Is the golden ratio a universal constant for self-replication? (2024)

Journal List
PLoS One
PMC6047800

As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice

PLoS One. 2018; 13(7): e0200601.

Published online 2018 Jul 16. doi:10.1371/journal.pone.0200601

PMCID: PMC6047800

PMID: 30011316

Yu Liu, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing^¤^* and David J. T. Sumpter, Methodology, Project administration, Supervision, Writing – review & editing

Danilo Roccatano, Editor

Author information Article notes Copyright and License information PMC Disclaimer

Associated Data

Supplementary Materials

Data Availability Statement

Abstract

The golden ratio, ϕ = 1.61803…, has often been found in connection with biological phenomena, ranging from spirals in sunflowers to gene frequency. One example where the golden ratio often arises is in self-replication, having its mathematical origins in Fibonacci’s sequence for “rabbit reproduction”. Recently, it has been claimed that ϕ determines the ratio between the number of different nucleobases in human genome. Such empirical examples continue to give credence to the idea that the golden ratio is a universal constant, not only in mathematics but also for biology. In this paper, we employ a general framework for chemically realistic self-replicating reaction systems and investigate whether the ratio of chemical species population follows “universal constants”. We find that many self-replicating systems can be characterised by an algebraic number, which, in some cases, is the golden ratio. However, many other algebraic numbers arise from these systems, and some of them—such as $\sqrt[3]{2} = 1.25992 \dots$ and 1.22074… which is also known as the 3rd lower golden ratio—arise more frequently in self-replicating systems than the golden ratio. The “universal constants” in these systems arise as roots of a limited number of distinct characteristic equations. In addition, these “universal constants” are transient behaviours of self-replicating systems, corresponding to the scenario that the resource inside the system is infinite, which is not always the case in practice. Therefore, we argue that the golden ratio should not be considered as a special universal constant in self-replicating systems, and that the ratios between different chemical species only go to certain numbers under some idealised scenarios.

Introduction

The golden ratio $(1 + \sqrt{5}) / 2 = 1.61803 \dots$ , usually denoted by the Greek letter ϕ (PHI), has attracted broad attention for a long time, with suggestions that they are aesthetically significant and ubiquitous in nature. These claims were made particularly prominent in Dan Brown’s book (and later film) The Da Vinci Code: (quote) Despite PHI’s seemingly mystical mathematical origins, Langdon explained, the truly mind-boggling aspect of PHI was its role as a fundamental building block in nature. Plants, animals, and even human beings all possessed dimensional properties that adhered with eerie exactitude to the ratio of PHI to 1… “PHI’s ubiquity in nature,” Langdon said, killing the lights, “clearly exceeds coincidence, and so the ancients assumed the number PHI must have been preordained by the Creator of the universe…” [1]

The basis of Dan Brown’s fictional discussion of the golden ratio ϕ originated from “scientific” studies (as documented by paper [2] and [3]). For example, the Great Pyramid of Khufu was claimed to be designed by ancient Egyptians to manifest ϕ, i.e., the ratio of the slant height to half of the length of the base was designed to be ϕ. Moreover, nautilus shell was believed to be a logarithmic spiral whose growth factor is ϕ. Other examples include the design of the Parthenon, Leonardo da Vinci’s paintings, the ratios in a human body, lengths of passages in the Aeneid.

These claims are, however, mostly false or misleading [2–4]. For example, although today the ratio of the slant height to half of the length of the base of the Great Pyramid happened to be ϕ approximately, the height was quite different when it was built over four thousand years ago, and there is no evidence that ancient Egyptians even knew the golden ratio [2]. A field study by Falbo showed that nautilus shell is quite far from a golden spiral [3]. For the Parthenon, there are many different numbers of its dimensions in the literature (probably because measurements are made between different points by different people), allowing a golden ratio enthusiast to choose whatever numbers to fit ϕ [2]. But sadly, despite the dubious nature of the claims, the golden ratio even remains a feature of some school textbooks, such as [5].

One area of biology in which the golden ratio has a genuine role is in phyllotaxis, i.e., the arrangement of leaves on the stem of a plant. The seeds in the head of a ripe sunflower fit neatly together into two interlocking families of helical spirals. The number of clockwise and anti-clockwise winding spirals are often both successive numbers of the Fibonacci sequence [6]. For example, a sunflower might have 34 clockwise spirals and 21 anti-clockwise spirals, where 34 and 21 are two successive numbers in the Fibonacci sequence, i.e., 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, …, which satisfies

X_t =X_t-1 +X_t-2

(1)

The limit of the ratio between two successive Fibonacci numbers is the golden ratio, namely, lim_t→∞X_t/X_t−1 = ϕ. These sequences are genuine biological phenomenon, not just in sunflowers, but in plants such as pinecones, aloe, pineapples, and cacti [6–8]. It should be stressed however that non-Fibonacci phyllotaxy is also abundant [7, 9]. Several theories have been proposed to explain the biochemical and mechanical mechanisms for these patterns (reviewed by paper [10]), including Alan Turing’s morphogenesis theory [11, 12].

