betaFeedback

WorksPhysics and Astronomy

Particle number projection on a spatial domain

Abstract

The formalism of particle number on a spatial domain for mean field wave functions with pairing is revisited to account for the case where finite dimensional basis are used. The formulas differ from the ones previously used in the literature. It is shown that the present formalism has the right limit in the well known case of zero pairing whereas the other formalism do not satisfy this basic requirement. By using a simple one-dimensional model we illustrate the differences in the results for particle number distribution probability obtained with the two methods.

Introduction

Both in nuclear fission and in reaction dynamics one has to deal with situations in the different stages of dynamical evolution where the system is composed of two or more fragments each of them to be treated as individual systems. The properties of those fragments have to be extracted from the total wave function of the system in a quantum mechanical consistent way. For instance, the determination of the (integer) number of particles in each of the fragments represents a problem: the total number of particles of the system is obviously an integer number but the probabilistic character of quantum mechanics and its characteristic probability density distribution does not guarantee that the same property holds for the fragments. The problem is aggravated if pairing correlations are present in the system and the optimal mean field treatment departs from the traditional Hartree-Fock (HF) theory and the Hartree-Fock-Bogoliubov (HFB) method with its characteristic breaking of the particle number symmetry is considered. The problem how to restore particle number symmetry in a fragment was first addressed by Simenel in Ref [1] were the use of a projector on particle number in a given region of space was introduced. The formalism was applied to the description of fusion reactions with Slater determinants describing the two fusioning nuclei. Later, the formalism was extended in Ref [2] to include Bardeen-Cooper-Schieffer (BCS) wave functions and the evaluation of overlaps was carried out with the Pfaffian formalism [3, 4, 5] to keep track of the sign of the overlaps. In the next years a flurry of publications dealing with the problem has appeared [2, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16] and the formalism extended even to deal with the restoration of rotational symmetry in the fragment [10, 15]. These developments have lead to another wave of papers discussing various effects related to fission fragment’s properties due to particle number restoration. In these publications, the formalism developed for HFB or BCS wave functions runs parallel to the old formalism developed to restore particle number symmetry [17, 18, 19, 20] with the only modification of replacing the particle number operator N^\hat{N} in the gauge angle rotation operator eiφN^e^{i\varphi\hat{N}} by its counterpart N^D\hat{N}_{\mathcal{D}} restricted to a very specific region of space (called a spatial domain D\mathcal{D} ). Unfortunately, those papers using BCS or HFB overlook the impact of using finite basis in the calculations [21, 22] and therefore they may obtain incorrect expressions for the required quantities as it will be shown below. It turns out that the expression used in [2] using the results of [5] based on pfaffians is applicable to the present case even thought the formalism of [5] assumes implicitly a basis closed under the symmetry operation. The reason for this fortunate coincidence is explained in [23] where it is also warned that the formulas in [5] for overlaps of general operators are not valid in the present case. On the other hand, the HF formalism used in Ref [1] is correct as the finitude of the basis is automatically accounted for by the determinant of the overlap matrix between the different orbitals involved [24]. To our knowledge the formalism for finite basis not closed under rotations [21, 22] has only been applied to the calculation of angular momentum projected potential energy surfaces for fission [25], the angular momentum distribution in fragments [15], and the evaluation of overlaps in fission when the oscillator bases have different values of the oscillator lengths [26]. The purpose of this paper is to develop the formalism of [21] in the present context of particle number restoration with BCS or HFB wave functions to provide the correct expressions for the calculation of particle number distribution probabilities. It is also proven how the present formalism provides the correct HF limit whereas the incorrect one of Refs [6, 7, 8, 9, 10, 11] does not. The differences between the two methods are illustrated in a simple one-dimensional example in the appendix.

Recently, angular momentum projection in a domain that is also widely discussed in the literature [10, 11, 15] to analyze the distribution of angular momentum in the fragments. The interest being triggered by experimental results obtained recently [27]. In this case, one also needs the formalism presented here (used in [15] but whitout a detailed explanation) that is expected to substantially differ from the one in [10, 11].

Particle number projection on a domain

To illustrate how projection on a domain D\mathcal{D} works in the case of using HFB wave functions, the case of particle number will be discussed in the following. To characterize the domain it is convenient to introduce the idempotent operator

Θ^D(r⃗)={Ir⃗⊂D0otherwise\hat{\Theta}_{\mathcal{D}}(\vec{r}) = \begin{cases} \mathbb{I} & \vec{r} \subset\mathcal{D} \\ 0 & \text{otherwise} \end{cases}

which is the identity operator if r⃗⊂D\vec{r} \subset\mathcal{D} and zero otherwise. The particle number operator in the domain is defined as the product, conveniently symmetrized, of N^\hat{N} times Θ^D\hat{\Theta}_{\mathcal{D}}

N^D=Θ^D(r⃗)N^Θ^D(r⃗)\hat{N}_{\mathcal{D}} = \hat{\Theta}_{\mathcal{D}}(\vec{r})\hat{N}\hat{\Theta}_{\mathcal{D}}(\vec{r})

The operator is used to define the projector on particle number in the domain D\mathcal{D}

P^DN=12π∫02πdφeiφ(N^D−N)(1)\hat{P}_{\mathcal{D}}^{N} = \frac{1}{2\pi}\int_{0}^{2\pi}d\varphi e^{i\varphi(\hat{N}_{\mathcal{D}}-N)} \tag*{(1)}

The treatment becomes more transparent if operators are expressed in terms of the creation and annihilation operators written in coordinate representation

ψ^†(r⃗)=∑kϕk∗(r⃗)ck†\hat{\psi}^{\dagger}(\vec{r}) = \sum_{k}\phi_{k}^{*}(\vec{r})c_{k}^{\dagger}
ψ^(r⃗)=∑kϕk(r⃗)ck\hat{\psi}(\vec{r}) = \sum_{k}\phi_{k}(\vec{r})c_{k}

where ϕk(r⃗)\phi_{k}(\vec{r}) is the wave function associated to the state ∣k⟩=ck†∣−⟩|k\rangle= c_{k}^{\dagger}|-\rangle. The operator N^D\hat{N}_{\mathcal{D}} is given by

