Article 2021 - Bernadhita

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

EXPECTATION VALUE OF HILBERT SPACE

B. H. S. Utami1,2, M. Usman3, Warsono3, Fitriani3

1
Doctoral Program of Mathematics and Natural Science, Faculty of Mathematics and Natural
Science, University of Lampung, Lampung, 35145 (Indonesia)
2
Department of Information System, STMIK Pringsewu, Lampung, 35373 (Indonesia)
3
Department of Mathematics and Natural Science, Faculty of Mathematics and Natural Science,
University of Lampung, Lampung, 35145 (Indonesia)
e-mail: bernadhitaherindri@yahoo.com

Abstract
This paper is aim to find the expectation value of Hilbert space.

Keywords: probability, Hilbert space, expectation value

1. Introduction

In applied statistics, statistics is defined as the science of making


inferences about the population characteristics. To get population information, of
course, it is done by observation. If the population is too large and it is not
possible to do so, then observations are made on a representative sample.
Sometimes, it is possible for scientists to generate datasets that follow a certain
probability distribution. In general, the population characteristics of concern are
the mean, variance, standard deviation, and proportion (Walpole et al, 2012).
In its application, the expectation value is identical to the mean which is
defined as the sum of the values of each sample observation divided by the
number of samples. Intuitively, we can use the relative frequency method to
calculate the population mean for the edge occurrences for the experiment of
tossing two coins a finite number of times. This value is commonly referred to the
mean value of the random variable X or theoretically called the expectation value
of the random variable X. The formula for continuous random variables is
obtained by approximating with a discrete random variable and noticing that the

1
formula for the expected value is a Riemann sum. Thus, expected values for
continuous random variables are determined by computing an integral.
Wider, the concept of estimating function can be considered as an element
of a Hilbert space. The expected value of random variable 𝒙 is a continuous linear
functional on Hilbert space (Small & McLeish, 1994). The collection of all
random variables with finite variance defined on a given probability space is a
Hilbert space and the properties of the conditional expectation needed to derive
the usual test theory formulas (Zimmerman, 1975). Research on embeddings on
probability measures using the Hilbert space concept was conducted by
Sriperumbudur in 2010. Previously, the expansion of expectation value in a
discrete Hilbert space was deepened by Asplund et al. in 2000. In 2020, Hahn et
al. introduces the Hilbert space averaging method on the construction of typical
states and on the aforementioned conditions of dynamical quantum typicality.
The expectation value approach in the Hilbert space is also carried out by
Klebanov et al. More specifically, his research establishes the analytical properties
of the linear conditional expectation in an infinite-dimensional Hilbert space
context. Inspired by previous research on expectation value in the Hilbert space,
this study discusses the definition of expectation value in the Hilbert space with
unitary elements. This study also gives examples of algebraic propositions
regarding the properties that apply to the expectation value in the Hilbert space.
2. Results
Probabilities are traditionally defined as measures on a sample space,
Small and McLeish define expected value as follows:
Definition 1. Given 𝐻 be a Hilbert space and 𝟏 be an element of 𝐻 then
the pair (𝐻, 𝟏) is a Hilbert space with unitary element if 〈𝟏, 𝟏〉 = 1. In other
words, the scalar multiplies of 𝟏 as constant elements of 𝐻. Furthermore, the
expected value of 𝒙 ∈ 𝐻 is defined by 𝐸(𝒙) = 〈𝒙, 𝟏〉.
These are examples of Hilbert space with unitary element:
Example 1. Let 𝐻 = ℝ3 be vector space with an inner product is defined by
3

〈𝑥, 𝑦〉 = ∑ 𝑥𝑖 𝑦𝑖
𝑖=1
and metric induced by this inner product is

2
1
𝑑(𝑥, 𝑦) = (∑3𝑖=1(𝑥𝑖 − 𝑦𝑖 )2 )2 .
As we know that Hilbert space is an inner product space that is complete under the
metric induced by the inner product (Roman, 2008), first we will show that every
sequence in ℝ3 converges in ℝ3 .
Given (𝑥𝑘 ) is a Cauchy sequence in ℝ3 , where
𝑥𝑘 = (𝑥𝑘,1 , 𝑥𝑘,2 , 𝑥𝑘,3 ).
From induced metric we have,
2
𝑑(𝑥𝑘 , 𝑥𝑚 )2 = ∑3𝑖=1(𝑥𝑘,𝑖 − 𝑥𝑚,𝑖 ) → 0 as 𝑘, 𝑚 → ∞.
For each coordinate position 𝑖, we have
2
(𝑥𝑘,𝑖 − 𝑥𝑚,𝑖 ) ≤ 𝑑(𝑥𝑘 , 𝑥𝑚 )2 .
2
In other words, (𝑥𝑘,𝑖 − 𝑥𝑚,𝑖 ) → 0 as 𝑘, 𝑚 → ∞ and the sequence (𝑥𝑘,𝑖 )𝑘=1,2,⋯ of

𝑖th coordinates is a Cauchy sequence in ℝ.