The Fibonacci process Eq (1) corresponds to self-replicating processes, having its origin in the growth pattern of the “Fibonacci rabbit”, an idealised scenario first considered by the Italian mathematician Fibonacci in his book Liber Abaci (1202). The idea is as follows: rabbits never die; it takes one month for a pair of infant rabbits to become a pair of adults; an adult pair always gives birth to an infant pair; the system starts with one pair of adult rabbits. This gives rise to Eq (1), where X_t is the number of adult pairs at month t, and the number of infant pairs at month t is X_t−1. So the ratio of the number of adult pairs over the number of infant pairs goes to ϕ as t → ∞.

Recently, the golden ratio was suggested to appear in another totally different field: the human genome. Dress et al. proposed that the growth pattern of repetitive DNAs is analogous to the pattern described by the Fibonacci process Eq (1) (repetitive DNAs are those built from a basic short DNA sequence that is repeated many times, often referred to as “junk DNA” and accounted for a large fraction of the whole human genome) [13]. Yamagishi and Shimabukuro argued that the distribution of the values of P_A + P_C for the 24 human chromosomes can also be explained by the Fibonacci process Eq (1) where P_A is the frequency of codon A and likewise, P_C is the frequency of codon C [14]. Moreover, Perez calculated that in the single-stranded whole human DNA, (P_C + P_G)/(P_AP_T) is approximately (3 − ϕ)/2 [15]. These empirical examples continue to give credence to the idea that ϕ is a universal constant, not only in mathematics but also for biology.

Another famous model that can generate ϕ is the L-system (a type of formal grammar) [16]. The specific L-system that generates ϕ has the following rules: for each generation, each Q becomes a Q and an S, while each S becomes a Q. The system is initialised with a string composed of only one Q. So at the first generation, the string becomes QS; at the second generation, the string becomes QSQ; then QSQQS, QSQQSQSQ, … The number of Q at each generation forms the Fibonacci sequence (as well as the number of S but with one generation lag). So the ratio of the number of Q over the number of S goes to ϕ. This specific L-system can actually be written as the following “reaction system”,

${\begin{array}{l} Q & \to Q + S \\ S & \to Q \end{array}$

(2)

This process is equivalent to the Fibonacci rabbit model. Note that Eq (2) also involves self-replication, because the two reactions add up to Q → Q + Q (by cancelling S on both sides) and Q is thus self-replicating. Similarly, the genealogical tree model [17, 18] which can also generate ϕ has the same argument.

These models do, from a mathematical viewpoint, give rise to the golden ratio, but our question is whether the golden ratio has a special role in natural self-replicating systems in general, as implied by the authors in papers [13–15]. Should we expect, in a wide range of different self-replicating systems, to find that the ratio of different components in the system is the golden ratio? And if not, are there other important mathematical constants that we should expect to see in connection to self-replicating systems?

To answer these questions we employ a framework of chemically realistic self-replicating reaction systems, introduced by us in paper [19]. This theory has practical significance to real chemistry and biology, and instantiations of this framework can capture a wide range of different natural chemical systems such as the citric acid cycle and the formose reaction. In this theory, a self-replicating system consists of a few chemical reactions, and self-replication is a system-level property: each reaction in the system is not able to self-replicate but the system self-replicates as a whole. If the golden ratio ϕ does have a special role in natural self-replicating systems in general, we would expect to see ϕ appears quite a lot in these systems.

This paper is organised as follows. Section Materials and methods reintroduces the framework in paper [19] for chemically realistic self-replicating reaction systems. The first part of Section Results elaborates on the scheme for population dynamics that we apply on self-replicating systems, and describe two self-replicating systems that are characterised by ϕ. In the second part of Section Results, we show that all of the analysable self-replicating systems up to a certain size are characterised by a limited number of distinct characteristic equations and numbers. In the third part of Section Results, we introduce the population dynamics for chemical reaction systems under the law of mass action, and find that the characteristic number of a self-replicating system is a transient behaviour corresponding to the scenario that the resource inside the system is infinite, which is not always the case in practice.

Materials and methods

Chemically realistic self-replicating reaction systems

In an earlier paper [19], we described a general framework to construct chemically realistic reaction systems. In this framework, a molecule is defined by its integer “mass”, i, and thus denoted $\bar{i}$ ; all reactions that conserve mass—the total mass on the reactant side adds up to those on the product side—are possible; only two types of reaction are allowed, synthesis of two molecules to create a molecule of greater mass (e.g., $\bar{2} + \bar{4} \to \bar{6}$ ) and decomposition into two molecules to create two molecules of lower mass (e.g., $\bar{6} \to \bar{1} + \bar{5}$ ); every reaction is chemically spontaneous. The last requirement constrains the details of the thermodynamic free energy of each molecule and reaction, which we do not need to consider in this paper, but it does have two consequences related to this paper. The first one is that a reaction and its reverse reaction, e.g., $\bar{2} + \bar{4} \to \bar{6}$ and $\bar{6} \to \bar{2} + \bar{4}$ , cannot both appear in one system. The second consequence will be introduced in the next paragraph.

