Fractional Conformal Map, Qubit Dynamics and the Leggett-Garg Inequality

Abstract

Any pure state of a qubit can be geometrically represented as a point on the extended complex plane through stereographic projection. By employing successive conformal maps on the extended complex plane, we can generate an effective discrete-time evolution of the pure states of the qubit. This work focuses on a subset of analytic maps known as fractional linear conformal maps. We show that these maps serve as a unifying framework for a diverse range of quantum-inspired conceivable dynamics, including (i) unitary dynamics,(ii) non-unitary but linear dynamics and (iii)  non-unitary and non-linear dynamics where linearity (non-linearity) refers to the action of the discrete time evolution operator on the Hilbert space. We provide a characterization of these maps in terms of Leggett-Garg Inequality complemented with No-signaling in Time (NSIT) and Arrow of Time (AoT) conditions.

Figure 1: Fractional Conformal Maps in two dimension and the Lüders bound

I Introduction

It is well known that the Bloch sphere can be identified with the Riemann sphere. The Riemann sphere, also known as the extended complex plane, serves as a one-point compactification of the complex plane (denoted as $C$) [1, 2, 3, 4]. This compactification is represented as $C\ \cup\ \infty=\tilde{C}$. Through stereographic projection, any point on the Bloch sphere can be mapped to a complex number $z$ on the extended complex plane. This geometric representation has been extensively explored in various contexts, as highlighted in works like [5, 1, 6, 7]. The successive application of a conformal map to a point on the extended complex plane, can be viewed as a discrete-time evolution of the Bloch vector on the Bloch sphere, representing the pure state. The complete set of conformal maps on the complex plane is characterized by locally invertible complex analytic functions, commonly known as Möbius transformations [9, 10, 11, 8].

The connection between the unitary evolution of a qubit, represented by a Bloch vector in two dimensions, and fractional linear conformal maps has been previously explored by Kim and Lee [5]. In this study, we extend this approach by delving into the exploration of all conceivable “fractional linear conformal maps” (FLC maps) in two dimensions and their classification based on the temporal correlations induced by such discrete-time qubit dynamics.

To probe the quantum nature of discrete-time qubit dynamics via temporal correlations, we employ the Leggett-Garg Inequality (LGI) [12, 14, 13]. In a simplified scenario involving three projective measurements and two steps of evolutions, we define the LGI using the LG parameter $K_{3}$ as: $-3\leq K_{3}=C_{12}+C_{23}-C_{13}\leq 1$, where $C_{ij}$ denotes two-time correlations. The violation of $K_{3}$ parameter serves as an indicator of non-classical behavior along with the conditions of No-signaling in Time (NSIT) and Arrow of Time (AoT) [15, 16], with the upper bound $3/2$ of $K_{3}$ referred to as the Lüders bound. Notably, recent theoretical studies and experimental observations have explored the potential violation of the Lüders bound, particularly in non-Hermitian systems [17, 18, 19, 20, 21, 24, 22, 23].

We categorize the parameter space of these FLC maps based on (a) the temporal correlations arising from discrete-time evolution and (b) the Linear or non-Linear action on the Hilbert space. (see FIG. 1). Employing a discrete-time version of LGI, we delineate the entire parameter space of these maps into three distinct classes: (i) exhibiting linear action on the Hilbert space respecting the Lüders bound, (ii) exhibiting non-linear action on the Hilbert space satisfying the Lüders bound, (iii) exhibiting non-linear action on the Hilbert space violating the Lüders bound.

The article is organized as follows: In section II, we introduce FLC maps, discussing the relationship between the extended complex plane and the Bloch sphere through stereographic projection. This section also comprehensively presents the dynamics of the pure state of the qubit induced by FLC maps. Moving on to section III, we devote our discussion to the interplay of linear and non-linear actions on the pure qubit state induced by these maps. In section IV we introduce the formalism for calculating the LGI for FLC maps and present analytical results. Finally, section V is reserved for discussions and concluding remarks.

II FLC maps induced dynamics

The stereographic projection is established by identifying the Bloch sphere with the Riemann sphere, enabling the definition of a projection $S:\mathcal{H}\longrightarrow\tilde{C}$ from the two-dimensional projective Hilbert space $\mathcal{H}$ to the extended complex plane $\tilde{C}$ [5]. A mathematical map, denoted as

$$f(z)=\frac{az+b}{cz+d},$$ (1)

where $a$, $b$, $c$, and $d$ are complex numbers with $ad-bc\not=0$, is characterized as a ‘fractional linear conformal map (FLC map)’ [11, 10]. In the Bloch sphere, the pure state of a qubit is represented by $\ket{\psi}=(\zeta_{1},\zeta_{2})=N(z,1)^{T}$, where $N=1/\sqrt{|z|^{2}+1}$ along with state-correspondent point on the extended complex plane being $z(=\zeta_{1}/\zeta_{2}$) [5],neglecting the overall phase. The discrete-time evolved state, corresponding to $z\mapsto f(z)$, is expressed as

$$\ket{\psi^{{}^{\prime}}}=\frac{1}{\sqrt{|f(z)|^{2}+1}}\ (f(z),1)^{T}$$ (2)

