Discussion about math, puzzles, games and fun. Useful symbols: ÷ × ½ √ ∞ ≠ ≤ ≥ ≈ ⇒ ± ∈ Δ θ ∴ ∑ ∫ π -¹ ² ³ °

You are not logged in.

- Topics: Active | Unanswered

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

Hi Au101

I thought this was straight forward but I'm now on explanation #3 and I still cannot get it right.

This is what ought to work.

Choose three sets of values for 's' and 't' so that you have the position vectors for three points in the plane. **

Transform them using T to get three position vectors for points in the new plane. (say A, B and C)

Find vectors that lie in the new plane; AB and BC would do.

Let the vector perpendicular to the new plane be

and make equations with a, b, and c by dot products AB.n = 0 and BC.n = 0

That gives two equations and three unknowns; not enough to find a single 'n',

but any multiple of 'n' will also work so choose a value for, say, 'a'

and then use the equations to find b and c. That will give you a solution for 'n'.

(If you want, you could try another value of 'a' and generate new values for b and c.

Although different, you will find this second solution is just a multiple of the first so either will do.)

So now you know 'n'.

You know that the transform of

is position vector for a point in the new plane so substitute into r.n = p to find the constant 'p'.

I've done all that and got an equation for the new plane.

Now to check it. The other position vectors should also fit this solution.

So far I haven't managed to acheive this!

So, I think my calculations have an error but I cannot find it. I don't think the logic is faulty (I hope).

Do you want to try this and post your figures. Then I can compare them with mine.

Bob

ps. Explanation #1 had a different line at the point marked **

Instead of transforming three position vectors, I tried transforming the two vectors

lying in the first plane and assuming the result was two vectors in the second plane.

That gave a completely different solution and I don't think it is correct.

Does the book make the following point clear:

When T transforms a plane it (i) transforms position vectors for points in the plane OR (ii) it transforms vectors that lie in the plane.

*Last edited by bob bundy (2011-01-01 22:30:45)*

You cannot teach a man anything; you can only help him find it within himself..........Galileo Galilei

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Hmmmm it's a tricky one, I copied out the question exactly as it stands in the book, but if it helps I can copy out the book's example of a problem which is similar to these questions, but not quite the same, perhaps that would shed some light:

*Last edited by Au101 (2011-01-02 01:36:24)*

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

I don't know if that will help but that's the book's angle, I can't find any further clarification in the book I'm afraid. I'd be happy to have a go at getting my own figures, except I'm not entirely sure I understand what to do.

*Last edited by Au101 (2011-01-01 06:30:48)*

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

Hi Au101,

Thanks for that.

I'm often making numeic errors, so I'm going to create Excel formulas for the planes and transformations so I don't keep slipping up.

Then I can sub in any plane and any matrix and quickly generate vectors. I'll use the book example to check my formulas.

Then I'll come back to you.

Looking at that example I think it has answered my question, anyway.

Bob

*Last edited by bob bundy (2011-01-01 05:51:04)*

You cannot teach a man anything; you can only help him find it within himself..........Galileo Galilei

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

Your last post:

Did you mean that? You cannot make a plane with the same vector for all three components.

Bob

You cannot teach a man anything; you can only help him find it within himself..........Galileo Galilei

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Oh, whoops, sorry, no I didn't I copied and pasted the matrix template for simplicity and forgot to change the entries! Sorry for the confusion, it should be the same as the last line of the preceding post, I'll just change it.

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

hi Au101,

Finally got it sorted. The method outlined in post 26 is sound. When I worked out a, c and b I got them in that order,

but failed to write them correctly for the vector 'n'.

The book method seems a bit complicated to me as you land up with such a tricky matrix multiplication to do.

Given what I'm like with little errors I'd never get it right that way. But if you want to try it, it'll probably be good for you.

To get the equation for the second plane in the required format you have got to find a way to compute 'n',

the vector that is perpendicular to the plane.

My method gets you two vectors in the plane, and it is fairly straight forward to get 'n' from them.

Then you can use any point in the plane to find 'p'.

If you are still finding this tough I have a 2D example I can send. I made it up when I was getting desperate to sort out why my results were inconsistent.

I wanted to check I had the theory soundly worked out.

And the answer to my question

When T transforms a plane it (i) transforms position vectors for points in the plane OR (ii) it transforms vectors that lie in the plane.

is definitely (i), which was what I had worked with all along.

Let me know if you have managed to complete the equation of the plane or want more help.

Best wishes,

Bob

ps. When this thread moved onto page 2 (at post 26), the page width went wrong and had stayed so for me ever since. How does it display for you?

*Last edited by bob bundy (2011-01-01 22:46:15)*

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Ooooh thanks ever so much for all of your tribulations Bob. I suppose that either method would suit me fine, I'm just not entirely sure I'm clear on your method, I don't suppose you could provide an example, or further instruction to clarify? (I'm sorry to request further tribulations). Otherwise the book method is perfectly fine, but I don't really see how to move from the book's finding the vector equation to finding an equation in the form

Oh and yes I'd meant to say, the screen is much wider than normal, it may have been to do with my using the LaTeX text environment, which does not drop lines automatically. I tried to fix it, but it doesn't seem to have made much of a difference I'm afraid.

*Last edited by Au101 (2011-01-02 01:53:34)*

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

hi

I see you are logged on at the moment. I've just come in from gardening to have lunch. I'll keep the computer running but you'll have to wait until later for the example as I want to take advantage of the light and fine weather to complete the garden jobs.