N^D=∫d3r⃗ψ^†(r⃗)ψ^(r⃗)ΘD(r⃗)\hat{N}_{\mathcal{D}} = \int d^{3}\vec{r}\hat{\psi}^{\dagger}(\vec{r})\hat{\psi}(\vec{r})\Theta_{\mathcal{D}}(\vec{r})

where now ΘD(r⃗)\Theta_{\mathcal{D}}(\vec{r}) is the Heaviside or step function in the domain D\mathcal{D}. It is also easy to obtain

eiφN^D=Θ^D‾(r⃗)+eiφΘ^D(r⃗)=I+(eiφ−1)Θ^D(r⃗)e^{i\varphi\hat{N}_{\mathcal{D}}} = \hat{\Theta}_{\overline{\mathcal{D}}}(\vec{r}) + e^{i\varphi}\hat{\Theta}_{\mathcal{D}}(\vec{r}) = \mathbb{I} + (e^{i\varphi}-1)\hat{\Theta}_{\mathcal{D}}(\vec{r})

where Θ^D‾(r⃗)\hat{\Theta}_{\overline{\mathcal{D}}}(\vec{r}) is the step function in the complementary domain D‾\overline{\mathcal{D}} and it is such that Θ^D‾(r⃗)+Θ^D(r⃗)=I\hat{\Theta}_{\overline{\mathcal{D}}}(\vec{r}) + \hat{\Theta}_{\mathcal{D}}(\vec{r}) = \mathbb{I}. The result of acting with the above operator on a wave function of the basis ϕk(r⃗)\phi_{k}(\vec{r}) is to multiply it by 1+(eiφ−1)ΘD(r⃗)1 + (e^{i\varphi}-1)\Theta_{\mathcal{D}}(\vec{r}) to give another wave function ϕ~k(r⃗)\widetilde{\phi}_{k}(\vec{r})

ϕ~k(r⃗)=(1+(eiφ−1)ΘD(r⃗))ϕk(r⃗).\widetilde{\phi}_{k}(\vec{r}) = \left(1 + (e^{i\varphi}-1)\Theta_{\mathcal{D}}(\vec{r})\right)\phi_{k}(\vec{r}).

These considerations use operators that are implicitly defined in the whole Hilbert space and therefore they are implicitly associated to a complete basis. When the basis used in the calculation is not complete, as it is usually the case in practical calculations where the number of basis states is finite, the new set of wave functions ϕ~k(r⃗)\widetilde{\phi}_{k}(\vec{r}) cannot be expressed in general as a linear combination of the basis’s elements. This is a fundamental remark as the operators defined in the subspace spanned by the finite basis do not share in general the same properties of the operators defined in the whole space. In those cases the overlap between HFB states and the associated generalized Wick’s theorem have to be formulated using the method discussed in Ref [22, 21] where a formula is given for the overlap between HFB states which are expressed in different bases not connected by a unitary transformation. This situation might seem paradoxical because N^D\hat{N}_{\mathcal{D}} is hermitian and its exponential (multiplied by ii) must be a unitary operator. However, this line of reasoning is only correct when considering the complete Hilbert space because the operator N^D\hat{N}_{\mathcal{D}} has non-zero matrix elements connecting every corner of the Hilbert space - see Appendix A for an example.

Projected overlap

Given a general HFB wave function ∣Ψ⟩|\Psi\rangle the goal is to determine ∣ΨDN⟩=P^DN∣Ψ⟩|\Psi_{\mathcal{D}}^{N}\rangle= \hat{P}_{\mathcal{D}}^{N}|\Psi\rangle in order to compute the particle number distribution probability

a(N)=⟨Ψ∣ΨDN⟩=12π∫02πdφe−iφN⟨Ψ∣eiφN^D∣Ψ⟩.(2)a(N) = \langle\Psi|\Psi_{\mathcal{D}}^{N}\rangle= \frac{1}{2\pi}\int_{0}^{2\pi}d\varphi e^{-i\varphi N}\langle\Psi|e^{i\varphi\hat{N}_{\mathcal{D}}}|\Psi\rangle. \tag*{(2)}

of the subsystem enclosed in D\mathcal{D}. According to Refs [21, 22] the more general overlap ⟨Ψ0∣eiφN^D∣Ψ1⟩\langle\Psi_{0}|e^{i\varphi\hat{N}_{\mathcal{D}}}|\Psi_{1}\rangle is given by

⟨Ψ0∣eiφN^D∣Ψ1⟩=det⁡Adet⁡R(3)\langle\Psi_{0}|e^{i\varphi\hat{N}_{\mathcal{D}}}|\Psi_{1}\rangle= \sqrt{\det A\det\mathcal{R}} \tag*{(3)}

with

A=Uˉ0T(RT)−1Uˉ1∗+Vˉ0TRVˉ1∗(4)A = \bar{U}_{0}^{T}(\mathcal{R}^{T})^{-1}\bar{U}_{1}^{*}+\bar{V}_{0}^{T}\mathcal{R}\bar{V}_{1}^{*} \tag*{(4)}

and

Rnm=⟨ϕn∣eiφN^D∣ϕm⟩=δkl+(eiφ−1)⟨ϕn∣Θ^D∣ϕm⟩.(5)\mathcal{R}_{nm} = \langle\phi_{n}|e^{i\varphi\hat{N}_{\mathcal{D}}}|\phi_{m}\rangle= \delta_{kl}+(e^{i\varphi}-1)\langle\phi_{n}|\hat{\Theta}_{\mathcal{D}}|\phi_{m}\rangle. \tag*{(5)}

In the above expressions Uˉi\bar{U}_{i} and Vˉi\bar{V}_{i} (i=0,1i=0,1) are the Bogoliubov amplitudes of each of the HFB wave functions (the bar over the symbols is kept for consistency with the notation of [22]) and ∣ϕn⟩|\phi_{n}\rangle is a generic state of the basis BB of dimension NN, which is assumed to be the same for both ∣Ψ0⟩|\Psi_{0}\rangle and ∣Ψ1⟩|\Psi_{1}\rangle. The matrix (ΘD)nm=⟨ϕn∣Θ^D∣ϕm⟩(\Theta_{\mathcal{D}})_{nm}=\langle\phi_{n}|\hat{\Theta}_{\mathcal{D}}|\phi_{m}\rangle is hermitian and therefore can be diagonalized by a unitary transformation TT to give real eigenvalues θl\theta_{l}. The operator Θ^D\hat{\Theta}_{\mathcal{D}} is idempotent when defined in the whole Hilbert space and therefore it should have eigenvalues θl\theta_{l} equal to zero or one. However this property does not apply in the case of a finite basis as the idempotent property requires a complete basis - see Appendix A. The overlap matrix R\mathcal{R} is diagonal also in the basis given by the TT transformation R=TrT+\mathcal{R}=T r T^{+} with eigenvalues

rl(φ)=1+(eiφ−1)θl.(6)r_{l}(\varphi) = 1+(e^{i\varphi}-1)\theta_{l}. \tag*{(6)}

From here it is straightforward to write (RT)−1=T∗r−1TT(\mathcal{R}^{T})^{-1}=T^{*}r^{-1}T^{T}. Please note that rlr_{l} is not a phase if θl\theta_{l} is different from zero or one - see Appendix A. This is manifestation of the use of finite basis not closed under the symmetry transformation. It is worth to emphasize at this point that the matrix AA of Eq 4 is not the one implicitly used in Refs [6, 7, 8, 9, 10, 11] ((RT)−1(\mathcal{R}^{T})^{-1} here is replaced by R∗\mathcal{R}^{*} there) and therefore the conclusions discussed in those papers, obtained in calculations with a finite basis, and in the case of non-zero pairing states are not correct. Please note that in the “standard” formulation for the particle number operator the overlap is given by

