Password Entropy in Lazarus (Delphi-like)

My today's goal was to create a function, which would evaluate the entropy of a given string password. It's mostly about how random the password is.

This little code is based on the C# code from Jeff Atwood answer post on Calculating entropy of a string

Based on the wikipedia article I checked this as ok, still I am unsure if the logarithms are ok, or not, but it seems to work.

Let me know, please, if you see any problem with this:

function TPassword.GetEntropy: Extended;

type
  TDictionary = specialize TFPGMap<Char, Integer>;

var
  Dictionary: TDictionary;
  I: Integer;
  Chr: Char;
  PasswordLength: Integer;
  Frequency: Extended;

begin

  // dictionary initialization
  Dictionary := TDictionary.Create;

  try

    // in this cycle we add distinct keys into mapped dictionary with value 1
    // if the key (char) is already there, we increase the value
    for I := 1 to Length(FPassword) do begin

      Chr := FPassword[I];

      if Dictionary.IndexOf(Chr) = -1
        then Dictionary.Add(Chr, 1)
        else Dictionary[Chr] := Dictionary[Chr] + 1;

    end;

    // now that we went through the string and mapped its content,
    // we have for password 'abbbgg' map like a=>1, b=>3, g=>2

    PasswordLength := Length(FPassword);

    Result := 0;

    Frequency := 0;

    // - in this cycle we go through each dictionary item
    // - first we take the number of its occurrences in the string
    // and divide it with password length, let's call it frequency
    // - then there is the main part where we multiply the frequency
    // with natural logarithm of frequency divided by natural logarithm of 2
    // and we are adding it to the result until we went through all keys

    for I := 0 to Dictionary.Count - 1 do begin

      Frequency := Dictionary.Data[I] / PasswordLength;

      Result := Result - (Frequency * (Ln(Frequency) / Ln(2)));

    end;

  finally

    // dictionary freeing
    Dictionary.Free;

  end;

end;