Equivalently, the operation induced by the fractional linear conformal map on the qubit state is represented as

$$|\psi^{{}^{\prime}}\rangle=N_{1}\Big{(}M|\psi\rangle\Big{)}=N_{1}\begin{pmatrix}a&b\\ c&d\end{pmatrix}|\psi\rangle=N_{1}\begin{pmatrix}az+b\\ cz+d\end{pmatrix}$$ (3)

where $M=\begin{pmatrix}a&b\\ c&d\end{pmatrix}$ is the matrix representation corresponding to the FLC map $f(z)$ with $N_{1}$ being the overall normalization of the qubit after the operation. This transformation ensures at least Positivity and Trace Preserving (PTP) properties, preserving the Hermiticity of the corresponding qubit state density matrix [25, 26]. The following relation schematically shows the direct correspondence between the Bloch sphere state and a point on the extended complex plane.

$$\ket{\psi}\longleftrightarrow z\mapsto z^{\prime}=f(z)\longleftrightarrow\ket{\psi}^{\prime}$$ (4)

III Linear vs. Non-linear Actions on the hilbert space

Action of a linear operator $\hat{O}$ on a pure state $|\psi\rangle=\eta_{1}|\psi_{1}\rangle+\eta_{2}|\psi_{2}\rangle$ can be expressed as

$$\hat{O}|\psi\rangle=\eta_{1}\ \hat{O}|\psi_{1}\rangle+\eta_{2}\ \hat{O}|\psi_{2}\rangle,$$ (5)

where $|\psi\rangle$ is decomposed in the orthonormal basis { $|\psi_{1}\rangle$ , $|\psi_{2}\rangle$} with $|\eta_{1}|^{2}+|\eta_{2}|^{2}=1$. A class of linear operator acting on the two dimensional Hilbert space can be written as

$$\hat{O}=\begin{pmatrix}a&b\\ -b^{*}&a^{*}\end{pmatrix}$$ (6)

with $|a|^{2}+|b|^{2}=r$, where $a,b\in\mathbb{C},r\in\mathbb{R}$. For determinant $r=1$, $\hat{O}$ is the unitary operator, while for $r\neq 1$, $\hat{O}$ corresponds to an operator which is unitary $\circ$ scaling. We will show that the parameter space of FLC maps (defined in Eq.(1)) includes both $r=1$ (unitary case explored by [5])and $r\neq 1$ case. On the contrary, $\hat{O}$ can violate linearity given in Eq.(5) if one relaxes the constraint $|a|^{2}+|b|^{2}=r$ such that,

$$\frac{\hat{O}|\psi\rangle}{\sqrt{\langle\psi|\hat{O}^{\dagger}\hat{O}|\psi\rangle}}\neq\tilde{N}\Big{(}\eta_{1}\frac{\hat{O}|\psi_{1}\rangle}{\sqrt{\langle\psi_{1}|\hat{O}^{\dagger}\hat{O}|\psi_{1}\rangle}}+\eta_{2}\frac{\hat{O}|\psi_{2}\rangle}{\sqrt{\langle\psi_{2}|\hat{O}^{\dagger}\hat{O}|\psi_{2}\rangle}}\Big{)}$$ (7)

where $\tilde{N}=\Big{|}\Big{|}\eta_{1}\frac{\hat{O}|\psi_{1}\rangle}{\sqrt{\langle\psi_{1}|\hat{O}^{\dagger}\hat{O}|\psi_{1}\rangle}}+\eta_{2}\frac{\hat{O}|\psi_{2}\rangle}{\sqrt{\langle\psi_{2}|\hat{O}^{\dagger}\hat{O}|\psi_{2}\rangle}}\Big{|}\Big{|}$. It is also shown that the parameter space of FLC maps also includes $|a|^{2}+|b|^{2}\neq r$ case. Although there can be a general class of non-linear operators not having the form of $\hat{O}$ but still following condition (7).

In general, any quantum evolution must respect linearity though unitarity can be compromised for an open system. On the other hand, quantum evolution interrupted by quantum measurement, followed by post selection can give rise to an effective dynamics having non-linear action on the space of states [21, 24]. In the next section we establish that FLC maps cover all possible quantum dynamics i.e. unitary, non-unitary but linear, non-unitary and non-linear.

Figure 2: Optimal LG $K_{3}$ values for a qubit undergoing dynamics induced by discrete FLC maps $f_{12}$, $f_{23}$, and $f_{13}$ (refer to Appendix-C.2). (a) Left diagram corresponds to the maps $f_{12}(z)=\frac{\alpha z+\beta}{\beta z+\alpha}$ and $f_{23}(z)=\frac{\gamma z+\delta}{\delta z+\gamma}$. (b) Right diagram corresponds to the maps $f_{12}(z)=\frac{\alpha z+\beta}{\beta^{*}z+\alpha^{*}}$ and $f_{23}(z)=\frac{\gamma z+\delta}{\delta^{*}z+\gamma^{*}}$. In both cases, the plot is independent of the initial qubit state on the Bloch sphere.

IV Discrete Time LGI with FLC Maps

In this section we consider evaluating three time LG parameter $K_{3}$ for discrete time evolution induced by FLC maps , corresponding to the dichotomic measurement operator $\hat{Q}=\sigma_{z}$. Steps followed for inducing the discrete evolution is given below.

Steps: (i) The state represented by $z=z_{1}$ on the extended complex plane is acted upon using the map $f_{12}(z)$. This map should be understood as the evolution of the system starting from time $t_{1}$ to time $t_{2}$.

(ii) The subsequent effective evolution over the next time interval from $t_{2}$ to $t_{3}$ is induced by the map $f_{23}(z)$ acting on the the state represented by $z=z_{2}=f_{12}(z_{1})$ at $t_{2}$.