⟨Ψ∣eiφN^∣Ψ⟩=eiφTr⁡[N]/2det⁡A\langle\Psi|e^{i\varphi\hat{N}}|\Psi\rangle=e^{i\varphi\operatorname{Tr}[N]/2}\sqrt{\det A}

with A=e−iφUˉTUˉ∗+eiφVˉTVˉ∗A=e^{-i\varphi}\bar{U}^{T}\bar{U}^{*}+e^{i\varphi}\bar{V}^{T}\bar{V}^{*}. It is customary to factor the e−iφe^{-i\varphi} in AA to cancel out the phase in the overlap to write ⟨Ψ∣eiφN^∣Ψ⟩=det⁡Aˉ\langle\Psi|e^{i\varphi\hat{N}}|\Psi\rangle=\sqrt{\det\bar{A}} with Aˉ=UˉTUˉ∗+e2iφVˉTVˉ∗\bar{A}=\bar{U}^{T}\bar{U}^{*}+e^{2i\varphi}\bar{V}^{T}\bar{V}^{*}. This is the strategy in [6, 7, 8, 9, 10, 11] that is clearly not justified in the present case of particle number on a domain. On the other hand, the formulas and results using Slater determinants [1], relying on the formulas derived in Ref [28], are correct. In the next subsection I will prove that the present formalism reduces in the zero-pairing limit to the result of Simenel and it will also be proven that this is not the case for the formulas in [6, 7, 8, 9, 10, 11].

The zero-pairing limit

As the formalism used in the HF case [1], based on the formulas of [28], is correct, it is important to check that the zero-pairing limit of the formalism presented above recovers Simenel’s results. In [21], the zero-pairing limit of Eq. (3) was discussed and by using a simple argument it was proven that the Löwdin formula [28] is recovered in this limit. Here we consider another derivation which explicitly shows the different cancellations required to go from Eq. (3) to Löwdin’s formula used in [1] and also allows to see how the incorrect approaches of Refs [6, 7, 8, 9, 10, 11] do not reproduce the correct formula. The limit is easily obtained by using Bloch-Messiah theorem [29, 30] providing the canonical form of the Bogoliubov amplitudes

Uˉi=DiuˉiCi\bar{U}_{i}=D_{i}\bar{u}_{i}C_{i}
Vˉi=Di∗vˉiCi(7)\bar{V}_{i}=D_{i}^{*}\bar{v}_{i}C_{i} \tag*{(7)}

in terms of the unitary transformation to the canonical basis DiD_{i}, the unitary transformation among quasiparticles CiC_{i} and the block diagonal matrices uˉi\bar{u}_{i} and vˉi\bar{v}_{i}. The matrix AA of Eq. (4) becomes

A=C1T(uˉ0T(RˉT)−1uˉ1∗+vˉ0TRˉvˉ1∗)C1∗=C1TAˉC1∗A=C_{1}^{T}\left(\bar{u}_{0}^{T}(\bar{R}^{T})^{-1}\bar{u}_{1}^{*}+\bar{v}_{0}^{T}\bar{R}\bar{v}_{1}^{*}\right)C_{1}^{*}=C_{1}^{T}\bar{A}C_{1}^{*}

where Rˉ=D0†RD1\bar{R}=D_{0}^{\dagger}RD_{1} and the matrix Aˉ\bar{A} is the BCS version of Eq. (4). Up to an irrelevant phase that becomes 1 when ∣Ψ0⟩=∣Ψ1⟩|\Psi_{0}\rangle=|\Psi_{1}\rangle and that can be chosen by fixing the phase of ⟨Ψ0∣Ψ1⟩\langle\Psi_{0}|\Psi_{1}\rangle, one has det⁡A=det⁡Aˉ\det A=\det\bar{A} and det⁡R=det⁡Rˉ\det R=\det\bar{R} and therefore the formula for the overlap becomes ⟨Ψ0∣eiφN^D∣Ψ1⟩=det⁡Aˉdet⁡Rˉ\langle\Psi_{0}|e^{i\varphi\hat{N}_{D}}|\Psi_{1}\rangle=\sqrt{\det\bar{A}}\sqrt{\det\bar{R}}. The next step is to consider the expression of Rˉ\bar{R} in block form

Rˉ=(Rˉ11Rˉ12Rˉ21Rˉ22)(8)\bar{R}=\begin{pmatrix}\bar{R}_{11}&\bar{R}_{12}\\\bar{R}_{21}&\bar{R}_{22}\end{pmatrix} \tag*{(8)}

where the matrix Rˉ11\bar{R}_{11} is the A×AA\times A matrix (here AA is the number of particles) of the overlaps between the occupied AA orbitals in the canonical basis of ∣Ψ0⟩|\Psi_{0}\rangle and those of eiφN^D∣Ψ1⟩e^{i\varphi\hat{N}_{D}}|\Psi_{1}\rangle. Next, one needs the formulas for the determinant

det⁡Rˉ=det⁡Rˉ11det⁡[Rˉ22−Rˉ21Rˉ11−1Rˉ12]\det\bar{R}=\det\bar{R}_{11}\det\left[\bar{R}_{22}-\bar{R}_{21}\bar{R}_{11}^{-1}\bar{R}_{12}\right]

and inverse

(RˉT)−1=(∙∙∙[Rˉ22T−Rˉ12T(Rˉ11T)−1Rˉ21T]−1)(\bar{R}^{T})^{-1}= \begin{pmatrix} \bullet& \bullet\\ \bullet& [\bar{R}_{22}^{T}-\bar{R}_{12}^{T}(\bar{R}_{11}^{T})^{-1}\bar{R}_{21}^{T}]^{-1} \end{pmatrix}

of the block matrix of Eq. (8). In the zero-pairing limit the uˉ0=uˉ1\bar{u}_{0}=\bar{u}_{1} matrices are diagonal with 0 in the first AA diagonal elements and 1 in the remaining ones, and the matrices vˉ0=vˉ1\bar{v}_{0}=\bar{v}_{1} are block diagonal with the first AA elements consisting of 2×22\times2 matrices (01−10)\begin{pmatrix}0&1\\-1&0\end{pmatrix} or 1 and the rest of the diagonal elements 0. Taking this form into account one obtains