By choosing some reactions that satisfy the requirements above, we can construct a chemically realistic reaction system, namely, an instantiation of the general framework. The following system is one of such instantiations,

$\begin{matrix} {\begin{matrix} \begin{matrix} \bar{2} & \to \bar{1} + \bar{1} \\ \bar{3} & \to \bar{1} + \bar{2} \\ \bar{1} + \bar{4} & \to \bar{5} \\ \bar{5} & \to \bar{2} + \bar{3} \end{matrix} \end{matrix} \end{matrix}$

(3)

We define the resource molecule to be the molecule that only appears on the reactant side ( $\bar{4}$ in this case). For any chemically realistic self-replicating system, there is at least one type of resource molecule. This is the second consequence of the “chemical spontaneity” requirement above. We define an intermediate molecule to be any molecule that appears on both the reactant side and the product side ( $\bar{1}$ , $\bar{2}$ , $\bar{3}$ and $\bar{5}$ in this case), and the waste molecule to be the molecule that only appears on the product side (there is no in this case).

In paper [19], we show that a chemically realistic reaction system is self-replicating if the following three criteria are satisfied:

1
For every reaction, at least one type of its reactants comes from the products of other reactions;
2
There is at least one intermediate molecule that appears on the reactant side fewer times than that on the product side;
3
There are no intermediate molecules that appear on the reactant side more often than on the product side.

In this paper we introduce an alternative, narrower criterion under which certain chemically realistic self-replicating systems are analysable. We say that a self-replicating system is analysable if it satisfies criterion 1, 2, and the following criterion 3*:

3*
On the reactant side of the whole system, each intermediate molecule species appears at most once.

Note that system Eq (3) satisfies all of criterion 1, 2 and 3*, and thus is an analysable self-replicating chemical reaction system.

Results

Systems characterised by the golden ratio ϕ

We now introduce a scheme for population dynamics for analysable self-replicating chemical reaction systems, using system Eq (3) as an example. Let N_i(t) be the population of molecule species $\bar{i}$ at generationt. We initialise the system with N₁(0) = N₂(0) = N₃(0) = N₅(0) = 1, i.e., one of each intermediate molecules. In addition, we assume that there is an infinite number of resource molecules $\bar{4}$ inside the system, namely N₄(t) = ∞ for all t.

We update molecule populations at generationt + 1 from t as follows: for each molecule species $\bar{i}$ , find the unique reaction that has $\bar{i}$ on the reactant side; and all of molecules $\bar{i}$ at generation t—namely N_i(t) of $\bar{i}$ —are transformed into the products of this corresponding reaction. The criterion 3* above guarantees that there is only one unique reaction that has molecule $\bar{i}$ on the reactant side.

Consider how we apply this updating procedure to system Eq (3):

Firstly, update N_i(0) to obtain N_i(1). For molecule $\bar{1}$ , due to the fact that the third reaction is the only reaction that has molecule $\bar{1}$ on the reactant side and N₁(0) = 1, we let this one of $\bar{1}$ transform into one of $\bar{5}$ through the third reaction. For molecule $\bar{2}$ , due to the fact that the first reaction is the only reaction that has molecule $\bar{2}$ on the reactant side and N₂(0) = 1, we let this one of $\bar{2}$ transform into two of $\bar{1}$ through the first reaction. For molecule $\bar{3}$ , because the second reaction is the unique reaction and N₃(0) = 1, we let this one of $\bar{3}$ transform into one of $\bar{1}$ and one of $\bar{2}$ . For molecule $\bar{5}$ , because the fourth reaction is the unique reaction and N₅(0) = 1, we let this one of $\bar{5}$ transform into one of $\bar{2}$ and one of $\bar{3}$ . Therefore, at generation t = 1, we have N₁(1) = 3, N₂(1) = 2, N₃(1) = 1 and N₅(1) = 1.
Then, update N_i(1) to obtain N_i(2). N₁(1) = 3 of molecule $\bar{1}$ are transformed into three of $\bar{5}$ through the third reaction; N₂(1) = 2 of $\bar{2}$ are transformed into four of $\bar{1}$ through the first reaction; N₃(1) = 1 of $\bar{3}$ is transformed into one $\bar{1}$ and one $\bar{2}$ through the second reaction; and N₅(1) = 1 of $\bar{5}$ is transformed into one $\bar{2}$ and one $\bar{3}$ through the fourth reaction. Therefore, at generation t = 2, we have N₁(2) = 5, N₂(2) = 2, N₃(2) = 1 and N₅(2) = 3.
Similarly, at generation t = 3, we have N₁(3) = 5, N₂(3) = 4, N₃(3) = 3 and N₅(3) = 5.
Continue this updating procedure. Finally we obtain a sequence of N₁(t), N₂(t), N₃(t) and N₅(t).

Table 1 shows the result over six generations. Note that this scheme can be applied to any analysable self-replicating system, through the following steps in general: (1) assume an infinite number of resource molecules inside the system; (2) initialise the system with one of each intermediate molecule; and (3) update molecule populations by generations to obtain sequences of molecule populations. This scheme defines a particular population dynamics for self-replicating systems.

Table 1

Molecule populations N_i(t) for the self-replicating chemical reaction system Eq (3) at generation t.

Generation t	N₁(t)	N₂(t)	N₃(t)	N₅(t)
0	1	1	1	1
1	3	2	1	1
2	5	2	1	3
3	5	4	3	5
4	11	8	5	5
5	21	10	5	11
6	25	16	11	21
⋮				⋮

Distribution of the characteristic number Λ

We now investigate a larger range of chemically realistic self-replicating reaction systems in order to determine which numbers typically characterise their behaviour. We consider all of the analysable self-replicating systems up to where the possibly largest molecule is $\bar{7}$ . We exhaustively checked every possible reaction system up to $\bar{7}$ to see whether it satisfies criterion 1, 2 and 3* in Section Materials and methods. Then we found 162 analysable self-replicating systems in total (currently we cannot go beyond $\bar{7}$ because the number of possible reaction systems increases so fast that the computation time becomes too long). The distribution of the characteristic numbers Λ for all of them is shown in Fig 2.