(iii) The composite evolution from $t_{1}$ to $t_{3}$ is induced by the composition of the above maps i.e. $f_{23}\circ f_{12}$ starting from the state at $t_{1}$.

$$z_{2}=f_{12}(z_{1})=\frac{a_{12}\ z_{1}+b_{12}}{c_{12}\ z_{1}+d_{12}};\ \ z_{3}=f_{23}(z_{2})=\frac{a_{23}\ z_{2}+b_{23}}{c_{23}\ z_{2}+d_{23}}$$

$$z_{3}=f_{23}(f_{12}(z_{1}))=\frac{a_{13}z_{1}+b_{13}}{c_{13}z_{1}+d_{13}}$$ (8)

where, $a_{13}=a_{12}a_{23}+c_{12}b_{23},\ b_{13}=b_{12}a_{23}+d_{12}b_{23},\ c_{13}=a_{12}c_{23}+c_{12}d_{23},\ d_{13}=b_{12}c_{23}+d_{12}d_{23}$.

IV.1 Joint Probabilities ($P_{ij}$) and correlation functions ($C_{ij}$)

In this subsection we define the temporal correlations $C_{ij}$’s expressed in terms of joint probabilities $P_{ij}$’s which are required to evaluate the LG parameter for the dichotomic observable $\hat{Q}=\sigma_{z}$.

$$C_{ij}=\sum_{\hat{Q}(t_{i}/t_{j})=\pm 1}\hat{Q}(t_{i})\hat{Q}(t_{j})P_{ij}(\hat{Q}(t_{i}),\hat{Q}(t_{j}))$$ (9)

$$\text{with}\quad P_{ij}(\hat{Q}(t_{i}),\hat{Q}(t_{j}))=$$

$$\frac{(|\langle\pm_{Q}|e^{-iH(t_{j}-t_{i})}|\pm_{Q}\rangle|^{2})(|\langle\pm_{Q}|e^{-iH(t_{i}-t_{1})}|\psi^{(0)}\rangle|^{2})}{\langle\pm_{Q}|e^{iH^{\dagger}(t_{j}-t_{i})}e^{-iH(t_{j}-t_{i})}|\pm_{Q}\rangle\langle\pm_{Q}|e^{iH^{\dagger}(t_{i}-t_{1})}e^{-iH(t_{i}-t_{1})}|\psi^{(0)}\rangle}$$ (10)

where, $i<j\ \{i,j=1,2,3\}$ and $\ket{\psi^{(0)}}$ is the initial state. Here $\hat{Q}(t_{k})$ denotes the measurement outcome either $+1$ (corresponding to the $\ket{\uparrow}=(1,0)^{T}$) or $-1$ (corresponding to the $\ket{\downarrow}=(0,1)^{T}$) of dichotomic observable $\hat{Q}=\sigma_{z}$. We must point out that for our calculation, the state $e^{-iH(t_{j}-t_{i})}|\phi\rangle$ is equivalent to the state $(f_{ij}(z_{|\phi\rangle}),1)^{T}$ where $z_{|\phi\rangle}$ is the point on the complex plane corresponding to the state $|\phi\rangle$ on the Bloch sphere via stereographic projection. Also $f_{ii}(z_{|\phi\rangle})=z_{|\phi\rangle}$ for all $\{i=1,2,3\}$.

Calculation of $C_{12}$ :  Assuming $t_{1}=0$, the initial state $|\psi^{(0)}\rangle(=|\psi(t_{1}=0)\rangle)$ is expressed in the eigenbasis of $\hat{Q}$ as,

$$\ket{\psi^{(0)}}=\frac{1}{\sqrt{|z_{\ket{\psi^{(0)}}}|^{2}+1}}\ (z_{\ket{\psi^{(0)}}},1)^{T}.$$ (11)

The probabilities of obtaining the $\ket{\uparrow}$ or $\ket{\downarrow}$ eigenstates of the measurement operator $\hat{Q}=\sigma_{z}$ at $t_{1}$ are then calculated (assuming $z_{\ket{\psi^{(0)}}}=re^{i\phi}$) as:

$$P_{t_{1}}(+)=|\braket{\uparrow}{\psi^{(0)}}|^{2}=\frac{r^{2}}{1+r^{2}}$$ (12)

$$P_{t_{1}}(-)=|\braket{\downarrow}{\psi^{(0)}}|^{2}=\frac{1}{1+r^{2}}$$ (13)

Thereafter the two eigenstates ({$\ket{\uparrow}$, $\ket{\downarrow}$}) are evolved using the map $f_{12}(z)$, where $z=\infty$ for $\ket{\uparrow}$ and $z=0$ for $\ket{\downarrow}$. This results in two states at time $t_{2}$ (denoted as $\ket{\psi(t_{2})}$) expressed as:

$$\ket{\psi(t_{2})}_{\ket{\uparrow}}=\frac{1}{\sqrt{|f(z=\infty)|^{2}+1}}\ (f(z=\infty),1)^{T}$$ (14)

$$\ket{\psi(t_{2})}_{\ket{\downarrow}}=\frac{1}{\sqrt{|f(z=0)|^{2}+1}}\ (f(z=0),1)^{T},$$ (15)

The above equations can be re-written using eq.(8) as:

$$\ket{\psi(t_{2})}_{\ket{\uparrow}}=\frac{1}{\sqrt{|a_{12}/c_{12}|^{2}+1}}\ (a_{12}/c_{12},1)^{T}$$ (16)

$$\ket{\psi(t_{2})}_{\ket{\downarrow}}=\frac{1}{\sqrt{|b_{12}/d_{12}|^{2}+1}}\ (b_{12}/d_{12},1)^{T}.$$ (17)