Aˉ=(IVTRˉ11IV00[Rˉ22T−Rˉ12T(Rˉ11T)−1Rˉ21T]−1)\bar{A}= \begin{pmatrix} I_{V}^{T}\bar{R}_{11}I_{V}&0\\ 0&[\bar{R}_{22}^{T}-\bar{R}_{12}^{T}(\bar{R}_{11}^{T})^{-1}\bar{R}_{21}^{T}]^{-1} \end{pmatrix}

where IVI_{V} is the A×AA\times A diagonal matrix with the first AA elements consisting of 2×22\times2 matrices (01−10)\begin{pmatrix}0&1\\-1&0\end{pmatrix} or 1. As det⁡IV=1\det I_{V}=1 one finally gets

det⁡Aˉdet⁡Rˉ=(det⁡Rˉ11)2\det\bar{A}\det\bar{R}=(\det\bar{R}_{11})^{2}

which gives, in the zero pairing limit and up to a phase

⟨Ψ0∣eiφN^D∣Ψ1⟩=det⁡Rˉ11.\langle\Psi_{0}|e^{i\varphi\hat{N}_{D}}|\Psi_{1}\rangle=\det\bar{R}_{11}.

In the ∣Ψ0⟩=∣Ψ1⟩|\Psi_{0}\rangle=|\Psi_{1}\rangle case where the phase is 1 one recovers Löwdin formula as used in [1]. Please note that if we were using the incorrect formulas of [6, 7, 8, 9, 10, 11] we should replace (RˉT)−1(\bar{R}^{T})^{-1} by Rˉ∗\bar{R}^{*} and therefore in that case det⁡Aˉdet⁡Rˉ=(det⁡Rˉ11)2det⁡Rˉ22∗/det⁡[Rˉ22T−Rˉ12T(Rˉ11T)−1Rˉ21T]\det\bar{A}\det\bar{R}=(\det\bar{R}_{11})^{2}\det\bar{R}_{22}^{*}/\det[\bar{R}_{22}^{T}-\bar{R}_{12}^{T}(\bar{R}_{11}^{T})^{-1}\bar{R}_{21}^{T}] which is not the correct limit.

The Pfaffian formula

In the previous formalism the overlap is given by a square root and therefore its sign is not defined in general [3]. This can become a real problem when the overlap depends on a continuous variable used as an integration variable. To circumvent the problem it was proposed to use a new formula based on techniques using fermion coherent states [3, 4, 5]. The formula, commonly denoted as “pfaffian formula” is given in terms of the Pfaffian of a skew symmetric matrix and it is formulated assuming the two HFB states are expressed in terms of the same basis (or equivalent bases up to a unitary transformation). When this is not the case, an extension of the pfaffian formalism given in Ref [4] solves the problem. In the following we will adapt the results of Ref [4] to our present situation. As discussed in [4] the relevant quantity is the extended overlap matrix

N=(IRR+I)\mathcal{N}=\begin{pmatrix}\mathbb{I}&\mathcal{R}\\\mathcal{R}^{+}&\mathbb{I}\end{pmatrix}

defined in the vector space union of the two bases B={∣ϕn⟩,n=0,…,N−1}B = \{|\phi_n\rangle, n = 0, \ldots, N - 1\} and B~={∣ϕ~n⟩=eiφN^D∣ϕn⟩,n=0,…,N−1}\widetilde{B} = \{|\widetilde{\phi}_n\rangle= e^{i\varphi\hat{N}_D}|\phi_n\rangle, n = 0, \ldots, N - 1\}. The matrix R\mathcal{R} is the one of Eq 5 with eigenvalues given in Eq 6 in terms of the real eigenvalues θl\theta_l of the matrix (ΘD)nm=⟨ϕn∣Θ^D∣ϕm⟩(\Theta_D)_{nm} = \langle\phi_n|\hat{\Theta}_D|\phi_m\rangle. Adapting the developments in Ref [4] to the present set up one obtains

⟨Ψ0∣eiφN^D∣Ψ1⟩=(−1)Npf⁡(N~D(1)−IN~D(0)∗−N~D(0)∗)(9)\langle\Psi_0|e^{i\varphi\hat{N}_D}|\Psi_1\rangle= (-1)^N \operatorname{pf}\left(\begin{array}{cc} \widetilde{N}_D^{(1)} & -I \\ \widetilde{N}_D^{(0)*} & -\widetilde{N}_D^{(0)*} \end{array}\right) \tag*{(9)}

where

N~D(0)=12(E+00E−)(M(0)−M(0)−M(0)M(0))(E+T00E−T)(10)\widetilde{N}_D^{(0)} = \frac{1}{2}\left(\begin{array}{cc} E_+ & 0 \\ 0 & E_- \end{array}\right)\left(\begin{array}{cc} M^{(0)} & -M^{(0)} \\ -M^{(0)} & M^{(0)} \end{array}\right)\left(\begin{array}{cc} E_+^T & 0 \\ 0 & E_-^T \end{array}\right) \tag*{(10)}

and

N~D(1)=12(F+00F−)(M(1)M(1)M(1)M(1))(F+T00F−T).(11)\widetilde{N}_D^{(1)} = \frac{1}{2}\left(\begin{array}{cc} F_+ & 0 \\ 0 & F_- \end{array}\right)\left(\begin{array}{cc} M^{(1)} & M^{(1)} \\ M^{(1)} & M^{(1)} \end{array}\right)\left(\begin{array}{cc} F_+^T & 0 \\ 0 & F_-^T \end{array}\right). \tag*{(11)}

In the above expressions, E±=n±1/2T+E_{\pm} = n_{\pm}^{1/2}T^+ and F±=n±1/2frT+F_{\pm} = n_{\pm}^{1/2}f_rT^+ with TT the matrix diagonalizing (ΘD)nm(\Theta_D)_{nm}. The diagonal matrices n±n_{\pm} contain the diagonal elements n±=1±∣r∣n_{\pm} = 1 \pm|r| where the complex quantities rlr_l are defined in Eq 6 and frf_r is the phase of the complex number rr. Finally, the matrices M(i)M^{(i)} are the matrices of the Thouless transformation defining ∣Ψi⟩|\Psi_i\rangle (see [4] for detailed definitions). In the limiting case where the matrix (ΘD)nm(\Theta_D)_{nm} is the identity, its eigenvalues are 1, the rlr_l are all the same rl=eiφ=frr_l = e^{i\varphi} = f_r, ∣r∣=1|r| = 1 and n+=2n_+ = 2, n−=0n_- = 0. With these values one recovers the results of [3, 5].

