please rank me as a setter

Everything about Sudoku that doesn't fit in one of the other sections

please rank me as a setter

Postby qwijibow » Mon Jan 23, 2006 3:59 pm

A few days ago, i asked you to give me feedback as a setter for a few sample sudoku problems i posted... i believe i was rated as useless, as my attempts to generate very difficult sudoku problems had more than one sollution.

I believe i have now fixed my short-commings.

I thought of a very fast, always unique method of writing sudoku problems of high difficulty.

My Aim here, was to create a sudoku puzzle that were as difficult as possable, while having only one possable solution.

Last time, my failing as a setter, was not beeing able to create extremely difficult puzzles with only one solution.

So please dont give me negative feedback if the puzzle is too difficult.

i would like to know please...
How difficult are these puzzles, if you attempt one of them, how long did it take you ?
and what techniques did you need to employ to solve the puzzle ?

if you attempted and failed, even better, le me know which one you attempted, and what your skill level is

here are a few that i wrote myself.

Code: Select all
| 1 | 4 |93 |
|249|   | 7 |
|   |7  |   |
-------------
|1 6| 5 |   |
|   |9  |   |
|958|   |74 |
-------------
|  1|43 | 9 |
|5  |  9|  1|
|   |  6|3 2|


Code: Select all
|   |864|52 |
|   |1  |8  |
|6  |3  |419|
-------------
|   |4 2|6 3|
| 5 |   |1 4|
| 8 |9  |   |
-------------
|  3|   |   |
|   | 3 |   |
|  9|  8| 45|


Code: Select all
|7  |  2|  5|
|  2|1  |   |
|4  |79 |1  |
-------------
|2  |3  |   |
|   | 85|   |
|18 |  9|  7|
-------------
|5 8| 4 |  1|
|  6|   |2  |
|   |  6|859|


I appologise if the formatting is bad:(
qwijibow
 
Posts: 6
Joined: 19 January 2006

Postby Carcul » Mon Jan 23, 2006 4:22 pm

Hi Qwijibow.

I have solved these puzzles, and here are my opinions (most advanced techniques used between brackets; order of puzzles from top to bottom in the above post):

Puzzle 1: easy (almost unique rectangle)

Puzzle 2: very easy (X-Wing).

Puzzle 3: extremely easy (naked pairs).

If you want to generate really hard puzzles, I think you have some work to do.:D

Regards, Carcul
Last edited by Carcul on Mon Jan 23, 2006 12:48 pm, edited 1 time in total.
Carcul
 
Posts: 724
Joined: 04 November 2005

Postby tarek » Mon Jan 23, 2006 4:44 pm

Hi Qwijibow,

I think your first puzzle is a good effort, superior to most newspaper puzzles (but not that difficult for members of this forum), it also lacks variety.

The other 2 were extremely very easy (needing only single elimination)

The difficulty gap between your 1st puzzle & what you would describe as the most difficult possible is HUGE.

Tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

Postby MCC » Mon Jan 23, 2006 5:01 pm

The pappocom rating for these are:

Puzzle 1 - V.hard
Puzzles 2 & 3 - Medium
MCC
 
Posts: 1275
Joined: 08 June 2005

Re: please rank me as a setter

Postby gsf » Mon Jan 23, 2006 5:08 pm

qwijibow wrote:A few days ago, i asked you to give me feedback as a setter for a few sample sudoku problems i posted... i believe i was rated as useless, as my attempts to generate very difficult sudoku problems had more than one sollution.

I believe i have now fixed my short-commings.

I thought of a very fast, always unique method of writing sudoku problems of high difficulty.

I think you may be going about this backwards (using the forum to rate your puzzles)
first write a puzzle rating program
then check that rating program against puzzles already posted and rated on the forums
then add the rating program to your puzzle generator which should have a random component
then run the generator and drop the easy puzzles

now you have a framework for tweaking both your rating component
and your puzzle generator -- over time and enough hard puzzles you
may be able to modify the generator to get more hard hits

note that some of the forum ratings (top95 etc.) are based on backtrack behavior
which is totally different from a human solver friendly techniques-required-to-solve rating
gsf
2014 Supporter
 
Posts: 7306
Joined: 21 September 2005
Location: NJ USA

Postby qwijibow » Mon Jan 23, 2006 7:39 pm

I think you may be going about this backwards (using the forum to rate your puzzles)
first write a puzzle rating program


good idea, but unfortunatly one of my weak points.

my AI chess programs in university were never very good at spotting which cobinations of moves were obviouse, and non obviose to human opponents.

i had no such idea puzzle ating programs could have a decent degree of accuracy. I will google for it.

The pappocom rating for these are:

Puzzle 1 - V.hard


Puzzle 1: easy (almost unique rectangle)


This is what puts me off puzzle rating programs.

Either the pappocom rating program is very inacurate, or Carcul is boasting.

Again.. thanks for the feedback, still a long way to go, but a good step us from useless.