Hence the joint probabilities can be evaluated as:

$$P_{12}(+,+)=\frac{r^{2}}{1+r^{2}}\times\frac{|a_{12}/c_{12}|^{2}}{|a_{12}/c_{12}|^{2}+1},$$ (18)

$$P_{12}(+,-)=\frac{r^{2}}{1+r^{2}}\times\frac{1}{|a_{12}/c_{12}|^{2}+1},$$ (19)

$$P_{12}(-,+)=\frac{1}{1+r^{2}}\times\frac{|b_{12}/d_{12}|^{2}}{|b_{12}/d_{12}|^{2}+1},$$ (20)

$$P_{12}(-,-)=\frac{1}{1+r^{2}}\times\frac{1}{|b_{12}/d_{12}|^{2}+1},$$ (21)

such that,

$$C_{12}=P_{12}(+.+)+P_{12}(\-,-)-P_{12}(-,+)-P_{12}(+,-)$$ (22)

which reduces to.

$$C_{12}=1-2z_{12}+2x_{12}(y_{12}+z_{12}-1),$$ (23)

where the $x_{12},y_{12},z_{12}$ lie between 0 and 1. The explicit forms are $x_{12}=\frac{r^{2}}{1+r^{2}}$, $y_{12}=\frac{|a_{12}/c_{12}|^{2}}{|a_{12}/c_{12}|^{2}+1}$ and $z_{12}=\frac{|b_{12}/d_{12}|^{2}}{|b_{12}/d_{12}|^{2}+1}$. Using the outlined prescription, the two-time correlation function $C_{ij}$ can be expressed as:

$$C_{ij}=1-2\ z_{ij}+2x_{ij}\ (y_{ij}+z_{ij}-1),$$ (24)

with $0\leq x_{ij},y_{ij},z_{ij}\leq 1$,and $i<j\ \{i,j=1,2,3\}$, where $x_{ij}=\frac{r^{2}}{1+r^{2}}$, $y_{ij}=\frac{|a_{ij}/c_{ij}|^{2}}{|a_{ij}/c_{ij}|^{2}+1}$ and $z_{ij}=\frac{|b_{ij}/d_{ij}|^{2}}{|b_{ij}/d_{ij}|^{2}+1}$ (see Appendix B for more details). Here $x_{ij}$’s depend on the state under evolution, while $y_{ij}$’s and $z_{ij}$’s depend on the elements of the maps $f_{12}$, $f_{23}$ and $f_{12}\circ f_{23}$. It is interesting to note that $C_{ij}$’s have no initial state dependence for $(y_{ij}+z_{ij}-1)=0$ which is similar to the condition of stationarity defined in the context of unitary dynamics in [41].

IV.2 Exploration of violation of Lüders bound

There is no clear prescription in the literature that which kind of qubit dynamics: (i) unitary (linear e.g closed system quantum dynamics),(ii) non-unitary (linear e.g open system Lindbladian quantum dynamics) and (iii)non-unitary (non-linear e.g. measurement induced non-Hermitian quantum dynamics) can possibly satisfy or violate Lüders bound. It is evident that the violation of Lüders bound is impossible within unitary (linear) dynamics of a qubit (see Appendix C.1) [21, 35]. It is also established in a recent work [35] that qubit dynamics represented by unital maps (a subset of non-unitary (linear) dynamics) can never exceed Lüders bound over the full parameter space. On the contrary, other recent works [21, 22, 24] have explored the violation of LG parameter $K_{3}$ beyond Lüders bound in non-Hermitian dynamics which is a subset of non-unitary (non-linear) dynamics. In this study, we are providing a mathematical constraint on the FLC map (although not exhaustive) to satisfy Lüders bound irrespective of its linear (or non-linear) nature. Firstly a demonstrative example of non-unitary(non-linear) qubit dynamics satisfying Lüders bound case is shown which is induced by FLC maps. A general $2\times 2$ non-unitary(non-linear) operator to evolve an initial qubit from a time instance $t_{i}$ to $t_{j}$, can be parametrized as,

$$\tilde{U}_{ij}=\alpha_{ij}\mathbf{I}+\beta_{ij}\mathbf{\sigma_{x}}+\gamma_{ij}\mathbf{\sigma_{y}}+\zeta_{ij}\mathbf{\sigma_{z}}=\begin{pmatrix}\alpha_{ij}+\zeta_{ij}&\beta_{ij}-i\gamma_{ij}\\ \beta_{ij}+i\gamma_{ij}&\alpha_{ij}-\zeta_{ij}\end{pmatrix}$$ (25)

with $\{i,j=1,2,3\And i<j\}$,  $\{\alpha_{ij},\ \beta_{ij},\ \gamma_{ij},\ \zeta_{ij}\in\mathbb{C}\}$ and $\{\mathbf{I},\mathbf{\sigma_{x/y/z}}\}$ being Pauli matrices. The FLC map corresponding to the $\tilde{U}_{ij}$ has the form:

$$f_{ij}(z)=\frac{(\alpha_{ij}+\zeta_{ij})z+(\beta_{ij}-i\gamma_{ij})}{(\beta_{ij}+i\gamma_{ij})z+(\alpha_{ij}-\zeta_{ij})}$$ (26)

Considering $\boxed{\zeta_{ij}=0;\ \gamma_{ij}=0}$, the FLC maps $f_{12}$ and $f_{23}$ reduces to

$$f_{12}(z)=\frac{\alpha_{12}z+\beta_{12}}{\beta_{12}z+\alpha_{12}},\ f_{23}(z)=\frac{\alpha_{23}z+\beta_{23}}{\beta_{23}z+\alpha_{23}}$$ (27)

Further numerical calculations (see Appendix C.2) and the plot (Fig. 2) confirms that the above FLC map respects Lüders bound of $3/2$ for the full parameter space.