The integration limit in the projector

The particle number projector is defined in general in Eq (1) with the 2π2\pi upper limit in the integral. It is often the case that the integration limit is reduced to the interval [0,π][0,\pi] by assuming that eiπ(N^D−N)e^{i\pi(\hat{N}_D-N)} is the identity when acting on a HFB state due to “number parity” super-selection rule. This statement is strictly true when the domain D\mathcal{D} is the whole space but it is not correct for a finite domain. For this property to be satisfied it is required the eigenvalues rlr_l of R\mathcal{R} (Eq (6)) to be −1-1 when φ=π\varphi= \pi which is only possible if θl=1\theta_l = 1. Therefore, it is mandatory to extend the integral in the projector to 2π2\pi as in this case the correct value of rl=1r_l = 1 is recovered irrespective of the value of θl\theta_l. On the other hand, owing to the property rl(2π−φ)=rl∗(φ)r_l(2\pi- \varphi) = r_l^*(\varphi) one can reduce the integration interval but at the price of replacing the integrand by twice its real part.

Reinterpretation of the formalism

In order to get more insight into the formalism it is convenient to introduce some ideas of condense matter physics and quantum information [31], first introduced in the context of fission in [16]. Let us define the non-orthogonal states ϕ~kD(r⃗)=ΘD(r⃗)ϕk(r⃗)\widetilde{\phi}_k^D(\vec{r}) = \Theta_D(\vec{r})\phi_k(\vec{r}) and their complementary ϕ~kD‾(r⃗)=ΘD‾(r⃗)ϕk(r⃗)\widetilde{\phi}_k^{\overline{D}}(\vec{r}) = \Theta_{\overline{D}}(\vec{r})\phi_k(\vec{r}) such that ϕk=ϕ~kD+ϕ~kD‾\phi_k = \widetilde{\phi}_k^D + \widetilde{\phi}_k^{\overline{D}}. One can orthogonalize those states by diagonalizing the hermitian matrix (ΘD)nm(\Theta_D)_{nm} with eigenvalues θl\theta_l to obtain the set of orthogonal states ϕkR(r⃗)=1θk∑lTlkϕ~lD(r⃗)\phi_k^R(\vec{r}) = \frac{1}{\sqrt{\theta_k}}\sum_l T_{lk}\widetilde{\phi}_l^D(\vec{r}) and ϕkL(r⃗)=11−θk∑lTlkϕ~lD‾(r⃗)\phi_k^L(\vec{r}) = \frac{1}{\sqrt{1-\theta_k}}\sum_l T_{lk}\widetilde{\phi}_l^{\overline{D}}(\vec{r}) which are also orthogonal for every kk and ll, i.e. ⟨ϕkR∣ϕlL⟩=0\langle\phi_k^R|\phi_l^L\rangle= 0. The set of orthogonal states ∣ϕkR⟩=Rk+∣⟩|\phi_k^R\rangle= R_k^+| \rangle are non-zero in the interior of the domain D\mathcal{D} whereas ∣ϕkL⟩=Lk+∣⟩|\phi_k^L\rangle= L_k^+| \rangle have the same property but in the complementary D‾\overline{\mathcal{D}}. Please note zero (one) eigenvalues have to be excluded in the set {ϕkR(ϕkL)}\{\phi_k^R(\phi_k^L)\}. With these definitions the basis states ∣ϕk⟩=ck+∣⟩|\phi_k\rangle= c_k^+| \rangle can be expanded as a linear combination

cl+=∑kTlk∗[θkRk++1−θkLk+](12)c_l^+ = \sum_k T_{lk}^*\left[\sqrt{\theta_k}R_k^+ + \sqrt{1-\theta_k}L_k^+\right] \tag*{(12)}

of states on the domain and outside it. This definition gives a new perspective to the values of θk\theta_k (see appendix below). The N^D\hat{N}_D operator can be written as N^D=∑kRk+Rk−\hat{N}_D = \sum_k R_k^+R_k^- what leads to eiφN^Dcl+e−iφN^D=∑kTlk∗[θkeiφRk++1−θkLk+]e^{i\varphi\hat{N}_D}c_l^+e^{-i\varphi\hat{N}_D} = \sum_k T_{lk}^*\left[\sqrt{\theta_k}e^{i\varphi}R_k^+ + \sqrt{1-\theta_k}L_k^+\right]. This result shows that the “number parity” associated to the cl+c_l^+ operators is not respected by the “number parity” of the Rk+R_k^+ implying, for instance, that a Slater determinant with an even number of particles has a non zero amplitude of having an odd number of particles in the domain D\mathcal{D}. In practical terms, the probability amplitude a(N)a(N) of a system with an even number of particles can also be different from zero for odd values of NN. From a quantum information perspective one can say that the original wave function is entangled with the one defined in the domain [31].

Application in a simple case

In order to appreciate the differences between the results obtained with the correct and the incorrect formalism as well as to study the entanglement in the system, I have carried calculations in a simple 1D model where the Hilbert space basis contains NB=40N_B = 40 elements corresponding to the spin 1/21/2, 1D Harmonic Oscillator wave functions with oscillator length b=1b = 1 and with quantum numbers n=0,…,NB/2−1n = 0,\ldots,N_B/2 - 1. The spin degree of freedom is explicitly considered giving a degeneracy factor of 2. For the pairing part, I have chosen a set of occupancies vn2=1/(1+exp⁡((n−n0)/σ))v_n^2 = 1/(1+\exp((n-n_0)/\sigma)) where n0n_0 and σ\sigma fix the shape of the orbital’s set occupancy. The particle number probability distribution a(N)a(N) has been computed with both the correct and the incorrect formulas in various situations including different domains D(x−x0)\mathcal{D}(x-x_0) with various values of x0x_0. The zero-pairing limit is also considered. The results, shown in appendix A clearly show how different the a(N)a(N) are except in those trivial cases where they should coincide (large negative values of x0x_0 to include in an effective way the whole Hilbert space). Finally, a more realistic 1D model is devised by using a Bogoliubov transformation tailored to produce a matter density with two fragments. Also in this case the a(N)a(N) differ pointing again to the necessity of considering the formalism of [21, 22] when dealing with finite size basis.

Although the model is discussed in detail in the appendix we show in Fig 1 two examples of a(N)a(N) distributions obtained in the zero-pairing limit and the two-fragment case. In both cases the correct distribution substantially differ from the incorrect one. The correct one is always peaked at the value of the average number of particles in the fragment. In the no-pairing case it is remarkable that a(N)a(N) significantly differs from zero for odd values of NN as a consequence of the entanglement with the fragment of the total wave function. On the other hand, in the two-fragment case the correct a(N)a(N) is very small for odd values of NN indicating a very low degree of entanglement associated to the clean definition of the fragment in this situation.

Summary and conclusions

In this paper we discuss how to compute properly the overlaps required in the projection on particle number on a domain D\mathcal{D} when dealing with HFB wave functions. It is shown that the formulas used in a series of recent papers are incorrect and the conclusions obtained in those papers dubious. The reason for the pitfall is not taking into account that the action of the symmetry operator on the finite basis leads to another basis which is not unitarily equivalent. The situation is exemplified in a simple case demonstrating the non-equivalence of some results when dealing with finite bases. Additionally, is is shown that the non-zero pairing limit of the present approach recovers Löwdin formula, contrary to previous formulations by other authors. The results discussed here call for a revision of a similar formalism for angular momentum projection on a domain.