Open in a separate window

Fig 2

Distribution of the characteristic numbers Λ for all the analysable self-replicating chemical reaction systems up to where the possibly largest molecule is $\bar{7}$ .

The blue lines correspond to all integer self-replicating systems, while the red bars correspond to all other systems.

The most frequent characteristic number is $Λ = 1.25992 \dots = \sqrt[3]{2}$ , which appeared 28 times. The following system Eq (12) is one such example,

$\begin{matrix} {\begin{matrix} \begin{matrix} \bar{1} + \bar{2} & \to \bar{3} \\ \bar{1} + \bar{3} & \to \bar{4} \\ \bar{4} & \to \bar{2} + \bar{2} \end{matrix} \end{matrix} \end{matrix}$

(12)

where molecule $\bar{1}$ is the resource. We have the recurrence relation

$\begin{matrix} (\begin{matrix} N_{2} (t + 1) \\ N_{3} (t + 1) \\ N_{4} (t + 1) \end{matrix}) = (\begin{matrix} 0 & 0 & 2 \\ 1 & 0 & 0 \\ 0 & 1 & 0 \end{matrix}) (\begin{matrix} N_{2} (t) \\ N_{3} (t) \\ N_{4} (t) \end{matrix}) \end{matrix}$

(13)

where the 3 × 3 matrix is the recurrence matrix A. Table 2 shows N_i(t) for several generations. We find that the population of any molecule species doubles every three generations, i.e.,

N_i(t +3) =2N_i(t) holds for any t, and i =2,3,4.

Table 2

Molecule populations N_i(t) for the self-replicating chemical reaction system Eq (12) at generation t.

Generation t	N₂(t)	N₃(t)	N₄(t)
0	1	1	1
1	2	1	1
2	2	2	1
3	2	2	2
4	4	2	2
5	4	4	2
6	4	4	4
⋮			⋮

Open in a separate window

Equivalently, N(t + 3) = 2N(t) holds for any t, where N(t) = (N₂(t), N₃(t), N₄(t))^⊺. We call this type of system as the integer self-replicating system, namely, those satisfying that

N(t +p) =A^p N(t) =RN(t) holds for any t,

where N(t) is the m × 1 vector for populations of intermediate molecules at generation t, the periodp is a positive integer, and the growth rateR is a positive integer. We further find that all of the eigenvalues of A in Eq (13) have the same absolute value and this value is $\sqrt[p]{R}$ ( $\sqrt[3]{2}$ in this case), which also means that the characteristic number $Λ = \sqrt[p]{R}$ . It is a special property compared with other systems such as Eqs (3) and (10) where the absolute values of the eigenvalues of their corresponding recurrence matrix are not all equal.

System Eq (12) is just one specific integer self-replicating system. The blue lines in Fig 2 represent all integer self-replicating systems, which are quite common. Their characteristic numbers, from small to large, are $1, \sqrt[4]{2}, \sqrt[3]{2}, \sqrt[4]{3}, \sqrt[4]{4}, \sqrt[3]{3}$ , and $\sqrt[3]{4}$ , respectively.

The second most frequent characteristic number in Fig 2 is Λ = 1.22074… This is known as the 3rd lower golden ratio, because of the fact that Λ³ = 1 + 1/Λ. In general, the number x that satisfies x^k = 1 + 1/x where k is an integer is proposed to be the kth lower golden ratio [21]. The following system is one example system characterised by the 3rd lower golden ratio.

${\begin{matrix} \begin{matrix} \bar{3} & \to \bar{1} + \bar{2} \\ \bar{4} & \to \bar{1} + \bar{3} \\ \bar{1} + \bar{5} & \to \bar{6} \\ \bar{6} & \to \bar{2} + \bar{4} \end{matrix} \end{matrix}$

where molecule $\bar{5}$ is the resource. It has the recurrence relation

$(\begin{matrix} N_{1} (t + 1) \\ N_{3} (t + 1) \\ N_{4} (t + 1) \\ N_{6} (t + 1) \end{matrix}) = (\begin{matrix} 0 & 1 & 1 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 1 \\ 1 & 0 & 0 & 0 \end{matrix}) (\begin{matrix} N_{1} (t) \\ N_{3} (t) \\ N_{4} (t) \\ N_{6} (t) \end{matrix})$

The eigenvalues of the recurrence matrix are 1.22074…, −0.24812… ± j1.03398…, and −0.72449…, respectively.

Note that the 1st lower golden ratio (namely, ϕ) appears three times; the 2nd lower golden ratio 1.32471… (also called the plastic number) appears four times; and the 4th lower golden ratio 1.16730… appears once (Fig 2).

