Right matrix divide / Inverse matrix

Good day,

As I am working on my model, I need to include bony landmarks. The reference
book I use uses a different axis for the segments (Paul Allard, 1995, in
this case). I therefore need to transform the bony landmarks coordinates
from their system to anybody’s. I can do all transformation by hand (i.e.
using matlab) and hardcode this information in the model, but I really don’t
like this idea because it means that the model is hard to maintain/modify. I
would rather calculate that transformation directly in the model, by
incorporating the reference axis from my source (A) and the reference axis
from the repository (b), and then use the right matrix division (Ax=b ->
x=A/b) to transform the data directly. I could therefore include, in the
model, the raw data from the book, and then transform it within anybody,
making the model a lot easier to maintain and more general.

In other words, I’d like to be able to calculate, within anybody, the
transformation matrix I would include in AnyFunTransform3DLin. For this, I
would need either a “AnyFunRDivide” or “AnyFunInvMat”, since x=A/b can
generally be calculated by x=A*Inv(b). Do any of those functions exist and I
missed it? Or this can’t be done? If it can’t be done, can’t it be added? Is
there another way to do this?

Thank you all!

Jean-Olivier

[Non-text portions of this message have been removed]

Hi Jean Olivier

I don’t think the functions you ask about exist at least not the
name;-). I guess you could make them yourself? Will you need a
rotation matrix for each segment?
I think you can find most of what you are seeking here:
http://www.anybodytech.com/495.0.html
Else please write again:D

Best regards
Christian, AnyBody Support

— In anyscript@yahoogroups.com, Jean-Olivier Racine
<jean-olivier.racine.1@…> wrote:
>
> Good day,
>
>
>
> As I am working on my model, I need to include bony landmarks. The
reference
> book I use uses a different axis for the segments (Paul Allard, 1995, in
> this case). I therefore need to transform the bony landmarks coordinates
> from their system to anybody’s. I can do all transformation by hand
(i.e.
> using matlab) and hardcode this information in the model, but I
really don’t
> like this idea because it means that the model is hard to
maintain/modify. I
> would rather calculate that transformation directly in the model, by
> incorporating the reference axis from my source (A) and the
reference axis
> from the repository (b), and then use the right matrix division (Ax=b ->
> x=A/b) to transform the data directly. I could therefore include, in the
> model, the raw data from the book, and then transform it within anybody,
> making the model a lot easier to maintain and more general.
>
>
>
> In other words, I’d like to be able to calculate, within anybody, the
> transformation matrix I would include in AnyFunTransform3DLin. For
this, I
> would need either a “AnyFunRDivide” or “AnyFunInvMat”, since x=A/b can
> generally be calculated by x=A*Inv(b). Do any of those functions
exist and I
> missed it? Or this can’t be done? If it can’t be done, can’t it be
added? Is
> there another way to do this?
>
>
>
> Thank you all!
>
> Jean-Olivier
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

Hello Jean-Olivier and Christian,

I send this email yesterday - unfortunately from a wrong account, so it was
not accepted by the group’s mailserver. Christian answered in the meantime,
but I think you should have my answer too.

Here comes the text from yesterday

x=A/b, AnyFunRDivide, and AnyFunInvMat - it is all good ideas. I guess
something like this will come in time, especially if a lot of you out there
demand for it;-)

In general, it has never been the plan that AnyScript should be a Matlab
clone, but we will try to follow demands from the users, and Matlab
certainly sets a standard. However the functions you mention are not yet on
the wish-list for the next version, so I cannot say when we will manage -
sorry.

When you speak of reference axis, isn’t that right-hand coordinate reference
system, and the A matrix is a 3 by 3 rotational transformation matrix? It
sounds like it in your description.

If so the matrices are orthogonal, which implies that inv(A) = A^T (or A’ in
Matlab/AnyScript syntax).

You can easily test this of course by computing A’*A.

I hope I understood you correctly about A, because then you should be able
to deal with your case in current AnyScript.