Acknowledgements

I would like to thank S.A.Giuliani for a carefull reading of the manuscript and several suggestions. This work has been supported by Spanish Agencia Estatal de Investigacion (AEI) of the Ministry of Science and Innovation under Grant No. PID2021-127890NB-I00.

Finite basis: 1D example

To better visualize the problematic associated to the use of finite basis one can consider a one-dimensional Harmonic Oscillator (HO) basis BN={ϕn(x),n=0,…,N−1}B_N = \{\phi_n(x), n = 0, \ldots, N-1\} and a domain D\mathcal{D} corresponding to positive values of xx. The action of eiφN^De^{i\varphi\hat{N}_{\mathcal{D}}} on the basis’ states leads to non-continuous, non-differentiable wave function at x=0x = 0

ϕ~n(x)=eiφN^Dϕn(x)={ϕn(x)x<0eiφϕn(x)x≥0\tilde{\phi}_n(x) = e^{i\varphi\hat{N}_{\mathcal{D}}}\phi_n(x) = \begin{cases} \phi_n(x) & x < 0 \\ e^{i\varphi}\phi_n(x) & x \ge0 \end{cases}

such that the scalar product with a general state ϕm(x)\phi_m(x) is given by

⟨ϕn∣ϕ~m⟩=δnm+(eiφ−1)θnm\langle\phi_n|\tilde{\phi}_m\rangle= \delta_{nm} + \left(e^{i\varphi}-1\right)\theta_{nm}

with

θnm=∫0∞dxϕn∗(x)ϕm(x)=2n+mn!m!4π∑pq(−)p+qΓ((n+m−2p−2q)/2)22p+2qp!q!(n−2p)!(m−2q)!.(13)\begin{aligned} \theta_{nm} = \int_0^\infty dx \phi_n^*(x)\phi_m(x) \\ &= \sqrt{\frac{2^{n+m}n!m!}{4\pi}}\sum_{pq}\frac{(-)^{p+q}\Gamma((n+m-2p-2q)/2)}{2^{2p+2q}p!q!(n-2p)!(m-2q)!}. \tag*{(13)} \end{aligned}

This is a peculiar matrix with θnn=1/2\theta_{nn}=1/2 and θnm=0\theta_{nm}=0 if n+mn+m is even and n≠mn\ne m. The matrix elements are non-zero even when nn or mm (or both) are greater than NN, i.e. ϕ~n(x)\tilde{\phi}_n(x) cannot be expressed as a linear combination of the states in basis BNB_N. On the other hand, we will assume that such expansion is possible in the case of using an infinite numerable basis. It is also worth to remark that the new set ϕ~n(x)\tilde{\phi}_n(x) is orthonormal, as the original one. This property is also preserved in the most general case presented in the main text. The matrix θnm\theta_{nm} is hermitian and therefore it can be diagonalized by a unitary transformation with real eigenvalues θl\theta_l. In the simplest N=2N=2 case one has θ11=θ22=1/2\theta_{11}=\theta_{22}=1/2 and θ12=θ21=1/2π\theta_{12}=\theta_{21}=1/\sqrt{2\pi} and the eigenvalues are θ0=0.1010\theta_0=0.1010 and θ1=0.8989\theta_1=0.8989, as can be easily checked. In the general case of a domain D\mathcal{D} consistent of x≥x0x \ge x_0 the formula A.1 can be easily generalized by considering incomplete gamma function. In order to check the differences between our results and previous ones we have performed calculations for simple BCS wave functions in the basis BNB_N with an occupation profile given by vn2=1/(1+exp⁡(n+1−n0)/σ)v_n^2 = 1/(1+\exp(n+1-n_0)/\sigma) with N=20N=20, n0=10.5n_0=10.5 and σ=1\sigma=1 giving an average number of 20 particles. In Fig A.2 we plot the results corresponding to θl\theta_l and a(N)a(N) for x0x_0 values of −10-10, −5-5, −2-2 and 0. For x0=−10x_0=-10 the domain essentially covers all the xx values where the spatial density (see below) is different from zero. As expected, all the θl=1\theta_l=1 and our formulation and the incorrect one given in the literature coincide with the present formulation. We also observe that a(N)=0a(N)=0 for odd values of NN as expected. For x0=−5x_0=-5 a tiny part of the density is left off the domain (see Fig A.3 below) and, as a consequence the two lowest θl\theta_l values differ from the nominal values 0 or 1. We observe how the a(N)a(N) probability strongly differs in the two cases with the incorrect one peaked at N=22N=22 and showing a completely different shape than the correct one. It is worth to mention that the incorrect a(N)a(N) gets its larger values at odd values of NN, contrary to the correct distribution where the a(N)a(N) for odd NN are quite small. The total number of particles in the domain, obtained by means of the formula for the particle number sumrule ∑Na(N)×N\sum_N a(N)\times N equals 19.98 (instead of 20) for the correct formula and takes the value 21.08 for the incorrect one. For x0=−2x_0=-2 there are only two θl\theta_l values different from zero or one. Five θl\theta_l values are zero and 13 equal 1 indicating that the Rk+R_k^+ basis contains 15 vectors, whereas the Lk+L_k^+ one contains 7 basis states. The maximum of the two probability distribution differ being the correct one centered at N=16N=16 and the incorrect at N=24N=24. The number of particle sumrule is 15.47 and 21.51 for the correct and incorrect results, respectively, wich is a manifestation of strong entanglement between the x0=−2x_0=-2 domain and the rest of the space. Finally, the x0=0x_0=0 corresponding to a domain including half of the matter distribution have a similar behavior for θl\theta_l and the probability distributions have similar shapes but their maximum are clearly shifted being the correct a(N)a(N) centered around the correct value of 10 (remember that the average number of particles in the whole system is 20) whereas in the incorrect case it is centered around 20. These results clearly indicate the inconsistency of the incorrect approach in estimating the probability distribution. In the case the a(N)a(N) for odd NN are significantly different from zero in the two cases.