It is worth noting from eq.(24) that, if certain ratio are constrained in the following manner:

$$|a_{ij}/c_{ij}|=|d_{ij}/b_{ij}|,\ \text{such that}\ y_{ij}+z_{ij}=1$$ (28)

then the upper bound of the LG parameter,

$$K_{3}=1-2z_{12}-2z_{23}+2z_{13}.$$ (29)

always remains below the Lüders bound (refer to Appendix C.2 for more detailed calculation). The maximization of the Lüders bound for the $K_{3}$ parameter is independent of the initial qubit states in the Bloch sphere. See Table 1 for examples of such FLC maps.

FLC Maps satisfying ratio constraint
$\Large{f_{ij}(z)}$	$\Large{f_{ij}(z)}$
(i) $\dfrac{az\pm b}{bz\pm a}$	iii) $\dfrac{az\pm b}{b^{*}z\pm a^{*}}$
(ii) $\dfrac{az+b}{-bz\pm a}$	iv)$\dfrac{az+b}{-b^{*}z\pm a^{*}}$

Table 1: FLC Maps satisfying ratio constraint and Lüders bound

V NSIT and AoT conditions

In this section, we relook at the considerations surrounding the interpretation of $K_{3}>1$ as an indicator of non-classical (quantum and beyond) dynamics. In recent years, various endeavors have been undertaken to address the noninvasive measurability loophole [28, 29] and the clumsiness loophole [31]. However, our focus here is exclusively on the statistical version of noninvasive measurability (NSIT) and arrow of time (AoT) conditions. It is crucial to recall that the simultaneous nonviolations of NSIT and AoT conditions ensure the existence of a global joint probability (see Appendix A) distribution [27], implying macroscopic realism. In the context of unitary dynamics, NSIT conditions are typically violated, while all AoT conditions are satisfied. Considering the dynamical process induced on a qubit by fractional linear conformal maps $f_{12}$, $f_{23}$, and $f_{13}$, explicit calculations (refer to Appendix A) reveal the following: (a) all two-time AoT conditions of the form $\text{AoT}_{{i(j)}}:P(m_{i})=\sum_{m_{j}=\pm 1}P(m_{i},m_{j})$ are satisfied (where $\{i,j=1,2,3\And i<j\}$; (b) three-time AoT conditions of types $\text{AoT}_{12(3)}:P(m_{1},m_{2})=\sum_{m_{3}=\pm 1}P(m_{1},m_{2},m_{3})$ and $\text{AoT}_{1(23)}:P(m_{1})\equiv\sum_{m_{2},m_{3}=\pm 1}P(m_{1},m_{2},m_{3})$ are always satisfied; and (c) all NSIT conditions of types $\text{NSIT}_{(i)j}:P(m_{j})=\sum_{m_{i}=\pm 1}P(m_{i},m_{j}),\quad\quad i<j$, and $\text{NSIT}_{1(2)3}:P(m_{1},m_{3})\equiv\sum_{m_{2}=\pm 1}P(m_{1},m_{2},m_{3})$ and $\text{NSIT}_{(1)23}:P(m_{2},m_{3})\equiv\sum_{m_{1}=\pm 1}P(m_{1},m_{2},m_{3})$ are generally not satisfied (here $P(m_{1},m_{2},m_{3})$ denotes the global joint probability with $m_{1},m_{2},m_{3}$ being the outcomes of dichotomic observable at time instances $t_{1},t_{2}$ and $t_{3}$ respectively. In general, the dynamics induced by the presented fractional linear conformal maps are found to be inconsistent with Macroscopic Realism (MR).

In the rest of this section we present an example of FLC map (having non-linear and non-unitary) action on the state space) for which NSIT of the type $\text{NSIT}_{1(2)3}$ is violated throughout the parameter space (except for a one parameter family) of map elements . We again take the FLC maps to be following,

$$f_{12}(z)=\frac{\alpha_{1}\ z+\beta_{1}}{\beta_{1}\ z+\alpha_{1}},\ \ \text{and}\ \ f_{23}(z)=\frac{\alpha_{2}\ z+\beta_{2}}{\beta_{2}\ z+\alpha_{2}},\\$$ (30)

$$z_{3}=f_{23}(f_{12}(z))=\frac{(\alpha_{1}\alpha_{2}+\beta_{1}\beta_{2})z+(\beta_{1}\alpha_{2}+\alpha_{1}\beta_{2})}{(\beta_{1}\alpha_{2}+\alpha_{1}\beta_{2})z+(\alpha_{1}\alpha_{2}+\beta_{1}\beta_{2})}$$

$$=\frac{a_{13}z_{1}+b_{13}}{c_{13}z_{1}+d_{13}}$$ (31)

A thorough calculation of joint probabilities (where outcome of the dichotomic observable is only $+1$ for all time instances (see Appendix D for other outcomes) using the above FLC maps yields (see Appendix A),

$$P(+^{1},+^{3})=\frac{r^{2}}{1+r^{2}}\Big{(}\frac{|\lambda_{1}\lambda_{2}+1|^{2}}{|\lambda_{1}\lambda_{2}+1|^{2}+|\lambda_{1}+\lambda_{2}|^{2}}\big{)}$$ (32)

$$P(+,+,+)+P(+,-,+)=\frac{r^{2}}{1+r^{2}}\frac{|\lambda_{1}\lambda_{2}|^{2}+1}{(|\lambda_{1}|^{2}+1)(|\lambda_{2}|^{2}+1)}$$ (33)

where $\lambda_{1}=\frac{\beta_{1}}{\alpha_{1}}$ and $\lambda_{2}=\frac{\beta_{2}}{\alpha_{2}}$.