Since ℝ is complete, we have
(𝑥𝑘,𝑖 ) → 𝑦𝑖 as 𝑘 → ∞.
If 𝑦 = (𝑦1 , 𝑦2 , 𝑦3 ), then
2
𝑑(𝑥𝑘 , 𝑦)2 = ∑𝑛𝑖=1(𝑥𝑘,𝑖 − 𝑦𝑖 ) → 0 as 𝑘 → ∞.
For each coordinate position 𝑖, we have
2
(𝑥𝑘,𝑖 − 𝑦𝑖 ) ≤ 𝑑(𝑥𝑘 , 𝑦)2 → 0.
Thus, we have every Cauchy sequence (𝑥𝑛 ) in ℝ3 is converges to 𝑦 element of
ℝ3 . In other words, ℝ3 is complete and ℝ3 is a Hilbert space.
Then we can find some possible unitary element in ℝ3 ,
1
1 0 0 √3
𝟏 = (0) , 𝟏∗ = (1) , 𝟏∗∗ = (0) , 𝟏∗∗∗ = 1
√3
0 0 1 1
( √3 )
which is multiplied by itself will produce 1. Thus, ℝ3 is Hilbert space with unitary
element.
Now, we will determine the expected value of Hilbert space as below.
For 𝒙 ∈ ℝ3 and unitary element 𝟏, we have
𝐸(𝒙) = 〈𝒙, 𝟏〉

3
3

= ∑ 𝒙 ∙ (1,0,0)
𝑖=1

= (𝑥1 , 𝑥2 , 𝑥3 ) ∙ (1,0,0)
= 𝑥1
For 𝒙 ∈ ℝ3 and unitary element 𝟏∗ , we have
𝐸(𝒙) = 〈𝒙, 𝟏〉
3

= ∑ 𝒙 ∙ (0,1,0)
𝑖=1

= (𝑥1 , 𝑥2 , 𝑥3 )(0,1,0)
= 𝑥2
For 𝒙 ∈ ℝ3 and unitary element 𝟏∗∗ , we have
𝐸(𝒙) = 〈𝒙, 𝟏〉
3

= ∑ 𝒙 ∙ (0,0,1)
𝑖=1

= (𝑥1 , 𝑥2 , 𝑥3 )(0,0,1)
= 𝑥3
For 𝒙 ∈ ℝ3 and unitary element 𝟏∗∗∗ , we have
𝐸(𝒙) = 〈𝒙, 𝟏〉
3
1 1 1
= ∑𝒙∙( , , )
𝑖=1
√3 √3 √3
1 1 1
= (𝑥1 , 𝑥2 , 𝑥3 ) ( , ) ,
√3 √3 √3
𝑥1 𝑥2 𝑥3
= + +
√3 √3 √3
Example 2. Let a Hilbert space of collection of square integrable functions on ℝ𝑛 ,
which is denoted by 𝐿2 (ℝ𝑛 ) and consist of all real-valued measurable functions 𝑓
that satisfy

∫ |𝑓(𝑥)|2 𝑑𝑥 < ∞.
ℝ𝑛

The resulting 𝐿2 (ℝ𝑛 )-norm of 𝑓 is defined by

‖𝑓‖𝐿2 (ℝ𝑛) = √∫ℝ𝑛|𝑓(𝑥)|2 𝑑𝑥.

4
The space 𝐿2 (ℝ𝑛 ) is naturally equipped with the following inner product as
follows:

〈𝑓, 𝑔〉 = ∫ℝ𝑛 𝑓(𝑥)𝑔(𝑥)𝑑𝑥 whenever 𝑓, 𝑔 ∈ 𝐿2 (ℝ𝑛 ).


Which is related to the 𝐿2 -norm since
1⁄
〈𝑓, 𝑓〉 2 = ‖𝑓‖𝐿2 (ℝ𝑛) .
We can define unitary element as
𝟏(𝑥) = 1.
When we operating
1 1

〈𝟏, 𝟏〉 = ∫ 1 𝑑𝑥 ∙ ∫ 1 𝑑𝑥 = 1.
0 0

Thus, 𝐿2 (ℝ𝑛 ) is Hilbert space with unitary element.


