The New Sudoku Players' Forum

[udosuk]

Prompted by JPF's interesting brainteaser here, I worked a bit manually on this problem:

Code: Select all

Fill the integers 1 to 9 into the following 3x3 square:

+-+-+-+
| | | |
+-+-+-+
| | | |
+-+-+-+
| | | |
+-+-+-+

Let SP=sum of product of each row + sum of product of each column + sum of product of each main diagonal

The smallest SP I can find is 436, while the largest is 1101.

I wonder if anyone can verify/prove these or find a different result.

This is not really a brainteaser but rather an interesting piece of recreational mathematics...

[Glyn]

Matt

Just a quick note highest I have is from {715392648}.
35+54+192+126+36+80+270+504=1297

Couldn't get your lowest I'll have another look if it's still outstanding.

[RW]

Let SP=sum of product of each row + sum of product of each column + sum of product of each main diagonal

The smallest SP I can find is 436, while the largest is 1101.

Are you sure you counted in the diagonals in 436? Could you post the two boxes where you found those results?

RW

[Glyn]

RW has hit it on the head. I didn't post my lowest as it didn't match Matt's

Grid is {365718492}
Sum = 90+56+72+84+54+80+6+20=462.

[JPF]

With only the main diagonal , I get :

Min = 442

Code: Select all

169
734
852

Max = 1169

Code: Select all

724
596
318

Edit :

With the main diagonal and the antidiagonal , I get :

Min = 462

Code: Select all

285
916
473

Max = 1321

Code: Select all

518
394
726

JPF

[udosuk]

Sorry guys for the mistakes and confusion. I'm embarassed (no "oops" emoticon here to display it).

By "main diagonals" I meant both the leading diagonal and non-leading diagonal (aka anti-diagonal). They're called "main" as opposed to the "broken diagonals". But after some research I think I've misused the term so from now on I'll only use the term diagonal. Sorry again!

And of course I forgot to add the product of the 2 diagonals in my initial results. After a comprehensive check using Excel (and a resulting 32MB .xls file) I can confirm that JPF's range of 462..1321 is correct. Perhaps a bit surprising to me is if you don't count the diagonals (i.e. only 3 rows and 3 columns) the range become 436..947:

Code: Select all

Min = 436

1 6 9
7 3 4
8 5 2

Max = 947

1 2 7
3 4 8
5 6 9

While you can rearrange the 436-grid to become the 462-grid, you can't do the equivalent for the 947-grid to become the 1321-grid. And while in many examples the min-grid and the max-grid are convertible to each other by the "10 minus" operation, it isn't the case here for both pairs. It is probably due to the usage of the multiplication operation.

Also from my Excel results I've confirmed 90 as the minimun largest product for the 3 rows & 3 columns in a grid. And 54 as the maximum smallest one. So for anybody who wants an encore of JPF's previous brainteaser:

Prove elegantly that there must always be at least a row or a column which has a product smaller than or equal to 54.

[RW]

Prove elegantly that there must always be at least a row or a column which has a product smaller than or equal to 54.

That's easy. Digit 9 is in one row and one column. One of them must have a product greater than or equal to 9*1*X (X>1) and the other a product greater than or equal to 9*2*Y ( Y>2). The smallest possible product for the second is 9*2*3=54.

RW

[udosuk]

That's easy. Digit 9 is in one row and one column. One of them must have a product greater than or equal to 9*1*X (X>1) and the other a product greater than or equal to 9*2*Y ( Y>2). The smallest possible product for the second is 9*2*3=54.

RW, what you just did was only to prove that there is at least a row or a column which is larger than or equal to 54. I'm looking for a proof for "smaller than or equal to 54".

The 1 must have 4 other buddies on the same row/column. One of them must be from {23456}. The largest possible product of that line is 1*6*9=54.

RW must write the correct proof on his own before peeping at mine.

Now, what if the diagonals are also involved?

[Glyn]

RW must write the correct proof on his own before peeping at mine.

How will you know?

[RW]

Note to self: Don't solve brainteasers late saturday night!

Ok, you're right udosuk. But the real solution is of course just as easy. You have digit 1 in two units. The maximum minimum products for these two would be 1*7*8=56 and 1*6*9=54. I'm sure udosuk already gave a better description in invisible ink, so I won't bother to improve this answer!

[udosuk]

How will you know?

See? I trust RW will always play it straight like Kimi Räikkönen!

Now, what if the diagonals are also involved?

Nothing changes, the minimum maximum product is still 90, the maximum minimum product is still 54.

[udosuk]

How about this:

Code: Select all

Fill the integers 1 to 9 into the following 3x3 square:

+-+-+-+
| | | |
+-+-+-+
| | | |
+-+-+-+
| | | |
+-+-+-+

Let PS6 = product of sum of each row * product of sum of each column

and PS8 = PS6 * product of sum of each diagonal

What are the ranges of PS6 & PS8 respectively?

[JPF]

PS6

Code: Select all

Min = 680400

123
468
579


Max = 11390625

159
672
834


PS8

Code: Select all

Min = 653184000

264
819
573

Max = 5464166400

537
491
628


JPF

[udosuk]

JPF, one of your answers is not unique. Can you (or your program) spot it?

[udosuk]

JPF, one of your answers is not unique. Can you (or your program) spot it?

Okay, I spot it for you.

Code: Select all

Min = 653184000

264
819
573

264
918
573

Two new challenges:


1. Most Narrow-Ranged Square

Fill all of {1..9} into a 3x3 square. Write down the sums and products of all rows/columns/diagonals. Let

range width of sums = max sum - min sum

range width of products = max product - min product

combined range width = range width of sums + range width of products

Now find the 3x3 square with the most narrow combined range.


2. The (Almost) Anti-Magic Square

Fill all of {1..9} into a 3x3 square so that all 18 numbers from {1..18} appear either as a cell value or a sum of a row/column/diagonal. Oh wait a minute, 9+8=17 is fewer than 18 by 1... In that case one of the 18 numbers is allowed to appear as a sum of 2 corners.


Both these challenges, of course, have unique solutions (allowing reflection/rotation).