside note, but didn't play a role in the ranking. What explains the relative rankings of PZ02 and PZ03 (as well as PP01 / PP06), is that the item with highest weight (by far) in the ranking was sheer speed. An indication about the ranking criteria was posted directly in the contest announcement.
Both of your programs work, both respect the rules and both have explanations. The differentiating factor is on speed: your program is consistently slower on average than Xeda's by a factor between 2 and 3 for n <= 100, though the n=250 figures show that your program is likely to be able to surpass Xeda's program at some point (unless the TI-BASIC interpreter plays yet another nasty trick), thanks to an asymptotically better algorithm.
Jacobly's ASM program can chew through n=3000 in less time than Xeda's and your program can chew through n=250.

We know who you are and the fantastic tricks you've pulled for years in TI-BASIC. In fact, when we saw you entered, we thought "OK, we probably have the winner here". But surprisingly, not this time. You made a program whose speed is reasonably close to that of the winner; the three other contestants in your category didn't.

Does the contents of my post seem relevant to you and provide enough insight into the ranking ?


maestroarte: PP08, c'est Excale, voir le premier post
Sa participation était bien sûr hors concours, mais très intéressante.

[Weregoose]

Cheers.

[Levak]

- c'est qui le p008?

Ta question devrais plutôt être "c'est quoi le résultat de P008?", non ?
Car d'après la News c'est Excale, mais il n'apparaît pas sur le graphique. En fait il a essayé de troller le concours just for fun mais a été attrapé

[The K]

Congratz to winners (and everyone who participated too !).

[Lepzulnag]

Bravo Bisam, Excale et les autres gagnants !

Finalement, ce qui est amusant c'est que d'après les corrections ce qui départage les meilleurs, c'est juste le moyen de générer les nombres premiers ou vérifier qu'un nombre est premier. L'aspect palindrome a été correctement géré par beaucoup de monde. D'ailleurs je ne pensais pas que le crible d’Ératosthène serait plus rapide qu'un algorithme récent !

Enfin la production d'Excale m'a bien fait rire Mettre tout son talent dans une farce, c'est une philosophie que j'apprécie J'ai regardé son code en sachant que quelque chose était spécial, et au début je me suis dit : "Mais... comment ?". L'algorithme était exactement le mien à une ou deux différences près. Finalement, en espionnant le chat hier soir et grâce à la curiosité de Nikitouzz j'ai compris. Bien joué Excale !

(A ce propos le classeur d'Excale ne marche pas sur mon émulateur, qui cependant n'est plus à jour.)

[critor]

Excale a caché du code assembleur dans son classeur tns.

Donc ça ne marche pas sous le logiciel Nspire qui ne reproduit pas le comportement de l'architecture ARM.

Il te faut l'émulateur communautaire ou une calculatrice avec l'OS 3.1 et Ndless.

[nikitouzz]

Moi et curiosité je suis impatient d'avoir la news