For $\text{NSIT}_{1(2)3}:P(+^{1},+^{3})=P(+,+,+)+P(+,-,+)$ to be obeyed both the eq.(32) and eq.(33) must match, yielding the following conditions,

$$\text{Re}(\lambda_{1}\lambda_{2})=0\quad\quad\text{and}\quad\quad\text{Re}(\lambda_{1})\text{Re}(\lambda_{2})=0$$ (34)

which in turn has two solutions,

$$\text{Condition \ 1:}\quad\quad\text{Re}(\lambda_{1})=\text{Im}(\lambda_{2})=0$$ (35)

$$\text{Condition \ 2:}\quad\quad\text{Re}(\lambda_{2})=\text{Im}(\lambda_{1})=0$$ (36)

This shows that the the $\text{NSIT}_{1(2)3}$ condition is only satisfied for i) $|\delta/\gamma|=0$ ( corresponding to eq.(35)), ii)   $|\beta/\alpha|=0$ ( corresponding to eq.(36))(see figure 2(a)). The above found parameter space also ensures that LG parameter $K_{3}$ is bounded by $1$ confirming classical behaviour of the dynamics. In general, for most of the parameter space of the FLC maps, optimal $K_{3}$ value is in the non-classical regime whenever NSIT conditions are violated.

VI Conclusion

In this study, We have established that the fractional linear conformal maps encompass a large variety of conceivable quantum dynamics interrupted (uninterrupted) by quantum measurements. We explore temporal correlations, quantified by the Leggett-Garg parameter $K_{3}$, in a two-level system subjected to the dynamics induced by fractional linear conformal maps, specifically denoted by $f_{12}$, $f_{23}$, and $f_{13}$. Our findings demonstrate that, when certain ratio constraints among the elements of these maps are met, the $K_{3}$ parameter remains within the confines of the Lüders bound, never surpassing the $3/2$ bound.

Additionally, we presented illustrative examples of the aforementioned maps in the form of a table, including the one falling under the category of unitary dynamics. Our investigation categorizes the classes of fractional linear conformal maps into three major distinct groups: (i) the action on the qubit state space is linear, and the Lüders bound is respected; (ii) the action on the state space is non-linear yet respects the Lüders bound; and (iii) the action on the state space is non-linear and violates the Lüders bound if specific ratio constraints are not satisfied.

VII Acknowledgement

S.P. offers his gratitude to the Council of Scientific and Industrial Research (CSIR), Govt. of India for financial support.

Appendix A Global joint probabilities

Using the maps $f{}_{12}$ , $f_{23}$ and $f_{13}$ (given in main text eq.(8)), we calculate the global three time joint probabilities as following,

$$P(+,+,+)={|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{2})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}}=\frac{|a_{23}/c_{23}|^{2}}{|a_{23}/c_{23}|^{2}+1}\times\frac{|a_{12}/c_{12}|^{2}}{|a_{12}/c_{12}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(+,+,-)={|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{2})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}}=\frac{1}{|a_{23}/c_{23}|^{2}+1}\times\frac{|a_{12}/c_{12}|^{2}}{|a_{12}/c_{12}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(+,-,+)={|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{2})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}}=\frac{|b_{23}/d_{23}|^{2}}{|b_{23}/d_{23}|^{2}+1}\times\frac{1}{|a_{12}/c_{12}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(+,-,-)={|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{2})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}}=\frac{1}{|b_{23}/d_{23}|^{2}+1}\times\frac{1}{|a_{12}/c_{12}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(-,+,+)={|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{2})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}}=\frac{|a_{23}/c_{23}|^{2}}{|a_{23}/c_{23}|^{2}+1}\times\frac{|b_{12}/d_{12}|^{2}}{|b_{12}/d_{12}|^{2}+1}\times\frac{1}{1+r^{2}}$$

$$P(-,+,-)={|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{2})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}}=\frac{1}{|a_{23}/c_{23}|^{2}+1}\times\frac{|b_{12}/d_{12}|^{2}}{|b_{12}/d_{12}|^{2}+1}\times\frac{1}{1+r^{2}}$$

$$P(-,-,+)={|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{2})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}}=\frac{|b_{23}/d_{23}|^{2}}{|b_{23}/d_{23}|^{2}+1}\times\frac{1}{|b_{12}/d_{12}|^{2}+1}\times\frac{1}{1+r^{2}}$$

$$P(-,-,-)={|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{2})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}}=\frac{1}{|b_{23}/d_{23}|^{2}+1}\times\frac{1}{|b_{12}/d_{12}|^{2}+1}\times\frac{1}{1+r^{2}}$$ (37)