For 𝑓 ∈ 𝐿2 (ℝ𝑛 ) and unitary element 𝟏 for all 𝑥 ∈ ℝ𝑛 , we have
𝐸(𝒙) = 〈𝒙, 𝟏〉

= ∫ 𝑓(𝑥) ∙ 𝟏(𝑥)𝑑𝑥
ℝ𝑛

= ∫ 𝑓(𝑥) ∙ 1 𝑑𝑥
ℝ𝑛

= ∫ 𝑓(𝑥)𝑑𝑥
ℝ𝑛

Using the properties of the inner product and unitary element, we obtain a
property which is expressed in the following proposition.
Proposition 1. Let H be a Hilbert space and 𝟏 is unitary element of 𝐻. If 𝒙 and 𝒚
element of 𝐻 then
𝐸(𝒙 + 𝒚) = 𝐸(𝒙) + 𝐸(𝒚).
Proof.
From definition we have,
𝐸(𝒙 + 𝒚) = 〈𝒙 + 𝒚, 𝟏〉.
Recall the linearity property of inner product, we have
𝐸(𝒙 + 𝒚) = 〈𝒙 + 𝒚, 𝟏〉 = 〈𝒙, 𝟏〉 + 〈𝒚, 𝟏〉 = 𝐸(𝒙) + 𝐸(𝒚).

We give an example as below.

5
Example 4. Let 𝐻 = ℝ3 be vector space with an inner product is defined by
3

〈𝑥, 𝑦〉 = ∑ 𝑥𝑖 𝑦𝑖
𝑖=1
and metric induced by this inner product is
1
𝑑(𝑥, 𝑦) = (∑3𝑖=1(𝑥𝑖 − 𝑦𝑖 )2 )2 .
where some possible unitary element in ℝ3 are
1
1 0 0 √3
∗ ∗∗ ∗∗∗ 1
𝟏 = (0) , 𝟏 = (1) , 𝟏 = (0), and 𝟏 = (√3)
0 0 1 1
√3

For 𝒙, 𝒚 ∈ ℝ3 and unitary element 𝟏, we have


𝐸(𝒙 + 𝒚) = 〈𝒙 + 𝒚, 𝟏〉
3

= ∑(𝒙 + 𝒚) ∙ (1,0,0)
𝑖=1

= (𝑥1 + 𝑦1 , 𝑥2 + 𝑦2 , 𝑥3 + 𝑦3 ) ∙ (1,0,0)
= 𝑥1 + 𝑦1
3 3

= ∑ 𝒙 ∙ (1,0,0) + ∑ 𝒚 ∙ (1,0,0)
𝑖=1 𝑖=1

= 〈𝒙, 𝟏〉 + 〈𝒚, 𝟏〉
= 𝐸(𝒙) + 𝐸(𝒚)
The similar way to 𝟏∗ , 𝟏∗∗ , and 𝟏∗∗∗ we have the property as follow.
Although the nature of the addition of two expectation values is general, the
multiplication of two expectation values does not apply to all cases in the Hilbert
space with unitary elements. The following is an example of the expectation value
of multiplying two elements in 𝐻 which is not the same as multiplying two
expectation values.
Example 5. Let 𝐻 = ℝ3 be vector space with an inner product is defined by
3

〈𝑥, 𝑦〉 = ∑ 𝑥𝑖 𝑦𝑖
𝑖=1
and metric induced by this inner product is
1
𝑑(𝑥, 𝑦) = (∑3𝑖=1(𝑥𝑖 − 𝑦𝑖 )2 )2 .
1
√3
1
Given a unitary element is 𝟏 = ( ), we have
√3
1
√3

𝐸(𝒙 ∙ 𝒚) = 〈𝒙 ∙ 𝒚, 𝟏〉 for 𝒙, 𝒚 ∈ ℝ3

6
3
1 1 1
= ∑(𝒙 ∙ 𝒚) ∙ ( , ) ,
𝑖=1
√3 √3 √3
1 1 1
= (𝑥1 𝑦1 , 𝑥2 𝑦2 , 𝑥3 𝑦3 ) ∙ ( ) , ,
√3 √3 √3
1 1 1
= 𝑥1 𝑦1 + 𝑥2 𝑦2 + 𝑥3 𝑦3
√3 √3 √3
On the other hand,
3 3
1 1 1
1 1 1
𝐸(𝒙) ∙ 𝐸(𝒚) = (∑ 𝒙 ∙ ( , , )) ∙ (∑ 𝒚 ∙ ( , , ))
𝑖=1
√3 √3 √3 𝑖=1
√3 √3 √3
1 1 1 1 1 1
= ((𝑥1 , 𝑥2 , 𝑥3 ) ∙ ( )) ∙ ((𝑦1 , 𝑦2 , 𝑦3 ) ∙ ( ,
, , , ))
√3 √3 √3 √3 √3 √3
1 1 1 1 1 1
= ( 𝑥1 + 𝑥2 + 𝑥3 ) ∙ ( 𝑦1 + 𝑦2 + 𝑦3 )
√3 √3 √3 √3 √3 √3
1 1 1
= 𝑥1 𝑦1 + 𝑥2 𝑦2 + 𝑥3 𝑦3
3 3 3
Thus, 𝐸(𝒙 ∙ 𝒚) ≠ 𝐸(𝒙) ∙ 𝐸(𝒚).
Proposition 2. Let H be a Hilbert space and 𝟏 is unitary element of 𝐻. If 𝒙
element of 𝐻 and 𝑎, 𝑏 ∈ ℝ then
𝐸(𝑎𝒙 + 𝑏𝒚) = 𝑎𝐸(𝒙) + 𝑏𝐸(𝒚).
Proof.
From Definition 1,
𝐸(𝑎𝒙 + 𝑏𝒚) = 〈𝑎𝒙 + 𝑏𝒚, 𝟏〉
= 〈𝑎𝒙, 𝟏〉 + 〈𝑏𝒚, 𝟏〉
= 𝑎〈𝒙, 𝟏〉 + 𝑏〈𝒚, 𝟏〉
= 𝑎𝐸(𝒙) + 𝑏𝐸(𝒚)

