Professional Documents
Culture Documents
Music, like all human art, is unabashedly telling, not only of our aptitude for
pattern recognition, but pattern synthesis as well. And so it is that we have done so, in
the form of fields such as mathematics, linguistics, philosophy, and even music. And
though these fields differ greatly, both axiomatically and axiologically, their parallels are
what seem to allow them grow from each other. Now, perhaps not all discovers that
apply to mathematics will have any use in music. Admittedly, I know not if it is possible
to reconcile all of the theorems proven in mathematics per musical analysis, or even if
the premise is laughable. However, there are many definitions provided by music
theorists, used for musical analysis, that can without doubt be explicated in
mathematical language. This is a direct consequence of the parallels between
mathematics, and music theory. Both these fields, though not exclusively, concern
methods of measurement, calculation, and classification. In regard to both of these,
mathematics has achieved tremendous depth, and so I believe it is in the best interest of
music theory to make methodologies and definitions readily comprehensible in
mathematics. In what follows, then, I will attempt to explicate the method by which the
calculations of musical analysis can be performed, by both humans and computers, with
ease. In order to move any further, however, a few mathematically concepts must be
established, which I will reapply when I introduce the music theoretic concepts.
Firstly, consider the Heaviside step function which, when it's argument is either
greater or less than '0', is valued at either '1' or '0', respectively (where an argument of '0'
is treated as positive). Then consider the following summation series;
H [ xn]
n=1
If 'x' is some positive value, then 'f(x-n)' will continue to be '1', until 'n' becomes
larger than 'x', which we can defined by the following inequalities;
n x 0=H [ xn]
xn 1=H [ xn]
Now, It is true that 'x' could very well not be an integer. However, since there is an
integer larger than any non-integer, the series will halt the addition at the point which 'n'
is an integer greater than 'x'. This then means that the number of times that '1' is added,
is equal to the closest integer that is not greater than 'x'. (where a difference of '0' is,
consistently, no greater than 'x')
x = H [ xn]
n=1
This definition, however, is useful because of the flexibility built into the function
through general treatment of the magnitudes of each variable, exemplified by the
equivalent relation;
x
x = H [ 1]
n=1 n
This is because, of course, if;
n x 0= H [ xn]
xn 1= H [ xn]
then surely;
x x
1 0=H [ 1]
n n
x x
1 1=H [ 1]
n n
Further exampled, though, assume that our original series was defined so that 'n' was
multiplied by 'k' , or;
H [ xkn]
n=1
H [ xnk ]
n=1
Again, we consider the inequality relations for the yields of 'f' and the relative
magnitudes of the terms within it's argument, as;
nk x 0=H [ xn k ]
xn k 1= H [ xnk ]
Such that the following relation is justified by exponentiation by '1/k';
1 1
k k
n x 0= H [ x n]
1 1
x k n 1= H [ x k n]
such that, again, we are able to resolve the given series as;
1
k k
x = H [ xn ]
n=1
But, even better, assume that 'k' is exponentiated to the degree of 'n', or;
H [ xk n ]
n=1
We, once more, establish the inequality relation for 'f', which in this case becomes;
k n x 0= H [ xk n ]
xk n 1=H [ xk n ]
Such that, if we take the natural log of both sides of the inequality, and divide away
the 'ln(k)', the inequality becomes;
ln x ln x
n 0=H [ n]
ln k ln k
ln x ln x
n 1=H [ n]
ln k ln k
Which, even still, leaves us with the following identity;
ln x n
= H [ xk ]
ln k n=1
So then, considering the wealth of possible relations implied by the existence of 'f',
defining 'f' would be of great interest to anyone who endeavours to solve an equation of
this form.
In order to do this, though, first consider the lambda function of any variable 'v', to
always be '0' when sgnv =1 , and 'v' when sgnv =1 . This is easily done by
summing the absolute value of 'v' to 'v', and dividing by two, which gives;
vv
v =
2
Now, if we divide this by 'v', though, we reduce the magnitude of the increasing side
of the function to '1', as it's magnitude was also 'v', but in the case of a negative
argument, the result is '0/v', which reduces to '0' in all cases barring when 'v' is equal to
'0'. However, to prevent the indeterminacy at this value, we must consider this same
form raised to the degree of '-v'. This is so that, when 'v' is '0', the limit of each
expression, as 'v' approaches '0' from the positive side, in the numerator and
denominator approach '1', rather than '0', or;
v v
lim v =lim v =1
x 0 x 0+
And also, that when sgnv =1 , the value does not reciprocate to become
Which is also equivalent to the derivative of the lambda function, exponentiated by '-
v', or;
H [v ]= ' v v
With this complete, we may move on to an introduction of the metrics and
classifications used in musical analysis, as well as a demonstration of how the above
work is applicable to it.
Music theory
At first, with very little information, it might seem intuitively correct that the
distance between any position the position just above it, on the staff, is equal to the
distance between any other two of the 'same relationship', with respect to each other.
Alas, it is not even correlated this way for the bi-partition of line to space and space
to line progressions of the 'same relationship'. However, an instrument like the piano
does successful establish a unitary progression from the 'tonic' to the 'octave' of a scale,
which is also periodic in form.
Above is a keyboard, each of it's keys labeled with a letter which refers to the class
of frequencies, which ideally reenforce each other, that each belongs to. As, we notice
already, there are keys that lie between those defined on the staff. These also follow a
continually increasing progression, but each key that immediately follows another, from
left to right on the keyboard, is in fact equidistant to any other two keys of the 'same
relationship'. This distance is referred to as the half-step, or semitone.
So now, we can define a function 'h(n)' which, given our tonic 'C', gives the distance,
in semitones, from 'C' to the corresponding note. This begets;
But, since both progressions are consistently periodic, it is definitely true that if;
bn=7b1
Then;
h n=12b
From only the first three values of 'h(n)', it is true that 'h(n)' behaves no differently than
'2(n-1)', or;
0n3 h n =2 n1
However, this changes when 'n' becomes larger than '3', which for the next four values of
'h(n)', the function stays consistently '1' step below what is expected for 'h(n)', given the
definition above, or;
3n7 h n=2n11
But, once again, after 'n' becomes larger than '7', the function disobeys the above relation
in like manner, or;
7n11 h n =2n111
However, since a subtraction of '1' semitone happens when 'n' is larger than '3', so must
this occur when 'n' is larger than the octave of '3', which then also occurs when 'n' is
larger than the octave of the octave of '3'. Subtraction of '1' semitone also occurs when
'n' is greater than the octave of '7', and this continues in the same manner as '3'.
However, one could explain this as a simple change of rules per the oddness, or
evenness of some expression. It is true though, that the following expression, where '(2a-
1)' is the form of all odd numbers, is continuously '' larger than the scale degree of the
(2a-1)th octave of '3';
7 2a10
2
And, similarly, for the (2a)th octave of '7';
7 2a 1
2
But, these both are numbers of the form;
a
11
7a
2
2
However, the number of times that 'n' is found to be larger than the above statement, as
'a' approaches infinity from '1', is the exact number of '1's to be subtracted from '2(n-1)',
per 'n', such that 'h(n)' may be defined as;
a
11
7a
2
h n=2 n1 H [n ]
a=1 2
Which, in general form, is also;
a
h n=2 n1 H [4n14a 11 ]
a=1
Q.E.D
Music theorem no.1
If one wishes play a scale of a mode ' m ' from some tonic given by ' S m n '
when ' n=1 ', it is possible to play the Ionian of the same tonic ' h n ', but starting
from ' m ', which results in a scale of mode and tonic given by ' h nm1 , which
is globally parallel to any other scale of the same mode. However, if one wishes the
tonic to remain unchanged, one can simply subtract the constant amount ' h m ', by
which the scale becomes displaced from the initial tonic due to the shift, which results in
a scale of mode ' m ' that is of the initial tonic.
S m n=h nm1h m
The simplification proof
Recalling the previously stated series identity for the floor function, given by;
v = H [ va]
a=1
Recall, also, the identity for 'h(n)' given explicitly, in terms of 'n', by;
hn=2n1 H [ 4n14a1a ]
a=1
Now, it is clear that the only obstacle to reducing the sum is the inseparability that
results from '-1' being exponentiated by the index. However, since the index is restricted
to being an even or odd integer, the expression ' 1a ' can be resolved by separating
the sum into a sum of it's even and odd indexed terms. This is done by replacing 'a' with
both '2a', for all evens, and '2a-1', for all odds. This begets;
2n1[ H [ 4n14 2a1 ] H [ 4n14 2a11
2a 2a1
]]
a=1 a=1
Reflecting back upon the scale formula which accounts for both mode and scale
degree, which is defined as;
S m n =h nm1h m
And, using the shifted sawtooth function of the mode, in order to restrict analysis to
pitch class;
m1
7t 1
7
Which, through direct substitution, yields;
4n 1 m3 4n1 m1
S m n=2 n1 t t
28 7 28 7
Triple and quadruple periodic scales
Since scales which are not defined as modes of ionian, such as melodic and
harmonic minor which are explicitly defined in terms of the '7' modes as;
+
S 2,4 n=S 2 n S 1 nS 5 n
and;
+
S 6,4 n=S 6 nS 1 nS 5 n
Respectively, such that the next group of scales, which are monotonically increasing,
are of the form;
S m1,m2 m3 n=S m1, AB n [S m2, B A nS m3, B A n]
Where 'A' or 'B' refer to the qualities of the respective scales (the quality is major, or
'A' when the 3rd scale degree is '4' semi-tones from the tonic, as opposed to the minor 'B'
which is '3' semitones away);
S 2,3 n=S 2 n [S 1 n S 4 n]
S 2,1 n=S 2 nS 4 nS 5 n
S 2,4 n=S 2 nS 1 n S 5 n
S 3,3 n=S 3 n [ S 1 nS 4 n]
S 3,1 n=S 3 n [ S 4 nS 5 n]
S 3,4 n=S 3 n [ S 1 nS 5 n]
S 6,3 n=S 6 n [ S 1 nS 4 n]
S 6,1 n=S 6 n [ S 4 nS 5 n]
S 6,4 n=S 6 n [ S 1 nS 5 n]
and the predominant majors;
S 1,3 n=S 1 n [ S 6 nS 3 n]
S 1,1 n=S 1 n [ S 2 nS 3 n]
S 1,4 n=S 1 n [ S 2 nS 6 n]
S 4,3 n=S 4 n [ S 6 nS 3 n]
S 4,1 n=S 4 n [ S 2 nS 3 n]
S 4,4 n=S 4 n [ S 2 nS 6 n]
S 5,3 n=S 5 n [ S 6 nS 3 n]
S 5,1 n=S 5 n[ S 2 nS 3 n ]
S 5,4 n=S 5 n[ S 2 nS 6 n]
The spacing chosen above, which is by twos, is the one which governs the formation of
the triad. In music, an interval is a group of two notes. Instead of this terminology
though, I will will refer the interval as a dyad, and collections of '3' notes as triads, and
tetrad for '4', ect. The fewer notes in the chord, however, the less is told of which scale
the notes are being selected from. The most ambiguous case being the monad.
Example Analysis of a chord progression
The order of the chords in the chord progression produces this, with each
chord defined by it's root, rather than imposing a certain quality;