Hi@all.
Is there a possibility of using bitwise operators and logical shifting in Modelica?
Full Version: Bitwise operations
Hello Djordjevic,
short answer: No.
A little longer answer: No, but you can use external C functions for this.
For example, this will work:
I hope this helps a little bit.
Roland
short answer: No.
A little longer answer: No, but you can use external C functions for this.
For example, this will work:
CODE
function XOR
input Integer x1;
input Integer x2;
output Integer y;
annotation (Include="
int xor(int x1, int x2) {
return x1^x2;
}");
external "C" y = xor(x1, x2);
end XOR;
model Tester
Integer x=1;
Integer y=3;
Integer z = XOR(x, y);
end Tester;
input Integer x1;
input Integer x2;
output Integer y;
annotation (Include="
int xor(int x1, int x2) {
return x1^x2;
}");
external "C" y = xor(x1, x2);
end XOR;
model Tester
Integer x=1;
Integer y=3;
Integer z = XOR(x, y);
end Tester;
I hope this helps a little bit.
Roland
Thank you Roland.
I wanted to make many booleaan values out of one integer variable. So that e.g. input=6 would result the output false, true, true, false.
So the idea was to use a simple for-loop like in C:
for (i=0; i<4; i++)
output[i] = (x & (0001)<<(i)));
Now that its not possible I wanted to try it with another for-loop such as. This time (hopefully) modelica:
for i in 3:0 loop
if (input>2^i) then
output[i] = true;
input = input - 2^i;
else
output[i] = false;
end if;
end for;
But I am experiencing some compiling errors, which I'm not used at C++.
Maybe Modellica still "thinks", that I am using an eqation like x=x+1, which is wrong, instead of interpreting it as x:=x+1. But that's only my guess.
Can you tell me what I'm doing wrong?
I wanted to make many booleaan values out of one integer variable. So that e.g. input=6 would result the output false, true, true, false.
So the idea was to use a simple for-loop like in C:
CODE
for (i=0; i<4; i++)
output[i] = (x & (0001)<<(i)));
Now that its not possible I wanted to try it with another for-loop such as. This time (hopefully) modelica:
CODE
for i in 3:0 loop
if (input>2^i) then
output[i] = true;
input = input - 2^i;
else
output[i] = false;
end if;
end for;
But I am experiencing some compiling errors, which I'm not used at C++.
Maybe Modellica still "thinks", that I am using an eqation like x=x+1, which is wrong, instead of interpreting it as x:=x+1. But that's only my guess.
I know the algorithm-part needs an algorothm keyword, which I also use, and the varables input and output are not the one that would physically come into the Model.Of course, I could write the code 4-times for a 4-bit varable, but it would be very unpleasent and ugly to convert a 32-bit variable by copy-pasting.
Can you tell me what I'm doing wrong?
Write a function in c or in modelica should solve this problem
try to give you a simple example. Be aware, the code listed below use outputValue[1] to represent the 2^3 and output[4] the 2^0. Of course you can change the order by your self.
function IntegerToBinary
input Real inputValue;
output Boolean outputValue[4];
protected
Real temp;
algorithm
temp:=inputValue;
for i in 1:4 loop
if (temp>=2^(4-i)) then
outputValue[i] :=true;
temp:=temp - 2^(4 - i);
else
outputValue[i] :=false;
end if;
end for;
end IntegerToBinary;
Thank you Roland.
I wanted to make many booleaan values out of one integer variable. So that e.g. input=6 would result the output false, true, true, false.
So the idea was to use a simple for-loop like in C:
for (i=0; i<4; i++)
output[i] = (x & (0001)<<(i)));
Now that its not possible I wanted to try it with another for-loop such as. This time (hopefully) modelica:
for i in 3:0 loop
if (input>2^i) then
output[i] = true;
input = input - 2^i;
else
output[i] = false;
end if;
end for;
But I am experiencing some compiling errors, which I'm not used at C++.
Maybe Modellica still "thinks", that I am using an eqation like x=x+1, which is wrong, instead of interpreting it as x:=x+1. But that's only my guess.
Can you tell me what I'm doing wrong?
try to give you a simple example. Be aware, the code listed below use outputValue[1] to represent the 2^3 and output[4] the 2^0. Of course you can change the order by your self.
CODE
function IntegerToBinary
input Real inputValue;
output Boolean outputValue[4];
protected
Real temp;
algorithm
temp:=inputValue;
for i in 1:4 loop
if (temp>=2^(4-i)) then
outputValue[i] :=true;
temp:=temp - 2^(4 - i);
else
outputValue[i] :=false;
end if;
end for;
end IntegerToBinary;
QUOTE(Djordjevic @ Nov 18 2008, 09:25 AM) 
Thank you Roland.
I wanted to make many booleaan values out of one integer variable. So that e.g. input=6 would result the output false, true, true, false.
So the idea was to use a simple for-loop like in C:
CODE
for (i=0; i<4; i++)
output[i] = (x & (0001)<<(i)));
Now that its not possible I wanted to try it with another for-loop such as. This time (hopefully) modelica:
CODE
for i in 3:0 loop
if (input>2^i) then
output[i] = true;
input = input - 2^i;
else
output[i] = false;
end if;
end for;
But I am experiencing some compiling errors, which I'm not used at C++.
Maybe Modellica still "thinks", that I am using an eqation like x=x+1, which is wrong, instead of interpreting it as x:=x+1. But that's only my guess.
I know the algorithm-part needs an algorothm keyword, which I also use, and the varables input and output are not the one that would physically come into the Model.Of course, I could write the code 4-times for a 4-bit varable, but it would be very unpleasent and ugly to convert a 32-bit variable by copy-pasting.
Can you tell me what I'm doing wrong?
Hello Djordjevic,
what is not correct in your code:
Arrays in Modelica start with the index 1 and not 0.
You have (probably) written equations - what you need to use is an algorithm as you also wrote. You need this because of
In the if statement you have to compare using '>=' instead of just '>'.
One suggestion would be:
Of course all numbers greater than 2^(nBytes - 1) will result in y[nBytes] set to true.
Roland
what is not correct in your code:
Arrays in Modelica start with the index 1 and not 0.
You have (probably) written equations - what you need to use is an algorithm as you also wrote. You need this because of
CODE
input := input - 2^i;
In the if statement you have to compare using '>=' instead of just '>'.
One suggestion would be:
CODE
model Test
parameter Integer x = 6;
parameter Integer nBytes = 4;
Boolean y[nBytes];
protected
Integer tmp;
algorithm
tmp := x;
for i in 1:nBytes loop
if tmp >= 2^(nBytes - i) then
y[nBytes - i + 1] := true;
tmp := tmp - integer(2^(nBytes - i));
else
y[nBytes - i + 1] := false;
end if;
end for;
end Test;
parameter Integer x = 6;
parameter Integer nBytes = 4;
Boolean y[nBytes];
protected
Integer tmp;
algorithm
tmp := x;
for i in 1:nBytes loop
if tmp >= 2^(nBytes - i) then
y[nBytes - i + 1] := true;
tmp := tmp - integer(2^(nBytes - i));
else
y[nBytes - i + 1] := false;
end if;
end for;
end Test;
Of course all numbers greater than 2^(nBytes - 1) will result in y[nBytes] set to true.
Roland
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.