excel vba signed right shift operator
'Many languages have a bitwise signed-right-shift operator: >>
'VBA does not. However, it can be emulated in a performant manner:
Dim n as Long
n = 6
n = (n And -2) \ 2 '1 signed-right-shift (6 >> 1) = 3
n = 8
n = (n And -4) \ 4 '2 signed-right-shifts (8 >> 2) = 2
n = 170
n = (n And -8) \ 8 '3 signed-right-shifts (170 >> 3) = 21
n = -170
n = (n And -16) \ 16 '4 signed-right-shifts (-170 >> 4) = -11
'Notice the pattern:
'n = (n And -2^shfits) \ 2^shifts
'Important: Exponentiation is SLOW. It is much faster to use the hard coded
' result of the exponentiation, as in the examples above.
n = -2023406815
n = (n And -16777216) \ 16777216 '24 signed-right-shifts (-2023406815 >> 24) = -121
'Imporatant: Since a Long Integer in VBA is signed, (n >> 31) needs special
'processing. The pattern is different:
n = 2147483647
n = n And -2147483648 '31 signed-right-shifts (-2147483647 >> 31) = 0
'Here is a VBA function that encapsulates the details:
Function ShiftRight&(ByVal n&, Optional ByVal shifts& = 1)
Dim d&
Select Case shifts
Case 1: d = 2&
Case 2: d = 4&
Case 3: d = 8&
Case 4: d = 16&
Case 5: d = 32&
Case 6: d = 64&
Case 7: d = 128&
Case 8: d = 256&
Case 9: d = 512&
Case 10: d = 1024&
Case 11: d = 2048&
Case 12: d = 4096&
Case 13: d = 8192&
Case 14: d = 16384&
Case 15: d = 32768
Case 16: d = 65536
Case 17: d = 131072
Case 18: d = 262144
Case 19: d = 524288
Case 20: d = 1048576
Case 21: d = 2097152
Case 22: d = 4194304
Case 23: d = 8388608
Case 24: d = 16777216
Case 25: d = 33554432
Case 26: d = 67108864
Case 27: d = 134217728
Case 28: d = 268435456
Case 29: d = 536870912
Case 30: d = 1073741824
Case 31: ShiftRight = CBool(n And &H80000000): Exit Function
Case 0: ShiftRight = n: Exit Function
End Select
ShiftRight = (n And -d) \ d
End Function
'Don't be off-put at the size of the function. This is many times faster than
'any other VBA function that carries out bitwise signed-right-shfits. It is
'thousands of times faster than calling Excel's
'WorksheetFunction.Bitrshift() method.
'However, remember that function calls are expensive in general. The inline
'examples above are approximately 20 times faster than calling this highly
'optimized function. But you have to get the math right. The function
'takes care of it for you.