Figure A.2. Left panels, distribution of eigenvalues θl\theta_l for different domains θ(x−x0)\theta(x-x_0) with x0=−10x_0=-10, −5-5, −2-2 and 0. Right panels, the corresponding probability of finding a wave function with particle number NN in the intrinsic wave function. The legend “Cor” corresponds to the results obtained with the correct formulas discussed in this paper whereas “Wro” stands for the incorrect results obtained with the approaches discussed in the main text

It is also interesting to compare the no-pairing limit of the two probabilities to the result obtained with Löwdin formula. The results shown in Figure 1a in the main text indicate that the correct formula gives exactly the same values than Löwdin formula as expected. The a(N)a(N) is peaked at N=10N=10 as expected for x0=0x_0=0 and the particle number sumrule is satisfied but with a

Figure 1. Particle number distribution a(N)a(N) as a function of NN in a) the zero-pairing limit and b) the two-fragment solution. In both panels the domain correspond to x0=0x_0 = 0 (i.e. positive values of xx) and the results corresponding to the correct formulation are given by full line and bullets whereas the ones of the incorrect formulation are given by circles and dotted lines. In panel a) the result obtained with Löwdin formula is also presented but it coincides with the correct one. In the two-fragment case the average number of particles in the fragment to the right is 5.4.

Figure A.3. Matter density corresponding to the two cases studied here. The one on the left is obtained by filling the lowest HO orbitals with a given occupancy distribution whereas the one on the right is obtained by using the DVR basis with ad-hoc occupancies used to obtain a density corresponding to two fragments.

large contribution of odd values of NN, indicating a large degree of entanglement in this case. The incorrect formula gives a a(N)a(N) peaked at N=20N=20 which is clearly not correct.

In order to better understand the situation we have performed an additional calculation where we have used a discrete variable representation (DVR) basis by diagonalizing the position operator in the harmonic oscillator basis BNB_N used before. The eigenvectors are labeled by the values of the eigenvalues which are roughly uniformly distributed in the xx axis. The occupancies in the new basis are taken as 0 for most of the states and vk2=0.9v_k^2=0.9 for those with DVR mesh points in between -4.6 and -2.25 (larger fragment) and in between 2.78 and 3.94 (smaller fragment). In this way we obtain the matter distribution depicted in Fig. A.3 which is shown along with the matter density of the previously discussed case. The total number of particles is 14.4 whereas in the smaller fragment (to the right) the number of particles is 5.4. The two fragments are well separated and therefore entanglement effects are expected to be small.

We finally provide the particle number probability distribution obtained with the wave function corresponding to two separate fragments taking as domain positive values of xx, i.e. x0=0x_0=0. The particle number probability distribution is depicted in Figure 1b for both the correct formulation (full line) and the incorrect one (dotted line). We observe how a(N)a(N) peaks at N=6N=6 in good agreement with the average number of fragments in the right hand side fragment (5.4) but this is not the case for the incorrect calculation. The number of particle sumrule is in this case 5.4 which is the expected value in absence of entanglement. This coincidence between the sum rule and the actual average value is a consequence of having a well separated fragment (see Fig. A.3) where the entanglement of the wave function is not relevant. On the other hand, the incorrect formula gives in this case a particle number sumrule of 15.42 which is clearly incorrect.