Bob

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Of course, I'm perfectly happy to wait, especially for answers as instructive as yours. However, I'm surprised to here about the weather, it's very cloudy here on (presumably) the other side of Essex.

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

I've got white fluffy clouds and some blue bits.

temp = 3 celcius

B

*Last edited by bob bundy (2011-01-02 02:04:28)*

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

hi Au101

OK. Here we go.

There are three forms for the equation of a plane (that I can think of, there may be more).

They are:

form 1

form 2; the dot product form

and

form 3

The first form is derived from the diagram in post 20.

This is the position vector for a point in the plane

and

are two vectors lying in the plane.

In the second form

is a vector that is perpendicular to all vectors lying in the plane. r and n are being 'dotted' together and the result is a constant.

If you then replace 'r' by

and do the dot product

you get the cartesian version that is the third form.

From this you can see that form 2 and form 3 are the same plane.

Less obviously, so is form 1.

Proof:

In form 1 put lambda = 1 and mu = 1 and you get x = 3, y = 3, z = 1

put lambda = 1 and mu = -1 and you get x = 1, y = -1, z = -1

put lambda = 0 and mu = 1 and you get x = 2, y = 3, z = 2

If you try each of these sets of values in x - y + z you get 1 every time.

So form 1 and form 3 have three non-collinear points in common. That's enough to prove they represent the same plane because you only need three points to define a plane (provided they are not in the same straight line).

Now to the question.

You know the equation of the plane before the transformation. The book method transforms points directly from this equation to get the equation after the transform.

I think the algebra for this is a bit horrid and you still have to convert it into form 2.

So my method was to pick three points in the plane and transform them. Let's call them A, B and C. That's just number work. ***

Now get two vectors in the plane by doing AB = AO - OB etc.

Any two will do because, provided they are not parallel, any two vectors will 'span' the plane ie. will enable you to reach all points in the plane.

Now to get the vector that's at right angles to both these vectors.

Imagine by some trick of gravity you can stand on the plane with 'up' meaning 'at right angles to' the plane. If I asked you to point a stick at right angles to the plane it would go straight up. If you draw any line on the 'ground' = 'the plane' , it would be at right angles to the stick. And it wouldn't matter how long the stick was. A stick twice as long, would still be at right angles to every line in the plane.

Call the vector at right angles

So do a dot product between one vector in the plane and 'n' and set it equal to zero. Do again for the second vector.

You've got 2 equations with a, b and c as unknowns. Choose any 'a' to make the calculations easy. That will enable you to work out 'b' and 'c'. That gives you a possible 'n' the vector perpendicular to the plane. You might think it is cheating to choose 'a'. But if you had chosen an 'a' that was twice as big, you'd have got 'b' and 'c' twice as big as before so all that would happen is you'd get '2n' for the perpendicular vector. You can use any vector that is at right angles so the first would do!

Now to get 'p'.

The equation

is the equation for all points, 'r', in the plane.

Back at point *** we had three possible points so 'sub' in any one set of x, y and z values and you'll get 'p'

Check by 'subbing' in the other values from *** to see if you get the same p.

Problem done!

Does that all make sense?

Bob

*Last edited by bob bundy (2011-01-02 06:07:51)*

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Wow what an explanation thank you so much.

So, firstly we begin by picking three points, presumably from

So we could say let

Which of course gives

Respectively, unless I'm very much mistaken. Now, if I have understood correctly so far we must now transform these points with **T**, which, I believe, gives:

But at this point I don't fully understand AB = AO - OB etc.

*Last edited by Au101 (2011-01-02 07:36:06)*

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

hi

Ok so far.

If we give your points letters, (don't forget, they are position vectors with respect to the origin, O)

then

Now do that again for BC = BO + OC

Can you continue from here?

Bob

*Last edited by bob bundy (2011-01-02 08:53:09)*

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Well, we only need one other vector, so we may as well do *BC*

So we now do a dot product between one vector in the plane and 'n' and set it equal to zero. Do again for the second vector.

Now if I can choose any value for *a* then I may as well go with 1, which gets us:

Okay so let's choose:

Now, at this point the most sensible thing to do would seem to be to multiply *n* by 10, but I happen to know that the book's answer actually requires me to multiply by -10. Now, unless I'm very much mistaken, the values of *n* make no difference at all, but it'd be nice to get the book's value of *n*, just to tie it off nicely. So we have:

Which leaves me with something of a problem, because the correct answer should be 120

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

hi Au101,

Interesting. I got 120. It'll take a while to find the differences and I'm in the middle of de-salting my car (again while the weather holds). I'll come back to this when the sun goes down.

Bob

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

Hi again,

I see what's wrong. You cannot use

because that's a position vector for the original plane. You need one for the new plane so choose from

Then it will work. :)

Bob

Offline

**Au101****Member**- Registered: 2010-12-01
- Posts: 351

Ah, okay, so let's say

And I think we have it, thank you so much

*Last edited by Au101 (2011-01-03 04:16:24)*

Offline

**bob bundy****Moderator**- Registered: 2010-06-20
- Posts: 6,916

You do indeed. Well done!

Bob

ps

I made it

which shows what I mean about many answers being possible.

*Last edited by bob bundy (2011-01-03 05:05:47)*

Offline