One thing that is maybe holding me back a little, is my reluctance to read other setting algorithms. I am attempting to make this progect as much my own work as passable.

im quite pleased to find that my method for finding a random completed grid is very differant from others i've read here.
qwijibow
 
Posts: 6
Joined: 19 January 2006

Postby PaulIQ164 » Mon Jan 23, 2006 7:56 pm

The main discrepancy is caused by differnet people's standards as to what is a difficult sudoku. Most people on this forum aren't happy unless their puzzles have a forcing chain, several zyxpqr-wings and several aquariums' worth of fish. But for the rest of the world, Pappocom V.Hards are about as hard as we like them. So your method appears to have generated (at least) one very difficult (by most people's standards) puzzle, which I'd say is a pretty impressive feat (depending on whether it was more by luck or judgement), though a lot of people on this forum, as we have seen, would consider it very easy (I'd suggest because they use auto-candidate programs - I don't think anyone really finds Pappocom V.Hards 'easy'.)
PaulIQ164
 
Posts: 533
Joined: 16 July 2005

Postby Carcul » Mon Jan 23, 2006 8:28 pm

Qwijibow wrote:Either the pappocom rating program is very inacurate, or Carcul is boasting.


I am not boasting. In puzzle 1, after the basic steps I only needed about 15 seconds to spot a pattern that solved the puzzle: so, it is reasonable for me to rate this grid as easy don't you think? The fact that other people might find this puzzle harder is another question, but I make myself very clear in my reply as I was giving a personal opinion (as you have asked) and not an "universal truth". Rating a puzzle, as I have already stated in another thread, is a very personal question. Perhaps the problem is in the fact that you might not have a clear idea on what is really a puzzle "as difficult as possible".

Regards, Carcul
Carcul
 
Posts: 724
Joined: 04 November 2005

Postby evert » Mon Jan 23, 2006 11:29 pm

puzzle 2 & 3 only singles
puzzle 1 from some point Simple Sudoku has no hint anymore
evert
 
Posts: 187
Joined: 26 August 2005

Postby tso » Tue Jan 24, 2006 1:31 am

qwijibow#1
Code: Select all
+-------+-------+-------+
| . 1 . | . 4 . | 9 3 . |
| 2 4 9 | . . . | . 7 . |
| . . . | 7 . . | . . . |
+-------+-------+-------+
| 1 . 6 | . 5 . | . . . |
| . . . | 9 . . | . . . |
| 9 5 8 | . . . | 7 4 . |
+-------+-------+-------+
| . . 1 | 4 3 . | . 9 . |
| 5 . . | . . 9 | . . 1 |
| . . . | . . 6 | 3 . 2 |
+-------+-------+-------+

Pappocom rates this as “Very Hard, but arguably unfair”, which is actually a distinct level beyond “Very Hard”. Pappocom can solve this puzzle, but will not create puzzles this hard. Actually, I’m surprised it validated a puzzle that appears to require a Nishio or similar tactic to solve -- I would have guessed it would have been rated “Invalid”.

Simple Sudoku does not rate. It found one naked pair -- but it was unable to finish the puzzle.

Sudoku Susser does not rate, but required these tactics to solve:
1 x Nishio
2 x Intersection Removal
1 x Simple Naked Sets
13 x Pinned Squares
Susser did not find a unique rectangle -- nor did I. The Nishio is pretty straight forward.

Sadman Sudoku rates this Diabolical and in fact cannot solve without trial and error. (Or “8” on an eight point scale.)

Paul’s Pages rated this “Outlaw” or 5 on a 5 point scale.

If you're used to looking for Nishio's, this puzzle will not be too tough as there is only one sticking point.


qwijibow#2
Code: Select all
+-------+-------+-------+
| . . . | 8 6 4 | 5 2 . |
| . . . | 1 . . | 8 . . |
| 6 . . | 3 . . | 4 1 9 |
+-------+-------+-------+
| . . . | 4 . 2 | 6 . 3 |
| . 5 . | . . . | 1 . 4 |
| . 8 . | 9 . . | . . . |
+-------+-------+-------+
| . . 3 | . . . | . . . |
| . . . | . 3 . | . . . |
| . . 9 | . . 8 | . 4 5 |
+-------+-------+-------+

Pappocom rates this as “Medium”.
Simple Sudoku required only singles.
Sudoku Susser: 11 x Pinned Squares
Sadman Sudoku rates this “Easy”. (Or “2” on an eight point scale.)
Paul’s Pages rated this “Hard” or 3 on a 5 point scale.


qwijibow#3
Code: Select all
+-------+-------+-------+
| 7 . . | . . 2 | . . 5 |
| . . 2 | 1 . . | . . . |
| 4 . . | 7 9 . | 1 . . |
+-------+-------+-------+
| 2 . . | 3 . . | . . . |
| . . . | . 8 5 | . . . |
| 1 8 . | . . 9 | . . 7 |
+-------+-------+-------+
| 5 . 8 | . 4 . | . . 1 |
| . . 6 | . . . | 2 . . |
| . . . | . . 6 | 8 5 9 |
+-------+-------+-------+


Pappocom rates this as “Medium”.
Simple Sudoku drequired only singles.
Sudoku Susser: 4 x Pinned Squares
Sadman Sudoku rates this “Easy”.
Paul’s Pages rated this “Hard”
tso
 
Posts: 798
Joined: 22 June 2005


Return to General