Including masking condition (NaN assignment) in anonymous function definition

If I want to mask some part of the plot (for example points located inside unit disk) I can use the following code:

[X,Y] = meshgrid(-6:.1:6) ;
[T,R] = cart2pol(X,Y) ;

f = @(t,r) r .* cos(t) ;

Z = f(T,R) ;      % (*)
Z( R< 1 ) = NaN ; % (*)

figure, surfc(X,Y,Z)

Is it possible to include the masking commands directly into anonymous function definition in order to avoid definition of auxiliary variable Z (lines with (*) ).

In my first attempt I followed the pattern of defining piece-wise continuous functions. Something like this

g = @(t,r) (r<2) .* (r .* cos(t) ) + (r>=2) .* NaN ; 
figure, surfc(X,Y, g(T,R) )

However, this code does not work because function g will always return NaN. How can it be done?

Answer 1

[Rik]

0/0

ans = NaN

You can use this to your advantage:

[X,Y] = meshgrid(-6:.1:6) ;
[T,R] = cart2pol(X,Y) ;

NaN_if_true_one_if_false=@(tf) (1-tf)./(1-tf);
g = @(t,r) (r .* cos(t) ) .* NaN_if_true_one_if_false(r<1);

surfc(X,Y,g(T,R))

%repeat original
f = @(t,r) r .* cos(t) ;
Z = f(T,R) ;      % (*)
Z( R< 1 ) = NaN ; % (*)
figure, surfc(X,Y,Z)

[tri stan]

This is what I was looking for. Thanks.