Best regards,

Michael


From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of Jean-Olivier Racine
Sent: Tuesday, February 06, 2007 16:20 PM
To: anyscript@yahoogroups.com
Subject: [anyscript] Right matrix divide / Inverse matrix

Good day,

As I am working on my model, I need to include bony landmarks. The reference
book I use uses a different axis for the segments (Paul Allard, 1995, in
this case). I therefore need to transform the bony landmarks coordinates
from their system to anybody’s. I can do all transformation by hand (i.e.
using matlab) and hardcode this information in the model, but I really don’t
like this idea because it means that the model is hard to maintain/modify. I
would rather calculate that transformation directly in the model, by
incorporating the reference axis from my source (A) and the reference axis
from the repository (b), and then use the right matrix division (Ax=b ->
x=A/b) to transform the data directly. I could therefore include, in the
model, the raw data from the book, and then transform it within anybody,
making the model a lot easier to maintain and more general.

In other words, I’d like to be able to calculate, within anybody, the
transformation matrix I would include in AnyFunTransform3DLin. For this, I
would need either a “AnyFunRDivide” or “AnyFunInvMat”, since x=A/b can
generally be calculated by x=A*Inv(b). Do any of those functions exist and I
missed it? Or this can’t be done? If it can’t be done, can’t it be added? Is
there another way to do this?

Thank you all!

Jean-Olivier

[Non-text portions of this message have been removed]

[Non-text portions of this message have been removed]

Hello guys,

x, not A, is a rotational matrix, but you got the idea right. Your comment
made me notice that I have a problem with my destination reference system,
it is not orthogonal. Your solution probably works, using the tools
Christian pointed out (thanks Christian, I completely forgot about that part
of the tutorial!). So once I get my data fixed, I guess I’ll be able to use
the orthogonal matrix simplification you described to get the desired
result.

Thanks guys!

Jean-Olivier


From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of AnyBody Support
Sent: 7 février 2007 03:06
To: anyscript@yahoogroups.com
Subject: Re: [anyscript] Right matrix divide / Inverse matrix

Hello Jean-Olivier and Christian,

I send this email yesterday - unfortunately from a wrong account, so it was
not accepted by the group’s mailserver. Christian answered in the meantime,
but I think you should have my answer too.

Here comes the text from yesterday

x=A/b, AnyFunRDivide, and AnyFunInvMat - it is all good ideas. I guess
something like this will come in time, especially if a lot of you out there
demand for it;-)

In general, it has never been the plan that AnyScript should be a Matlab
clone, but we will try to follow demands from the users, and Matlab
certainly sets a standard. However the functions you mention are not yet on
the wish-list for the next version, so I cannot say when we will manage -
sorry.

When you speak of reference axis, isn’t that right-hand coordinate reference
system, and the A matrix is a 3 by 3 rotational transformation matrix? It
sounds like it in your description.

If so the matrices are orthogonal, which implies that inv(A) = A^T (or A’ in
Matlab/AnyScript syntax).

You can easily test this of course by computing A’*A.

I hope I understood you correctly about A, because then you should be able
to deal with your case in current AnyScript.

Best regards,

Michael


From: <mailto:anyscript%40yahoogroups.com> anyscript@yahoogroups.com
[mailto: <mailto:anyscript%40yahoogroups.com> anyscript@yahoogroups.com] On
Behalf
Of Jean-Olivier Racine
Sent: Tuesday, February 06, 2007 16:20 PM
To: <mailto:anyscript%40yahoogroups.com> anyscript@yahoogroups.com
Subject: [anyscript] Right matrix divide / Inverse matrix

Good day,

As I am working on my model, I need to include bony landmarks. The reference
book I use uses a different axis for the segments (Paul Allard, 1995, in
this case). I therefore need to transform the bony landmarks coordinates
from their system to anybody’s. I can do all transformation by hand (i.e.
using matlab) and hardcode this information in the model, but I really don’t
like this idea because it means that the model is hard to maintain/modify. I
would rather calculate that transformation directly in the model, by
incorporating the reference axis from my source (A) and the reference axis
from the repository (b), and then use the right matrix division (Ax=b ->
x=A/b) to transform the data directly. I could therefore include, in the
model, the raw data from the book, and then transform it within anybody,
making the model a lot easier to maintain and more general.

In other words, I’d like to be able to calculate, within anybody, the
transformation matrix I would include in AnyFunTransform3DLin. For this, I
would need either a “AnyFunRDivide” or “AnyFunInvMat”, since x=A/b can
generally be calculated by x=A*Inv(b). Do any of those functions exist and I
missed it? Or this can’t be done? If it can’t be done, can’t it be added? Is
there another way to do this?