Similarly we procced to calculate the two time joint ($t_{1},t_{2})$ probabilities:

$$P(+^{1},+^{2})=|\langle\uparrow|\psi(t_{2})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}=\frac{|a_{12}/c_{12}|^{2}}{|a_{12}/c_{12}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(+^{1},-^{2})=|\langle\downarrow|\psi(t_{2})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}=\frac{1}{|a_{12}/c_{12}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(-^{1},+^{2})=|\langle\uparrow|\psi(t_{2})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}=\frac{|b_{12}/d_{12}|^{2}}{|b_{12}/d_{12}|^{2}+1}\times\frac{1}{1+r^{2}}$$

$$P(-^{1},-^{2})=|\langle\downarrow|\psi(t_{2})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}=\frac{1}{|b_{12}/d_{12}|^{2}+1}\times\frac{1}{1+r^{2}}$$ (38)

where $P(\pm^{i},\pm^{j})$ denotes that the measurement is done at time instances $t_{i}$ first and then $t_{j}$. Next we have the two time joint ($t_{1},t_{3})$ probabilities following:

$$P(+^{1},+^{3})=|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{1})\rangle|^{2}=\frac{|a_{13}/c_{13}|^{2}}{|a_{13}/c_{13}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(+^{1},-^{3})=|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle+_{z}|\psi(t_{1})\rangle|^{2}=\frac{1}{|a_{13}/c_{13}|^{2}+1}\times\frac{r^{2}}{1+r^{2}}$$

$$P(-^{1},+^{3})=|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}=\frac{|b_{13}/d_{13}|^{2}}{|b_{13}/d_{13}|^{2}+1}\times\frac{1}{1+r^{2}}$$

$$P(-^{1},-^{3})=|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{1})\rangle|^{2}=\frac{1}{|b_{13}/d_{13}|^{2}+1}\times\frac{1}{1+r^{2}}$$ (39)

Next we have the two time joint ($t_{2},t_{3})$ probabilities following:

$$P(+^{2},+^{3})=|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{2})\rangle|^{2}=\frac{|a_{23}/c_{23}|^{2}}{|a_{23}/c_{23}|^{2}+1}\times\frac{|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}{1+|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}$$

$$P(+^{2},-^{3})=|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\uparrow}}|^{2}|\langle\uparrow|\psi(t_{2})\rangle|^{2}=\frac{1}{|a_{23}/c_{23}|^{2}+1}\times\frac{|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}{1+|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}$$

$$P(-^{2},+^{3})=|\langle\uparrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{2})\rangle|^{2}=\frac{|b_{23}/d_{23}|^{2}}{|b_{23}/d_{23}|^{2}+1}\times\frac{1}{1+|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}$$

$$P(-^{2},-^{3})=|\langle\downarrow|\psi(t_{3})\rangle_{\ket{\downarrow}}|^{2}|\langle\downarrow|\psi(t_{2})\rangle|^{2}=\frac{1}{|b_{23}/d_{23}|^{2}+1}\times\frac{1}{1+|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}$$ (40)

where $z_{1}=re^{i\phi}$ which corresponds to the initial unmeasured state $|\psi(t_{1})\rangle=|\psi^{(0)}\rangle=\frac{1}{\sqrt{|z_{1}|^{2}+1}}(z_{1},1)$.

The one time ($t_{i}$) probabilities:

$$P(+^{1})=|\langle\uparrow|\psi(t_{1})\rangle|^{2}=\frac{r^{2}}{1+r^{2}},\quad\quad P(-^{1})=|\langle\downarrow|\psi(t_{1})\rangle|^{2}=\frac{1}{1+r^{2}}$$

$$P(+^{2})=|\langle\uparrow|\psi(t_{2})\rangle|^{2}=\frac{|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}{1+|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}},\quad\quad P(-^{2})=|\langle\downarrow|\psi(t_{2})\rangle|^{2}=\frac{1}{1+|(a_{12}z_{1}+b_{12})/(c_{12}z_{1}+d_{12})|^{2}}$$

$$P(+^{3})=|\langle\uparrow|\psi(t_{3})\rangle|^{2}=\frac{|(a_{13}z_{1}+b_{13})/(c_{13}z_{1}+d_{13})|^{2}}{1+|(a_{13}z_{1}+b_{13})/(c_{13}z_{1}+d_{13})|^{2}},\quad\quad P(-^{3})=|\langle\downarrow|\psi(t_{3})\rangle|^{2}=\frac{1}{1+|(a_{13}z_{1}+b_{13})/(c_{13}z_{1}+d_{13})|^{2}}$$ (41)

Appendix B Sample calculation of $C_{ij}$:

Following the same argument given in the main text, $C_{23}$ can be expressed as:

$$C_{23}=1-2z_{23}+2x_{23}(y_{23}+z_{23}-1),$$ (42)

where $y_{23}=\frac{|a_{23}/c_{23}|^{2}}{|a_{23}/c_{23}|^{2}+1}$ and $z_{23}=\frac{|b_{23}/d_{23}|^{2}}{|b_{23}/d_{23}|^{2}+1}$.

Likewise the joint probabilities for the correlation $C_{13}$ are as follows:

$$P_{13}(+,+)=\frac{r^{2}}{1+r^{2}}\times y_{13}$$ (43)

$$P_{13}(+,-)=\frac{r^{2}}{1+r^{2}}\times(1-y_{13})$$ (44)

$$P_{13}(-,+)=\frac{1}{1+r^{2}}\times z_{13}$$ (45)

$$P_{13}(-,-)=\frac{1}{1+r^{2}}\times(1-z_{13})$$ (46)

$$C_{13}=1-2z_{13}+2x_{13}(y_{13}+z_{13}-1),$$ (47)

where $y_{13}$ and $z_{13}$ are given by:

$$x_{13}=\frac{r^{2}}{1+r^{2}},\quad\quad y_{13}=\frac{|a_{13}/c_{13}|^{2}}{|a_{13}/c_{13}|^{2}+1}$$ (48)

$$z_{13}=\frac{|b_{13}/d_{13}|^{2}}{|b_{13}/d_{13}|^{2}+1}\\$$ (49)