In order to explain why certain characteristic numbers arise, we note that the m × m recurrrence matrix A = (a_jk) for any analysable self-replicating chemical reaction system must satisfy the following conditions:

It is a non-negative integer square matrix. This is because any reaction involves only integer number of molecules.
All entries on the main diagonal are 0, namely a_jj = 0. This is because no molecule can appear on both sides of a reaction (e.g., $\bar{2} + \bar{3} \to \bar{2}$ is not allowed, since the mass has to be conserved).
a_jk and a_kj cannot be both larger than 0, that is, a_jk ⋅ a_kj = 0. This is because a reaction and its reverse reaction cannot both appear in one system.
The sum of any column is either 0, 1 or 2. This is because of criterion 3* in Section Materials and methods and also because each reaction produces no more than two molecules.
The sum of at least one column is 2. This is because of criterion 2 in Section Materials and methods.
The sum of any row is at least 1. This is because of criterion 1 in Section Materials and methods.

Therefore, the characteristic equation of A has the general form (referring to S1 Appendix for the derivation):

λ^m -σ₃λ^m-3 +σ₄λ^m-4 -⋯ +(-1)^m-1σ_m-1λ +(-1)^mσ_m =0

where σ₃, σ₄, ⋯, σ_m are all integers. Note that (1) this polynomial is always monic, i.e., the leading coefficient is 1; (2) the term λ^m−1 is zero because of condition 2 of A mentioned above; (3) the term λ^m−2 is zero because of condition 3 of A mentioned above.

For all of the chemically realistic self-replicating systems we investigated (where the largest molecule is $\bar{7}$ ), there are at most six types of intermediate molecules (since at least one molecule species must be the resource). Therefore, we only have four cases: For any 3-intermediate-molecule system, the characteristic equation is

λ³ -det(A) =0

(14)

For any 4-intermediate-molecule system, the characteristic equation is

λ⁴ -σ₃λ +det(A) =0

For any 5-intermediate-molecule system, the characteristic equation is

λ⁵ -σ₃λ² +σ₄λ -det(A) =0

For any 6-intermediate-molecule system, the characteristic equation is

λ⁶ -σ₃λ³ +σ₄λ² -σ₅λ +det(A) =0

Here we list four properties of the characteristic number Λ for self-replicating systems:

Any Λ appeared in Fig 2 is the largest roots of either of the four characteristic equations above, which also means that any Λ is an algebraic number.
The self-replicating systems that is characterised by the golden ratio ϕ have at least four types of intermediate molecules.
From Eq (14), we see that for any 3-intermediate-molecule system, the absolute values of all of A’s eigenvalues are equal, and thus $Λ = \sqrt[3]{det (A)}$ . Therefore, any 3-intermediate-molecule self-replicating system is an integer self-replicating system with p = 3 and R = det(A).
From our simulations, we observed that for any integer self-replicating system, some of its recurrence matrix A’s eigenvalues might be zero and all of other eigenvalues have the same absolute value $\sqrt[p]{R}$ where p and R are some positive integers. Based on this observation, we propose the following hypothesis.
Hypothesis: If any m × m matrix A satisfies all the six conditions above, and also satisfies that there exist positive integers p and R, and a non-negative and non-zero m × 1 vector N such that A^pN = RN, then
$| λ_{1} | = | λ_{2} | = \dots = | λ_{u} | = \sqrt[p]{R} and λ_{u + 1} = λ_{u + 2} = \dots = λ_{m} = 0$
where λ₁, λ₂, ⋯, λ_m are the m eigenvalues of A, and 1 ≤ u ≤ m.
This hypothesis holds numerically for all the cases we have studied, and is a very striking result, but we have been unable to prove it rigorously. We can construct a specific type of graph that satisfies all the six conditions for A (each vertex of the graph represents each type of molecule in the system while each edge is assigned the value of the corresponding entry of A), and then prove that this type of graph has the properties the hypothesis says. But we have not been able to prove that these conditions for A guarantee this type of graph (personal communication with Volodymyr Mazorchuk [22]).

Population dynamics under the law of mass action

The scheme Eq (5) we applied on analysable self-replicating systems corresponds to a particular population dynamics. However, this population dynamics is inconsistent with most chemical reactions in physical scenarios: Based on the law of mass action, the rate of a chemical reaction is, in general, equal to the arithmetic product of a predefined reaction rate constant and the concentrations of reactants. The rate constant depends on physical properties of the chemicals involved, while the concentrations depend on the physical conditions under which the reaction occurs.

In paper [19], we derived formulas for reaction rates under the law of mass action, in physical scenarios: (1) all molecules are ideally gaseous; (2) the self-replicating system is a well-mixed system and kept in a box under constant pressure and temperature; (3) the resource molecule population in this box is kept constant but finite, achieved by a presumed external unlimited reservoir of resource molecules. Under the law of mass action, for a synthesis reaction $\bar{i} + \bar{j} \to \bar{i + j}$ , the reaction rate is (in unit s⁻¹)

γ_+ij =ω_+ij·N_i·N_j/N

where ω_+ij is the rate constant for this synthesis reaction, N_i = N_i(τ) is the population of molecule $\bar{i}$ in the box at time τ, and N = N(τ) is the total population of all the molecules in the box at time τ. Note that the variable “time” τ is continuous, representing the physical time, which is different from the abstract discrete variable “generation” t in the former population dynamics Eq (5). For a decomposition reaction $\bar{i + j} \to \bar{i} + \bar{j}$ , the reaction rate is (in unit s⁻¹)

γ_-ij =ω_-ij·N_i+j

where ω_−ij is the rate constant for this decomposition reaction. The physical conditions and the properties of chemicals assumed in paper [19] guarantee that rate constants ω_+ij or ω_−ij for all reactions in the self-replicating system are identical, thus denoted ω. Finally, we use ordinary differential equations (ODEs) to describe its population dynamics.

We now investigate system Eq (3), which is characterised by the golden ratio ϕ under the population dynamics Eq (5) we used previously. Here, the mass action population dynamics for this system are

$\begin{matrix} {\begin{matrix} \begin{matrix} d N_{1} / d τ & = - ω U N_{1} / N + 2 ω N_{2} + ω N_{3} \\ d N_{2} / d τ & = - ω N_{2} + ω N_{3} + ω N_{5} \\ d N_{3} / d τ & = - ω N_{3} + ω N_{5} \\ d N_{5} / d τ & = - ω N_{5} + ω U N_{1} / N \end{matrix} \end{matrix} \end{matrix}$

(15)

where U is the constant population of resource molecule $\bar{4}$ in the box, and N = U + N₁ + N₂ + N₃ + N₅. There are only two parameters. We set ω = 1.09986 × 10¹¹s⁻¹, the same as in paper [19], which is determined by the chemical properties and physical conditions. Note that in physical scenarios, U is always finite. We arbitrarily set U = 10¹⁰.

Solutions of Eq (15), with initial condition N₁(0) = N₂(0) = N₃(0) = N₅(0) = 1, are shown in Fig 3a, in log-normal scale. After a transient period in the beginning, the populations go into an exponential growth phase (approximately from 0.5 to 2.5 × 10⁻¹⁰s). The straight line in log-normal scale implies an exponential growth. After this phase, growing slows down. Fig 3b shows the normalised molecule population n_i = N_i/(N₁ + N₂ + N₃ + N₅), to compare with Fig 1. We observe that lim_τ→∞(N₁(τ)/N₂(τ)) ≡ lim_τ→∞(n₁(τ)/n₂(τ)) = ∞ which is not ϕ.

Open in a separate window

Fig 3

The mass action population dynamics of system Eq (3) in a physical scenario.

(a) Solutions of Eq (15) in log-normal scale, i.e., x-axis is in normal scale and y-axis is in logarithmic scale. (b) The normalised molecule populations for the same period as in (a).

During the exponential growth phase, however, the normalised population n_i remains almost unchanged (Fig 3b). We further find that n_i during this phase is approximately identical to the corresponding limit value in Fig 1. That is, during the exponential growth phase in Fig 3b, n₁ ≈ 0.38196, n₂ = n₅ ≈ 0.23606, n₃ ≈ 0.14589, and thus n₁/n₂ ≈ ϕ. By solving Eq (15) for various values of U, we find that the exponential growth phase always occurs, and only occurs when the sum of populations of intermediate molecules is much smaller than population of the resource molecule, namely, U/(U + N₁ + N₂ + N₃ + N₅) → 1. That is why, in Fig 3a, the exponential growth phase stops around 3 × 10⁻¹⁰s, when N₁ + N₂ + N₃ + N₅ becomes of comparable magnitude of U. Indeed, if we let U/(U + N₁ + N₂ + N₃ + N₅) → 1, Eq (15) becomes a linear ODE system, and its solutions are exponential functions. We thus have

$\begin{matrix} lim_{τ \to \infty} \frac{N_{1} (τ)}{N_{2} (τ)} = ϕ, if \frac{U}{U + N_{1} + N_{2} + N_{3} + N_{5}} \to 1 \end{matrix}$

Therefore, the characteristic number of the system are transient behaviours corresponding to the scenario that all other molecules are much fewer than the resource inside the system, equivalently, the resource inside the system is infinite. It is interesting to note that the Jacobian of the ODEs Eq (15) at the fixed point N₁ = N₂ = N₃ = N₅ = 0 is (A − I), where A is the recurrence matrix for this self-replicating system Eq (3), namely, A in Eq (4) (it is in fact a general case that for an analysable self-replicating system, the Jacobian at the zero fixed point of its corresponding ODEs is always (A − I) where A is its corresponding recurrence matrix). In physical scenarios, infinite resource is not possible, so we do not often expect the characteristic number of a self-replicating system to manifest itself.

The transitory nature of characteristic numbers in growth is not the only reason to doubt their universal significance. Note that we also made a strong assumption about the rate constants above that they are all identical, namely, ω_ij = ω. As rate constants are determined by the properties of involved chemicals, different reactions are very unlikely to have the same rate constant, in more realistic situations. This observation adds another reason we would not expect ϕ to be widely observed in real biological systems.

There is yet another factor that limits the generality of the characteristic number of self-replicating systems. Take the following system Eq (16) as an example,

$\begin{matrix} {\begin{matrix} \begin{matrix} \bar{1} + \bar{1} & \to \bar{2} \\ \bar{3} & \to \bar{1} + \bar{2} \\ \bar{5} & \to \bar{1} + \bar{4} \\ \bar{2} + \bar{6} & \to \bar{8} \\ \bar{8} & \to \bar{3} + \bar{5} \end{matrix} \end{matrix} \end{matrix}$

(16)

where molecule $\bar{6}$ is the resource. Under the mass action population dynamics, this system evolves as in Fig 4. However, it is not analysable by the population dynamics Eq (5): According to criterion 1, 2 and 3 in Section Materials and methods, system Eq (16) is a chemically realistic self-replicating system, but intermediate molecule $\bar{1}$ appears twice on the reactant side, which violates criterion 3*. Therefore, it is not analysable by Eq (5), and it thus has no such recurrence matrix. This stricter criterion 3* is essential in allowing us to apply Eq (5). But, it greatly limits the number of systems that we can look at.

Open in a separate window

Fig 4

The mass action population dynamics of system Eq (16) in a physical scenario.

(a) Solutions of its corresponding ODEs in log-normal scale. Note that populations of molecule $\bar{3}$ , $\bar{5}$ and $\bar{8}$ are always the same. (b) The normalised molecule populations for the same period as in (a).

The factors listed above, combined with the low number of analysable systems which give rise to ϕ, lead us to dismiss the idea that the golden ratio, or any other constant, provides any form of universal characterisation of self-replicating systems.

Conclusion

A wide range of chemically realistic self-replicating systems can be described by recurrence matrices that can be further characterised by an algebraic number, which is the largest absolute value among all of the eigenvalues of the recurrence matrix. In some cases, the characteristic number is the golden ratio ϕ. Yet, ϕ is just one of the many numbers that might appear. These could be otherwise an nth root of an integer, a generalised golden ratio or certain other algebraic numbers. The characteristic number of a self-replicating system is a transient behaviour, corresponding to the scenario that the resource inside the system is infinite. Moreover, while many systems can be characterised by an algebraic number, there remain many more that cannot be characterised in this way.

Our work suggests that there is no reason to believe that ϕ or any other specific number will characterise self-replicating systems. We do not look at data from any specific biological system here, such as ratios of codon frequencies in DNA, but we rather argue that when we study the vast array of all biological systems, then constants such as ϕ appear only in a very limited range of situations. It would thus be highly unlikely that the specific, complex self-replicating system that underlies, for example, our DNA is one such system. Our results contradict the claim of some other authors (e.g., in paper [15]) that ϕ is universal.

We conclude that (1) ϕ has its particular position in chemically realistic self-replicating reaction systems in general but it is not that special since many other constants also appear, and (2) that these constants, including ϕ, appear in chemically realistic self-replicating systems only when there is an infinite number of resources inside the system, which is not usually the case in practice. Whenever the golden ratio itself, or some linear function of the golden ratio, is found to characterise the ratio of chemicals in a system, the most likely explanation is that the relationship occurred by chance.

We suggest, instead, that a more useful approach is to develop models that characterise self-replication and investigate their general properties. We see our chemical system and approach as a useful step in this direction.

Supporting information

S1 Appendix

Characteristic polynomial of a matrix.

(PDF)

Click here for additional data file.^{(104K, pdf)}

Acknowledgments

Y. L. thanks Erasmus Mundus Action 2 programme (Lotus Scholarship), funded by the European Commission, for support of his Ph.D. study in Uppsala University, where the work on this paper was undertaken.

Funding Statement

Erasmus Mundus Action 2 programme (Lotus Scholarship, http://www.lotus.ugent.be/index.asp?p=1131&a=1131), funded by the European Commission, supported Yu Liu for his Ph.D. study in Uppsala University, where the work on this paper was undertaken. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability

All relevant data are within the paper.

References

1. Brown D. The Da Vinci Code. Doubleday; 2003. [Google Scholar]

2. Markowsky G. Misconceptions about the golden ratio. The College Mathematics Journal. 1992;23:2–19. 10.2307/2686193 [CrossRef] [Google Scholar]

3. Falbo C. The golden ratio–a contrary viewpoint. The College Mathematics Journal. 2005;36:123–134. 10.1080/07468342.2005.11922119 [CrossRef] [Google Scholar]

4. Fischler R. How to find the golden number without really trying. Fibonacci Quarterly. 1981;19:406–410. [Google Scholar]

5. Serra M. Discovering Geometry: An Inductive Approach. Key Curriculum Press, Berkeley; 1989. [Google Scholar]

6. Stewart I. The Mathematics of Life. Basic Books; 2011. [Google Scholar]

7. Brousseau A. Fibonacci statistics in conifers. Fibonacci Quarterly. 1969;7:525–532. [Google Scholar]

8. Coxeter HSM. Introduction to Geometry. 2nd edWiley; 1989. [Google Scholar]

9. Swinton J, Ochu E, Consortium TMTS. Novel Fibonacci and non-Fibonacci structure in the sunflower: results of a citizen science experiment. Royal Society Open Science. 2016;3(5). 10.1098/rsos.160091 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

10. Newell AC, Pennybacker M. Fibonacci patterns: common or rare?Procedia IUTAM. 2013;9(Supplement C):86–109. 10.1016/j.piutam.2013.09.009 [CrossRef] [Google Scholar]

11. Rueda-Contreras MD, Aragon JL. Alan Turing’s chemical theory of phyllotaxis. Revista Mexicana de Fisica. 2014;60:1–12. [Google Scholar]

12. Turing AM. The chemical basis of morphogenesis. Philosophical Transactions of the Royal Society of London Series B, Biological Sciences. 1952;237(641):37–72. 10.1098/rstb.1952.0012 [CrossRef] [Google Scholar]

