Merge pull request #90 from yasserbakkar97/master

Add halfprecision_yaser_beker.py

Week06/halfprecision_yaser_beker.py

@@ -0,0 +1,57 @@
+class HalfPrecision:
+    def __init__(self, number):
+        if not isinstance(number, float):
+            raise TypeError("Input must be a float.")
+        self.number = number
+
+    def __str__(self):
+        # Check for special cases: NaN and infinity
+        if self.number != self.number or self.number == float('inf') or self.number == float('-inf'):
+            # Special case: infinity
+            if self.number == float('inf'):
+                return "0111110000000000"
+            # Special case: negative infinity
+            elif self.number == float('-inf'):
+                return "1111110000000000"
+            # Special case: NaN
+            else:
+                return "0111111000000000"
+
+        # Check for zero
+        if self.number == 0.0:
+            # Check for negative zero
+            if str(self.number)[0] == '-':
+                return "1000000000000000"
+            else:
+                return "0000000000000000"
+
+        # Get the sign bit
+        sign_bit = '0' if self.number >= 0 else '1'
+
+        # Convert the number to positive for simplicity
+        number = abs(self.number)
+
+        # Check for large numbers
+        if number >= 65536.0:
+            return "0111110000000000"
+
+        # Calculate the exponent and adjust the number
+        exponent = 0
+        while number >= 2.0:
+            number /= 2.0
+            exponent += 1
+        while number < 1.0:
+            number *= 2.0
+            exponent -= 1
+
+        # Convert the exponent to biased representation
+        exponent += 15
+        exponent_bits = format(exponent, '05b')
+
+        # Convert the mantissa
+        mantissa = format(int((number - 1.0) * 2**10), '010b')
+
+        # Combine the bits
+        binary_string = sign_bit + exponent_bits + mantissa
+
+        return binary_string