Appendix C Derivation of Lüders bound for (i) unitary (linear) case of qubit, (ii) general non-unitary (non-linear) case of FLC Maps

C.1 Unitary Case

For unitary FLC maps (given in main text eq.(8)) the elements of $f_{12}$, $f_{23}$ and $f_{13}$ are related by,

$$d_{ij}=a^{*}_{ij},\quad c_{ij}=-b^{*}_{ij}$$ (50)

together with

$$|a_{ij}|^{2}+|b_{ij}|^{2}=1\ \text{for}\ i<j,\{i,j=1,2,3\}$$

Hence the expressions of $C_{ij}$’s simplify to,

$$C_{12}=1-2|b_{12}|^{2},\ C_{23}=1-2|b_{23}|^{2}$$

$$C_{13}=1-2|b_{13}|^{2}$$

$$=1-2\Big{(}|a_{12}|^{2}|b_{23}|^{2}+|b_{12}|^{2}|a_{23}|^{2}+2\text{Re}(a_{12}a_{23}b_{12}b^{*}_{23})\Big{)}$$ (51)

Taking the quantities $a_{12}=\cos(\theta_{1})e^{i\gamma_{1}},\ b_{12}=\sin(\theta_{1})e^{i\gamma_{2}},\ a_{23}=\cos(\theta_{2})e^{i\gamma_{3}},\ b_{23}=\sin(\theta_{2})e^{i\gamma_{4}}$, the expression of $K_{3}$ becomes

$$K_{3}=\cos 2\theta_{1}+\cos 2\theta_{2}-\cos 2\theta_{1}\cos 2\theta_{2}$$

$$-\sin 2\theta_{1}\sin 2\theta_{2}\cos 2\gamma$$ (52)

where $\gamma=\gamma_{1}+\gamma_{2}+\gamma_{3}-\gamma_{4}$.

Eq.(52) ensures that $K_{3}$ is upper bounded by Lüders bound.

C.2 General case

Following the ratio constraints (given in main text eq.(28)) the expression of LG parameter becomes $K_{3}=1-2z_{12}-2z_{23}+2z_{13}$. Assuming the following quantities:   $b_{12}/d_{12}=r_{1}e^{i\theta_{1}},\ b_{23}/d_{23}=r_{2}e^{i\theta_{2}},\ a_{23}/b_{23}=r_{3}e^{i\theta_{3}},\ c_{23}/d_{23}=r_{4}e^{i\theta_{4}}$, we obtain

$$z_{12}=\frac{r_{1}^{2}}{1+r_{1}^{2}},\quad\quad\quad z_{23}=\frac{r_{2}^{2}}{1+r_{2}^{2}}$$

$$z_{13}=\frac{A}{A+B}$$ (53)

with $A=r_{2}^{2}\Big{(}r_{1}^{2}r_{3}^{2}+1+2r_{1}r_{3}\cos(\theta_{1}+\theta_{3})\Big{)}$ and $B=r_{1}^{2}r_{4}^{2}+1+2r_{1}r_{4}\cos(\theta_{1}+\theta_{4})$. Employing the ratio constraint $|\frac{a_{12}}{c_{12}}|=|\frac{d_{12}}{b_{12}}|$ and $|\frac{a_{23}}{c_{23}}|=|\frac{d_{23}}{b_{23}}|$, it can be established

$$\boxed{r_{4}^{2}=r_{3}^{2}r_{2}^{4}}$$ (54)

. From the ratio constraint $|\frac{a_{13}}{c_{13}}|=|\frac{d_{13}}{b_{13}}|$ we also find another relation ,

$$r_{2}^{2}\Big{(}\frac{r_{1}^{2}+r_{3}^{2}+2r_{1}r_{3}\cos\theta_{3}}{r_{1}^{2}+r_{4}^{2}+2r_{1}r_{4}\cos\theta_{4}}\Big{)}=\big{(}\frac{r_{1}^{2}r_{4}^{2}+1+2r_{1}r_{4}\cos\theta_{4}}{r_{1}^{2}r_{3}^{2}+1+2r_{1}r_{3}\cos\theta_{3}}$$ (55)

Likewise $\theta_{1}$ can be set to zero (without loss of generality). Hence $K_{3}$ parameter becomes a function of four unknown parameters { $r_{1},\ r_{2},\ \theta_{3},\ \theta_{4}$ }. It is checked numerically that, $K_{3}$ respects Lüders bound for the full parameter space of { $r_{1},\ r_{2},\ \theta_{3},\ \theta_{4}$ } .

Appendix D NSIT conditions

It is evident from (Appendix 37-40) $\text{NSIT}_{1(2)3}$ is violated for all spin projections i.e.:

$$P(+^{1},+^{3})\neq P(+,+,+)+P(+,-,+)$$ (56)

$$P(+^{1},-^{3})\neq P(+,+,-)+P(+,-,-)$$ (57)

$$P(-^{1},+^{3})\neq P(-,+,+)+P(-,-,+)$$ (58)

$$P(-^{1},-^{3})\neq P(-,+,-)+P(-,-,-)$$ (59)

Likewise all other NSIT conditions of type $\text{NSIT}_{(1)2}$, $\text{NSIT}_{(2)3}$, $\text{NSIT}_{(1)3}$ and $\text{NSIT}_{(1)23}$ are also violated for arbitrary parameter values.