13. Dress A, Giegerich R, Grünewald S, Wagner H. Fibonacci-Cayley numbers and repetition patterns in genomic DNA. Annals of Combinatorics. 2003;7:259–279. 10.1007/s00026-003-0185-z [CrossRef] [Google Scholar]

14. Yamagishi MEB, Shimabukuro AI. Nucleotide frequencies in human genome and Fibonacci numbers. Bulletin of Mathematical Biology. 2008;70(3):643–653. 10.1007/s11538-007-9261-6 [PubMed] [CrossRef] [Google Scholar]

15. Perez JC. Codon populations in single-stranded whole human genome DNA are fractal and fine-tuned by the Golden Ratio 1.618. Interdisciplinary Sciences: Computational Life Sciences. 2010;2(3):228–240. [PubMed] [Google Scholar]

16. Prusinkiewicz P, Lindenmayer A. The Algorithmic Beauty of Plants. New York: Springer-Verlag; 1996. [Google Scholar]

17. Basin SL. The Fibonacci sequence as it appears in nature. Fibonacci Quarterly. 1963;1:53–56. [Google Scholar]

18. Hutchison LAD, Myres NM, Woodward SR. Growing the family tree: the power of DNA in reconstructing family relationships. In: Proceedings of the First Symposium on Bioinformatics and Biotechnology (BIOT-04, Colorado Springs); 2004. p. 42–49.

19. Liu Y, Sumpter DJT. Spontaneous emergence of self-replication in chemical reaction systems. bioRxiv. 2018. [PMC free article] [PubMed] [Google Scholar]

20. Levin O. Discrete Mathematics: An Open Introduction. CreateSpace Independent Publishing Platform; 2015. [Google Scholar]

21. Krcadinac V. A new generalization of the golden ratio. Fibonacci Quarterly. 2006;44(4):335–340. [Google Scholar]

22. Mazorchuk V. Notes on non-negative matrices and the Perron–Frobenius theorem; 2017. [Google Scholar]

Articles from PLOS ONE are provided here courtesy of PLOS

FAQs

Is the golden ratio a universal constant for self-replication? ›

Such empirical examples continue to give credence to the idea that the golden ratio is a universal constant, not only in mathematics but also for biology.

Know More ›

Is the golden ratio everywhere? ›

The mysterious Golden Ratio can be seen all over the place in nature. They are, found across the cosmos. From the human ear's curve to the human bone structure, to the growth pattern of leaves and flowers, to the spirals of a nautilus seashell or the spirals of galaxies, spirals can be seen everywhere [4] .

Does the golden ratio apply to everything? ›

The Golden Ratio is a mathematical ratio you can find almost anywhere, like nature, architecture, painting, and music. When specifically applied to design specifically, it creates an organic, balanced, and aesthetically pleasing composition.

Discover More Details ›

Does the golden ratio repeat? ›

In math, Phi represents a number that starts with 1.618033988749895… And goes on forever without repeating! That's one reason Phi is an irrational number.

Find Out More ›

What is the significance of the golden ratio? ›

The golden ratio has been used to analyze the proportions of natural objects and artificial systems such as financial markets, in some cases based on dubious fits to data. The golden ratio appears in some patterns in nature, including the spiral arrangement of leaves and other parts of vegetation.

What did Fibonacci say about the golden ratio? ›

The golden ratio is derived by dividing each number of the Fibonacci series by its immediate predecessor. In mathematical terms, if F(n) describes the nth Fibonacci number, the quotient F(n)/ F(n-1) will approach the limit 1.618... for increasingly high values of n. This limit is better known as the golden ratio.

Why is the golden ratio irrational? ›

Because of the very slow convergence of this sequence, we say that the golden ratio is most difficult to approximate by a rational number. More poetically, the golden ratio has been called the most irrational of the irrational numbers.

What is an example of the golden ratio in real life? ›

For example, the measurement from the navel to the floor and the top of the head to the navel is the golden ratio. Animal bodies exhibit similar tendencies, including dolphins (the eye, fins and tail all fall at Golden Sections), starfish, sand dollars, sea urchins, ants, and honey bees.

Show Me More ›

Where is the golden ratio in the world? ›

If we use mankind's own Meridian at 180 degrees as the origin point, the Golden Ratio point is at 42 degrees, 29 minutes and 32.05 seconds, which is 298 km to the east of Mecca. The image below shows the midpoint between Alaska and Russia.

Explore More ›

Where can you find the golden ratio in real life? ›

Golden Ratio is one of the most common mathematical ratios in nature. We see this ratio everywhere from majestic landscapes like the Pyramids of Giza and the Mona Lisa to modern-day logos such as Twitter and Pepsi. Golden ratios are unique because of their golden proportion.

Discover More ›

Which celebrity is closest to golden ratio? ›

Jodie Comer

The winner of Dr. Julian De Silva's 2023 study is actor Jodie Comer — our current Golden Ratio goddess whose face is 94.52% accurate to the Greek Golden Ratio.

Learn More Now ›

What place has the golden ratio? ›

Art historians have found other examples of the golden ratio in the Mona Lisa, the Parthenon in ancient Athens, and the Great Pyramid of Giza.

Learn More Now ›