Thank you all!

Jean-Olivier

[Non-text portions of this message have been removed]

[Non-text portions of this message have been removed]

[Non-text portions of this message have been removed]

Hello Jean-Olivier

You are very welcome. We are glad to be helpful.

Best regards
Christian, AnyBody Support

— In anyscript@yahoogroups.com, Jean-Olivier Racine
<jean-olivier.racine.1@…> wrote:
>
> Hello guys,
>
>
>
> x, not A, is a rotational matrix, but you got the idea right. Your
comment
> made me notice that I have a problem with my destination reference
system,
> it is not orthogonal. Your solution probably works, using the tools
> Christian pointed out (thanks Christian, I completely forgot about
that part
> of the tutorial!). So once I get my data fixed, I guess I�ll be able
to use
> the orthogonal matrix simplification you described to get the desired
> result.
>
>
>
> Thanks guys!
>
> Jean-Olivier
>
>
>
> _____
>
> From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com]
On Behalf
> Of AnyBody Support
> Sent: 7 f�vrier 2007 03:06
> To: anyscript@yahoogroups.com
> Subject: Re: [anyscript] Right matrix divide / Inverse matrix
>
>
>
> Hello Jean-Olivier and Christian,
>
> I send this email yesterday - unfortunately from a wrong account, so
it was
> not accepted by the group’s mailserver. Christian answered in the
meantime,
> but I think you should have my answer too.
>
> Here comes the text from yesterday
>
> x=A/b, AnyFunRDivide, and AnyFunInvMat - it is all good ideas. I guess
> something like this will come in time, especially if a lot of you
out there
> demand for it;-)
>
> In general, it has never been the plan that AnyScript should be a Matlab
> clone, but we will try to follow demands from the users, and Matlab
> certainly sets a standard. However the functions you mention are not
yet on
> the wish-list for the next version, so I cannot say when we will
manage -
> sorry.
>
> When you speak of reference axis, isn’t that right-hand coordinate
reference
> system, and the A matrix is a 3 by 3 rotational transformation
matrix? It
> sounds like it in your description.
>
> If so the matrices are orthogonal, which implies that inv(A) = A^T
(or A’ in
> Matlab/AnyScript syntax).
>
> You can easily test this of course by computing A’A.
>
> I hope I understood you correctly about A, because then you should
be able
> to deal with your case in current AnyScript.
>
> Best regards,
>
> Michael
>
> _____
>
> From: <mailto:anyscript%40yahoogroups.com> anyscript@yahoogroups.com
> [mailto: <mailto:anyscript%40yahoogroups.com>
anyscript@yahoogroups.com] On
> Behalf
> Of Jean-Olivier Racine
> Sent: Tuesday, February 06, 2007 16:20 PM
> To: <mailto:anyscript%40yahoogroups.com> anyscript@yahoogroups.com
> Subject: [anyscript] Right matrix divide / Inverse matrix
>
> Good day,
>
> As I am working on my model, I need to include bony landmarks. The
reference
> book I use uses a different axis for the segments (Paul Allard, 1995, in
> this case). I therefore need to transform the bony landmarks coordinates
> from their system to anybody’s. I can do all transformation by hand
(i.e.
> using matlab) and hardcode this information in the model, but I
really don’t
> like this idea because it means that the model is hard to
maintain/modify. I
> would rather calculate that transformation directly in the model, by
> incorporating the reference axis from my source (A) and the
reference axis
> from the repository (b), and then use the right matrix division (Ax=b ->
> x=A/b) to transform the data directly. I could therefore include, in the
> model, the raw data from the book, and then transform it within anybody,
> making the model a lot easier to maintain and more general.
>
> In other words, I’d like to be able to calculate, within anybody, the
> transformation matrix I would include in AnyFunTransform3DLin. For
this, I
> would need either a “AnyFunRDivide” or “AnyFunInvMat”, since x=A/b can
> generally be calculated by x=A
Inv(b). Do any of those functions
exist and I
> missed it? Or this can’t be done? If it can’t be done, can’t it be
added? Is
> there another way to do this?
>
> Thank you all!
>
> Jean-Olivier
>
> [Non-text portions of this message have been removed]
>
> [Non-text portions of this message have been removed]
>
>
>
>
>
> [Non-text portions of this message have been removed]
>