Here we give an example.
1
√3
1
Example 9. For 𝒙, 𝒚 ∈ ℝ3 and unitary element 𝟏 = ( ), we have √3
1
√3

𝐸(𝑎𝒙 + 𝑏) = 〈𝑎𝒙 + 𝑏, 𝟏〉
3
1 1 1
= ∑(𝑎𝒙 + 𝑏𝒚) ∙ ( , ) ,
𝑖=1
√3 √3 √3
1 1 1
= (𝑎𝑥1 + 𝑏𝑦1 , 𝑎𝑥2 + 𝑏𝑦2 , 𝑎𝑥3 + 𝑏𝑦3 ) ∙ ( , , )
√3 √3 √3

7
1 1 1
=( (𝑎𝑥1 + 𝑏𝑦1 ) + (𝑎𝑥2 + 𝑏𝑦2 ) + (𝑎𝑥3 + 𝑏𝑦3 ))
√3 √3 √3
1 1
= 𝑎(𝑥1 + 𝑥2 + 𝑥3 ) + 𝑏(𝑦1 + 𝑦2 + 𝑦3 )
√3 √3
3 3
1 1 1 1 1 1
= 𝑎∑𝒙∙( , ,
) + 𝑏∑𝒚 ∙ ( , , )
𝑖=1
√3 √3 √3 𝑖=1
√3 √3 √3
= 𝑎〈𝒙, 𝟏〉 + 𝑏〈𝒚, 𝟏〉
= 𝑎𝐸(𝒙) + 𝑏𝐸(𝒚)

1
As simulation, we generate data considering 𝑛 is a convergent Cauchy sequence in
ℝ. Let 𝐻 = ℝ be vector space with an inner product is defined by
〈𝒙, 𝒚〉 = 𝑥 ∙ 𝑦
and metric induced by this inner product is
𝑑(𝑥, 𝑦) = √𝑥 − 𝑦
and 1 as unitary element of ℝ. From definition of expectation value in ℝ, we have
𝐸(𝒙) = 〈𝒙, 𝟏〉 = 4.27854. Thus, we make a normal probability plot to see
whether the data is normally distributed or not as shown in the following figure.

Normal Probability Plot


3

0
Z-values

0 0.2 0.4 0.6 0.8 1 1.2


-1

-2

-3

-4
Data values

Figure 1. Normal probability plot of data


Based on the plot in Figure 1, the points do not follow a straight line. Thus, the
data can be said to be not normally distributed and more likely to have an
exponential distribution.

8
References
Busemeyer, J. R., & Wang, Z. (2018). Hilbert space multidimensional theory.
Psychological Review, 125(4), 572–591. https://doi.org/10.1037/rev0000106
Hahn, N., Guhr, T., & Waltner, D. (2020). Hilbert space average of transition
probabilities. Physical Review E, 101(6), 1–19.
https://doi.org/10.1103/PhysRevE.101.062135
Klebanov, I., Sprungk, B., & Sullivan, T. J. (2020). The linear conditional
expectation in Hilbert space. 1–32. http://arxiv.org/abs/2008.12070
Roman, S. (2008). Advanced Linear Algebra. Springer.
Smalll, C. G., & McLeish, D. L. (1994). Hilbert Space Methods in Probability
and Statistical Inference. John Wiley and Sons, Inc.
Sriperumbudur, B. K., Gretton, A., Fukumizu, K., Schölkopf, B., & Lanckriet, G.
R. G. (2010). Hilbert Space Embeddings and Metrics on Probability
Measures. Journal of Machine Learning Research, 11, 1517–1561.
Walpole, Ronald E.;Myers, Raymond H.; Myers, Sharon L; Ye, K. (2012).
Probability and Statistics for Engineers and Scientistist (9th ed.). Prentice
Hall Pearson.
Zimmerman, D. W. (1975). Probability Space, Hilbert Space, and The Axioms of
Test Theory. Psychometrika, 40(3), 395–412.

You might also like