References

References

  1. [1]C. Simenel, Particle transfer reactions with the time-dependent hartree-fock theory using a particle number projection technique, Phys. Rev. Lett. 105 (2010) 192701. doi:10.1103/PhysRevLett.105.192701 URL https://link.aps.org/doi/10.1103/PhysRevLett.105.192701DOI
  2. [2]G. Scamps, D. Lacroix, Effect of pairing on one- and two-nucleon transfer below the coulomb barrier. A time-dependent microscopic description, Phys. Rev. C 87 (2013) 014605. doi:10.1103/PhysRevC.87.014605. URL https://link.aps.org/doi/10.1103/PhysRevC.87.014605
  3. [3]L. M. Robledo, Sign of the overlap of Hartree-Fock-Bogoliubov wave functions, Phys. Rev. C 79 (2009) 021302. doi:10.1103/PhysRevC.79.021302. URL http://link.aps.org/doi/10.1103/PhysRevC.79.021302DOI
  4. [4]L. M. Robledo, Technical aspects of the evaluation of the overlap of Hartree-Fock-Bogoliubov wave functions, Phys. Rev. C 84 (2011) 014307. doi:10.1103/PhysRevC.84.014307. URL http://link.aps.org/doi/10.1103/PhysRevC.84.014307
  5. [5]G. F. Bertsch, L. M. Robledo, Symmetry restoration in Hartree-Fock-Bogoliubov based theories, Phys. Rev. Lett. 108 (2012) 042505. doi:10.1103/PhysRevLett.108.042505. URL http://link.aps.org/doi/10.1103/PhysRevLett.108.042505DOI
  6. [6]M. Verriere, N. Schunck, T. Kawano, Number of particles in fission fragments, Phys. Rev. C 100 (2019) 024612. doi:10.1103/PhysRevC.100.024612. URL https://link.aps.org/doi/10.1103/PhysRevC.100.024612DOI
  7. [7]A. Bulgac, Projection of good quantum numbers for reaction fragments, Phys. Rev. C 100 (2019) 034612. doi:10.1103/PhysRevC.100.034612. URL https://link.aps.org/doi/10.1103/PhysRevC.100.034612DOI
  8. [8]D. Lacroix, S. Ayik, Counting statistics in finite fermi systems: Illustrations with the atomic nucleus, Phys. Rev. C 101 (2020) 014310. doi:10.1103/PhysRevC.101.014310. URL https://link.aps.org/doi/10.1103/PhysRevC.101.014310DOI
  9. [9]M. Verriere, N. Schunck, D. Regnier, Microscopic calculation of fission product yields with particle-number projection, Phys. Rev. C 103 (2021) 054602. doi:10.1103/PhysRevC.103.054602. URL https://link.aps.org/doi/10.1103/PhysRevC.103.054602DOI
  10. [10]A. Bulgac, Restoring broken symmetries for nuclei and reactions fragments, Phys. Rev. C 104 (2021) 054601. doi:10.1103/PhysRevC.104.054601. URL https://link.aps.org/doi/10.1103/PhysRevC.104.054601DOI
  11. [11]A. Bulgac, I. Abdurrahman, S. Jin, K. Godbey, N. Schunck, I. Stetcu, Fission fragment intrinsic spins and their correlations, Phys. Rev. Lett. 126 (2021) 142502. doi:10.1103/PhysRevLett.126.142502. URL https://link.aps.org/doi/10.1103/PhysRevLett.126.142502DOI
  12. [12]N. Schunck, D. Regnier, Theory of nuclear fission, Progress in Particle and Nuclear Physics 125 (2022) 103963. doi:https://doi.org/10.1016/j.ppnp.2022.103963. URL https://www.sciencedirect.com/science/article/pii/S0146641022000242
  13. [13]A. Bulgac, I. Abdurrahman, K. Godbey, I. Stetcu, Fragment intrinsic spins and fragments’ relative orbital angular momentum in nuclear fission, Phys. Rev. Lett. 128 (2022) 022501. doi:10.1103/PhysRevLett.128.022501. URL https://link.aps.org/doi/10.1103/PhysRevLett.128.022501DOI
  14. [14]G. Scamps, Y. Hashimoto, Transfer probabilities for the reactions 14,20O + 40O in terms of multiple time-dependent hartree-fock-bogoliubov trajectories, Phys. Rev. C 96 (2017) 031602. doi:10.1103/PhysRevC.96.031602. URL https://link.aps.org/doi/10.1103/PhysRevC.96.031602
  15. [15]P. Marević, N. Schunck, J. Randrup, R. Vogt, Angular momentum of fission fragments from microscopic theory, Phys. Rev. C 104 (2021) L021601. doi:10.1103/PhysRevC.104.L021601. URL https://link.aps.org/doi/10.1103/PhysRevC.104.L021601DOI
  16. [16]Y. Qiang, J. Pei, K. Godbey, Quantum entanglement in nuclear fission, Physics Letters B 861 (2025) 139248. doi:https://doi.org/10.1016/j.physletb.2025.139248. URL https://www.sciencedirect.com/science/article/pii/S0370269325000085
  17. [17]B. Bayman, A derivation of the pairing-correlation method, Nuclear Physics 15 (1960) 33–38. doi:https://doi.org/10.1016/0029-5582(60)90279-0. URL https://www.sciencedirect.com/science/article/pii/0029558260902790
  18. [18]K. Dietrich, H. J. Mang, J. H. Pradal, Conservation of particle number in the nuclear pairing model, Phys. Rev. 135 (1964) B22–B34. doi:10.1103/PhysRev.135.B22. URL https://link.aps.org/doi/10.1103/PhysRev.135.B22DOI
  19. [19]H. J. Mang, J. O. Rasmussen, M. Rho, Comparisons of bcs nuclear wave functions with exact solutions. ii. particle-number-conserving method, Phys. Rev. 141 (1966) 941–948. doi:10.1103/PhysRev.141.941. URL https://link.aps.org/doi/10.1103/PhysRev.141.941
  20. [20]P. Ring, P. Schuck, The Nuclear Many-Body Problem, Springer-Verlag, Berlin, 2000.
  21. [21]L. M. Robledo, Practical formulation of the extended Wick theorem and the Onishi formula, Phys. Rev. C 50 (1994) 2874–2881. doi:10.1103/PhysRevC.50.2874. URL http://link.aps.org/doi/10.1103/PhysRevC.50.2874DOI
  22. [22]L. M. Robledo, Formulation of the generator coordinate method with arbitrary bases, Phys. Rev. C 105 (2022) L021307. doi:10.1103/PhysRevC.105.L021307. URL https://link.aps.org/doi/10.1103/PhysRevC.105.L021307DOI
  23. [23]L. M. Robledo, Pfaffian formulas for non equivalent bases (2025). arXiv:2503.23539. URL https://arxiv.org/abs/2503.23539
  24. [24]P.-O. Löwdin, Quantum theory of many-particle systems. II. Study of the ordinary Hartree-Fock approximation, Phys. Rev. 97 (1955) 1490–1508. doi:10.1103/PhysRev.97.1490. URL https://link.aps.org/doi/10.1103/PhysRev.97.1490DOI
  25. [25]P. Marević, N. Schunck, Fission of 240Pu with symmetry-restored density functional theory, Phys. Rev. Lett. 125 (2020) 102504. doi:10.1103/PhysRevLett.125.102504. URL https://link.aps.org/doi/10.1103/PhysRevLett.125.102504DOI
  26. [26]L. M. Robledo, Operator overlaps in harmonic oscillator bases with different oscillator lengths, Phys. Rev. C 105 (2022) 044317. doi:10.1103/PhysRevC.105.044317. URL https://link.aps.org/doi/10.1103/PhysRevC.105.044317DOI
  27. [27]J. N. Wilson, D. Thisse, M. Lebois, N. Jovancević, D. Gjestvang, R. Canavan, M. Rudigier, D. Étasse, R.-B. Gerst, L. Gaudefroy, E. Adamska, P. Adsley, A. Algora, M. Babo, K. Belvedere, J. Benito, G. Benzoni, A. Blazhev, A. Boso, S. Bottoni, M. Bunce, R. Chakma, N. Cieplika-Oryńczak, S. Courtin, M. L. Cortés, P. Davies, C. Delafosse, M. Fallot, B. Fornal, L. Fraile, A. Gottardo, V. Guadilla, G. Häfner, K. Hauschild, M. Heine, C. Henrich, I. Homm, F. Ibrahim, Å. W. Iskra, P. Ivanov, S. Jazrawi, A. Korgul, P. Koseoglou, T. Kröll, T. Kurtukian-Nieto, L. Le Meur, S. Leoni, J. Ljungvall, A. Lopez-Martens, R. Lozeva, I. Matea, K. Miernik, J. Nemer, S. Oberstedt, W. Paulsen, M. Piersa, Y. Popovitch, C. Porzio, L. Qi, D. Ralet, P. H. Regan, C. Reznykova, V. Sánchez-Tembleque, S. Siem, C. Schmitt, P.-A. Säderström, C. Säder, G. Tocaben, V. Vedia, D. Verney, N. Warr, B. Wasilewska, J. Wiederhold, M. Yavahchova, F. Zeiser, S. Ziliani, Angular momentum generation in nuclear fission, Nature 590 (7847) (2021) 566–570. doi:10.1038/s41586-021-03304-w. URL https://doi.org/10.1038/s41586-021-03304-w
  28. [28]P.-O. Löwdin, Quantum theory of many-particle systems. I. Physical interpretations by means of density matrices, natural spin-orbitals, and convergence problems in the method of configurational interaction, Phys. Rev. 97 (1955) 1474–1489. doi:10.1103/PhysRev.97.1474. URL http://link.aps.org/doi/10.1103/PhysRev.97.1474DOI
  29. [29]C. Bloch, A. Messiah, The canonical form of an antisymmetric tensor and its application to the theory of superconductivity, Nuclear Physics 39 (1962) 95–106. doi:https://doi.org/10.1016/0029-5582(62)90377-2. URL https://www.sciencedirect.com/science/article/pii/0029558262903772
  30. [30]B. Zumino, Normal forms of complex matrices, Journal of Math- ematical Physics 3 (5) (1962) 1055–1057. arXiv:https: //pubs.aip.org/aip/jmp/article-pdf/3/5/1055/19196721/1055_1_online.pdf, doi:10.1063/1.1724294 URL https://doi.org/10.1063/1.1724294
  31. [31]I. Klich, Lower entropy bounds and particle number fluctuations in a fermi sea, Journal of Physics A: Mathematical and General 39 (4) (2006) L85. doi:10.1088/0305-4470/39/4/L02. URL https://dx.doi.org/10.1088/0305-4470/39/4/L02

Paper details

vlnc:W_eqdnucp6
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents