betaFeedback

WorksPhysics and Astronomy

Symmetry and Redundancy in Cremer-Pople Phases for Ring Puckering

Abstract

Cremer-Pople (CP) coordinates provide an exact description of ring puckering, yet direct sampling of their phases generates many symmetry-equivalent representations of the same puckering pattern. We develop a general analytical framework that identifies these equivalences before Cartesian reconstruction and geometry optimization. For homogeneous rings of arbitrary size, explicit transformations of the CP phases and the signed special coordinate of even-membered rings generate the complete symmetry orbit of the ordered elevation pattern. We classify the redundancies into four types and derive a factorized orbit-size formula linking CP multiplicity to puckering symmetry. These results yield exact single-phase fundamental intervals and sequentially reduced multiphase covering domains, including the restriction to a positive special coordinate. The framework is extended to rings containing one heteroatom or two identical heteroatoms, with reductions determined by the operations that preserve their placement. We also introduce a semidiscrete conformational nomenclature that combines symbolic amplitude distributions with explicit active phases. In an application to cyclohexadecane, symmetry reduction alone decreases the number of sampled CP points by approximately a factor of thirty. Combined with selective phase-grid coarsening for three ordinary modes, the reduction approaches two orders of magnitude. These developments provide a systematic basis for generating, classifying, and efficiently sampling ring puckering patterns in CP space.

Introduction

Flexible rings undergo large-amplitude puckering motions that cannot be described by a single dihedral angle. Their conformation is instead a collective property of all ring atoms, and its proper description has been studied for more than half a century. A classic example is the analysis of cyclopentane by Kilpatrick, Pitzer, and Spitzer,[1] who showed that the puckered forms of a five-membered ring can be described by just two parameters, i.e., by an amplitude and by a phase angle, with interconversion taking place through pseudorotation. This established the amplitude–phase picture of ring puckering and strongly influenced later work on cyclic conformational dynamics.

The description was later generalized to rings of arbitrary size through the Cremer–Pople (CP) coordinates.[2] These coordinates define a unique mean plane and describe the out-of-plane elevations {zj}\{z_j\} of the ring atoms relative to this plane using N−3N-3 independent puckering coordinates. These consist of an amplitude and a phase for each ordinary mode, and, for even-membered rings, one additional signed coordinate with no associated phase. For an NN-membered ring, the CP amplitudes qmq_m and phases ϕm\phi_m are defined by$[2]

qmcos⁡ϕm=2N∑j=1Nzjcos⁡(2πm(j−1)N),(1)q_m \cos\phi_m = \sqrt{\frac{2}{N}} \sum_{j=1}^{N} z_j \cos\left(\frac{2\pi m(j-1)}{N}\right), \tag*{(1)}
qmsin⁡ϕm=−2N∑j=1Nzjsin⁡(2πm(j−1)N),(2)q_m \sin\phi_m = -\sqrt{\frac{2}{N}} \sum_{j=1}^{N} z_j \sin\left(\frac{2\pi m(j-1)}{N}\right), \tag*{(2)}

with m=2,3,…,Mm=2,3,\ldots,M, where M=(N−1)/2M=(N-1)/2 if NN is odd and M=N/2−1M=N/2-1 if NN is even.

If NN is even, there is one additional coordinate at m=N/2m=N/2,

qN/2=1N∑j=1N(−1)j−1zj,(3)q_{N/2} = \sqrt{\frac{1}{N}} \sum_{j=1}^{N} (-1)^{j-1}z_j, \tag*{(3)}

which has no phase and may take either sign. The inverse relation gives the elevations as functions of the amplitudes and phases. For odd NN,

zj=2N∑m=2Mqmcos⁡(ϕm+2πm(j−1)N),(4)z_j = \sqrt{\frac{2}{N}} \sum_{m=2}^{M} q_m \cos\left(\phi_m+\frac{2\pi m(j-1)}{N}\right), \tag*{(4)}

and for even NN,

zj=2N∑m=2Mqmcos⁡(ϕm+2πm(j−1)N)+1NqN/2(−1)j−1.(5)\begin{aligned} z_j &= \sqrt{\frac{2}{N}} \sum_{m=2}^{M} q_m \cos\left(\phi_m+\frac{2\pi m(j-1)}{N}\right) \\ &\quad+ \sqrt{\frac{1}{N}}q_{N/2}(-1)^{j-1}. \tag*{(5)} \end{aligned}

Because CP coordinates describe the out-of-plane elevations of the ring atoms, we examine the symmetry of this elevation pattern on a reference regular polygon. This symmetry need not coincide with the point-group symmetry of the full three-dimensional molecular structure. This is the same level at which Pickett and Strauss[3] analyzed ring puckering, treating cycloalkane conformations as symmetry-adapted out-of-plane displacements of a planar regular NN-gon of DNhD_{Nh} symmetry. Their analysis classified the possible puckering modes, counted the symmetry-related forms, identified low-barrier pseudorotational paths, and derived selection rules for the associated conformational dynamics.

Boeyens and Evans[4] later established the one-to-one correspondence between the symmetry-adapted coordinates of Pickett and Strauss and the Cremer–Pople puckering parameters, showing that the two formalisms are mathematically equivalent. They also emphasized that the symmetry of the displacement pattern {zj}\{z_j\} need not coincide with that of the full three-dimensional molecular structure. More recently, Barone has incorporated local curvilinear ring-puckering components into a general construction of symmetry-adapted, non-redundant internal coordinates.1

Previous studies have exploited CP symmetry in conformational classification and sampling. Allen et al.⁶⁻⁸ partitioned the CP conformational spaces of cycloheptane and cyclooctane into symmetry-equivalent asymmetric units. Symmetry-based reductions have also been used for pyranose sugars energy landscapes,⁹ seven-membered-ring metadynamics,¹⁰ and potential-energy scans of four- to six-membered rings.¹¹

Other studies improved sampling by restricting the explored region of CP space¹²,¹³ or by using compact CP collective variables.¹⁴,¹⁵ These applications motivate a unified formulation of the symmetry action in CP space and its use in constructing reduced search domains for arbitrary ring size.

Our previous work¹⁶ introduced a ring reconstruction algorithm (RRA) for generating chemically reasonable geometries from CP coordinates and performing systematic conformational searches. However, symmetry multiplicities were determined only a posteriori by comparing optimized structures. Identifying equivalent CP representations before reconstruction and optimization can avoid redundant calculations while retaining at least one representative of every symmetry-inequivalent puckering pattern.

In this work, we develop an analytical framework for rings of arbitrary size that connects CP redundancy with puckering symmetry and provides symmetry-reduced domains for conformational searches. First, we determine which symmetry operations produce identical CP coordinates, establishing the multiplicity and number of distinct CP images before Cartesian reconstruction and geometry optimization. Second, we derive symmetry-reduced phase domains for conformational searches, retaining at least one representative of every symmetry-inequivalent puckering pattern. The underlying transformations relate CP representations through cyclic relabeling, index reflection, or inversion of the elevations through the mean plane; each resulting representation is referred to as a CP image.

The reductions are developed for unsubstituted cycloalkanes and rings containing one heteroatom or two identical heteroatoms, retaining only index operations that preserve the atom-type pattern. Their use for other systems requires preservation of chemical connectivity and specified stereochemistry. For energy searches in an achiral environment, fully mirror-related molecular geometries are included in the same equivalence class because they have identical energies.

Specifically, we derive (i) explicit transformations of the CP phases and, for even-membered rings, the signed special coordinate; (ii) a closed-form orbit-size formula relating CP multiplicity to puckering symmetry; and (iii) exact single-phase fundamental intervals and sequentially reduced multiphase covering domains with residual boundary redundancy. Their integration with the reconstruction and search strategy of Ref. 16 is illustrated by a systematic search for low-energy conformers of cyclohexadecane.

Symmetry Transformations of Cremer–Pople Coordinates

When a conformational search is carried out directly in CP space, different phase points may encode the same puckering conformation because of the symmetry of the reference regular polygon. The relevant operations act on the ordered index set j=1,…,Nj=1,\ldots,N and, when appropriate, on the global sign of the elevations, so that the resulting transformed CP images belong to the same equivalence class. The goal of this section is to derive the explicit phase transformations that generate all such equivalent images.

All transformations are derived using the original Cremer–Pople orientation convention, in which the positive CP normal is the one from which the ring atoms are viewed in clockwise order. Thus, whenever an index transformation reverses this orientation, the transformed CP image is re-expressed using the original reference normal.

Throughout this derivation, the out-of-plane elevations introduced previously are treated as an ordered vector z=(z1,…,zN)\mathbf{z}=(z_1,\ldots,z_N), expressed in the common CP-normal orientation. This vector, rather than the full Cartesian coordinates of the molecule, is the primary object of the analysis. Accordingly, the labels shown in the figures denote the ordered CP elevation indices, i.e., the positions of the components of z\mathbf{z}, rather than fixed chemical atom labels.

Indistinguishable atoms

Let M={m1,…,mr}\mathcal{M}=\{m_1,\ldots,m_r\} denote the set of active phase-bearing modes, i.e., those indices m∈{2,…,M}m\in\{2,\ldots,M\} for which qm≠0q_m\ne0. The indices are ordered as m1<⋯<mrm_1<\cdots<m_r, and r=∣M∣r=|\mathcal{M}| is the number of active phase-bearing modes.

For even-membered rings with a nonzero special coordinate, the complete active-mode set is M±=M∪{N/2}\mathcal{M}_{\pm}=\mathcal{M}\cup\{N/2\}, with r±=r+1r_{\pm}=r+1. When sampling is restricted to qN/2>0q_{N/2}>0, we denote this set by M+\mathcal{M}_{+} and its size by r+=r+1r_{+}=r+1. If the special coordinate is absent or zero, the complete active-mode set is M\mathcal{M}.

All operations that involve ring-index changes are understood modulo NN, and all phase transformations are taken modulo 2π2\pi. For indistinguishable ring atoms, the relevant operations are: (i) cyclic index rotation, (ii) index reflection with orientation reversal, and (iii) global inversion of the elevations.

(i) Cyclic rotation of indices. Let

Rk:zj↦zj′=zj+k,k=0,1,…,N−1,(6)\mathcal{R}_k:z_j\mapsto z'_j=z_{j+k},\qquad k=0,1,\ldots,N-1, \tag*{(6)}

where the indices are taken modulo NN. In the present convention, Rk\mathcal{R}_k is a relabeling operation: the elevation assigned to ring position jj after the operation is the elevation that, before the operation, belonged to position j+kj+k, as illustrated in Figure 1. Substituting Eq. (4) or Eq. (5) into Eq. (6), the contribution from the phase-bearing modes becomes

Cyclic rotation of the ordered CP elevation vector for $N=5$

Figure 1. Cyclic rotation of the ordered CP elevation vector for N=5N=5: (a) reference CP image, mean plane, and direction associated with the C51C^1_5 transformation; (b) transformed CP image, in which the elevation component assigned to position jj is taken from the original component at position j+1j+1.

zj,ϕ′=2N∑m∈Mqmcos⁡(ϕm+2πm(j+k−1)N),(7)z'_{j,\phi}=\sqrt{\frac{2}{N}}\sum_{m\in\mathcal{M}}q_m\cos\left(\phi_m+\frac{2\pi m(j+k-1)}{N}\right), \tag*{(7)}

so that the transformed phase is

ϕm′=ϕm+2πmkN.(8)\phi'_m=\phi_m+\frac{2\pi mk}{N}. \tag*{(8)}

Here zj,ϕ′z'_{j,\phi} denotes the contribution from the phase-bearing modes; for odd NN, it is the complete transformed elevation. If NN is even, the special coordinate transforms as

qN/2′=(−1)kqN/2.(9)q'_{N/2}=(-1)^kq_{N/2}. \tag*{(9)}

The factor (−1)k(-1)^k reflects the alternating character of the m=N/2m=N/2 mode; shifting the labeling by one ring position reverses its sign.

(ii) Index reflection with orientation reversal. Consider the index operation

Cℓ:j↦j′=N−j+ℓ+2,ℓ=0,1,…,N−1,(10)C_\ell:j\mapsto j'=N-j+\ell+2,\qquad\ell=0,1,\ldots,N-1, \tag*{(10)}

again modulo NN. This operation reflects the ordered index set of the reference polygon. The parameter ℓ\ell specifies the position of the reflection axis. For even NN, even values of ℓ\ell correspond to axes through two opposite atoms, whereas odd values correspond to axes through the midpoints of two opposite bonds. For odd NN, each reflection axis passes through one atom and the midpoint of the opposite bond; the fixed atom satisfies 2j=ℓ+22j=\ell+2 modulo NN.

Because this reflection reverses the index orientation, the CP normal is also reversed. Following the Cremer–Pople orientation convention stated above, the transformed elevations are re-expressed with the reference CP normal. Writing jj as the output index, the resulting action is

zj′=−zN−j+ℓ+2.(11)z'_j=-z_{N-j+\ell+2}. \tag*{(11)}

The minus sign in Eq. (11) accounts for the reversal of the CP normal. The transformed elevations are obtained by substituting the CP expansion into Equation (11). Absorbing the overall minus sign into the cosine using −cos⁡θ=cos⁡(θ+π)-\cos\theta=\cos(\theta+\pi) gives

zj,ϕ′=2N∑m∈Mqmcos⁡[2πm(j−1)N+2πmℓN+ϕm+π].(12)z'_{j,\phi}=\sqrt{\frac{2}{N}}\sum_{m\in\mathcal{M}}q_m\cos\left[\frac{2\pi m(j-1)}{N}+\frac{2\pi m\ell}{N}+\phi_m+\pi\right]. \tag*{(12)}

Using the evenness of the cosine, this expression becomes

zj,ϕ′=2N∑m∈Mqmcos⁡[2πm(j−1)N−2πmℓN−ϕm−π].(13)z'_{j,\phi}=\sqrt{\frac{2}{N}}\sum_{m\in\mathcal{M}}q_m\cos\left[\frac{2\pi m(j-1)}{N}-\frac{2\pi m\ell}{N}-\phi_m-\pi\right]. \tag*{(13)}

The comparison with the standard CP expansion gives

ϕm′=−ϕm−2πmℓN+π.(14)\phi'_m=-\phi_m-\frac{2\pi m\ell}{N}+\pi. \tag*{(14)}

Thus, reflected indexing reverses each phase and introduces an axis-dependent shift, while the reversal of the CP normal contributes the additional π\pi offset.

If NN is even, the special mode m=N/2m=N/2 must be treated separately because it has no associated phase variable. Since NN is even, writing

N−j+ℓ+1=(j−1)+(N−2j+ℓ+2)(15)N-j+\ell+1=(j-1)+(N-2j+\ell+2) \tag*{(15)}

gives

(−1)N−j+ℓ+1=(−1)j−1(−1)N−2j+ℓ+2=(−1)j−1(−1)ℓ.(16)\begin{aligned} (-1)^{N-j+\ell+1}&=(-1)^{j-1}(-1)^{N-2j+\ell+2}\\ &=(-1)^{j-1}(-1)^\ell. \tag*{(16)} \end{aligned}

Thus, including the minus sign induced by CP-normal reversal, the transformed contribution becomes

zj,N/2′=(−1)ℓ+1qN/2N(−1)j−1,(17)z'_{j,N/2}=\frac{(-1)^{\ell+1}q_{N/2}}{\sqrt{N}}(-1)^{j-1}, \tag*{(17)}

so that the special coordinate transforms as

qN/2′=(−1)ℓ+1qN/2.(18)q'_{N/2}=(-1)^{\ell+1}q_{N/2}. \tag*{(18)}

Therefore, for the phase-bearing modes, this normal reversal is represented by a π\pi shift of the phases; for the special mode, it is represented by a sign change of the amplitude.

(iii) Global inversion of the elevations. Global inversion reflects all atomic elevations through the mean plane. This operation can be encoded by a sign parameter s∈{+1,−1}s \in\{+1,-1\}:

Is:zj′=szj,s∈{+1,−1}.(19)\mathcal{I}_{s}: z'_j = sz_j,\qquad s \in\{+1,-1\}. \tag*{(19)}

For s=+1s=+1 this is the identity, while for s=−1s=-1 it reverses the sign of every elevation. Since −cos⁡(θ)=cos⁡(θ+π)-\cos(\theta)=\cos(\theta+\pi), inversion acts on the phase variables as

ϕm′=ϕm+π1−s2.(20)\phi'_m = \phi_m + \pi\frac{1-s}{2}. \tag*{(20)}

The effect of the global inversion is the same as applying a σh\sigma_h operation over the geometry as shown in Figure 2.

Global inversion of the elevations for N = 5

Figure 2. Global inversion of the elevations for N=5N=5: (a) shows the reference ring, mean plane, and the σh\sigma_h plane which coincides with the mean plane; (b) shows the ring after the inversion.

For even NN, the special coordinate transforms as

qN/2′=sqN/2.(21)q'_{N/2}=s q_{N/2}. \tag*{(21)}

Combining index reflection (when present), cyclic rotation, and global inversion gives the general transformation law for the phase variables:

ϕm′=(−1)pϕm+2πmkN−p2πmℓN+pπ+π1−s2,(22)\phi'_m = (-1)^p\phi_m+\frac{2\pi m k}{N}-p\frac{2\pi m\ell}{N}+p\pi+\pi\frac{1-s}{2}, \tag*{(22)}

where p∈{0,1}p \in\{0,1\} distinguishes the rotation branch (p=0p=0) from the reflection branch (p=1p=1), k,ℓ=0,…,N−1k,\ell=0,\ldots,N-1, and s∈{±1}s \in\{\pm1\}. The composition convention in Eq. (22) applies the index reflection first, followed by the cyclic rotation and global inversion. All phases are understood modulo 2π2\pi.

In the reflection branch, varying ℓ\ell at fixed k=0k=0, or varying kk at fixed ℓ\ell, generates the same set of index reflections. Using NN rotations and NN reflections, each with two choices of ss, gives 4N4N formal operations.

For even NN, the special coordinate transforms as

qN/2′=s(−1)k+p(ℓ+1)qN/2.(23)q'_{N/2}=s(-1)^{k+p(\ell+1)}q_{N/2}. \tag*{(23)}

The cyclic permutation contributes (−1)k(-1)^k, while reflected indexing contributes (−1)ℓ(-1)^\ell when p=1p=1. The additional factor (−1)p(-1)^p accounts for the reversal of the CP normal, and ss represents the independent global inversion.

These transformations can also be associated with the conventional symmetry-operation labels of the regular reference polygon. The correspondence between the parameters (k,p,ℓ,s)(k,p,\ell,s) and the DNhD_{Nh} labels is listed in Table 1 and II. In both tables, ℓ\ell is irrelevant in the rotation branch, and k=0k=0 is used in the reflection branch to avoid duplicate parametrizations.

ppsskk or ℓ\ellϕm′\phi'_mDNhD_{Nh}
0+1+10ϕm\phi_mEE
0+1+1k=1,…,N−1k=1,\ldots,N-1ϕm+κmk\phi_m+\kappa_m kCNkC_N^k
0−1-10ϕm+π\phi_m+\piσh\sigma_h
0−1-1k=1,…,N−1k=1,\ldots,N-1ϕm+κmk+π\phi_m+\kappa_m k+\piSNk aS_N^{k\,a}
1+1+1ℓ=0,…,N−1\ell=0,\ldots,N-1−ϕm−κmℓ+π-\phi_m-\kappa_m\ell+\piC2ℓC_2^{\ell}
1−1-1ℓ=0,…,N−1\ell=0,\ldots,N-1−ϕm−κmℓ-\phi_m-\kappa_m\ellσvℓ\sigma_v^{\ell}

Table 1. Parameters (k,p,ℓ,s)(k,p,\ell,s), induced phase mapping ϕm′\phi'_m, and corresponding DNhD_{Nh} labels for odd NN, where κm=2πm/N\kappa_m=2\pi m/N. Reflection rows use k=0k=0. All phase mappings are understood modulo 2π2\pi.

Symmetry-transformed and Cartesian-derived CP coordinates

The transformations derived above act directly on the CP representation, that is, on the ordered elevation vector and on the corresponding CP coordinates. In the convention used throughout this work, the ring labels denote CP elevation indices, i.e. the positions of the components of z=(z1,…,zN)\mathbf{z}=(z_1,\ldots,z_N), rather than material atom labels followed through a Cartesian transformation.

This distinction becomes important when CP coordinates are recomputed from a transformed Cartesian geometry. This situation is illustrated in Figure 3 for the vertical reflection σvℓ=0\sigma_v^{\ell=0} of an envelope conformation of cyclopentane. In the reference CP image [Figure 3(a)], the ordered CP indices define the positive CP normal. After the selected-index operation [Figure 3(b)], two equivalent readings are possible, depending on what the labels are taken to represent. If the labels denote

Reflected indexing with orientation reversal for an envelope conformation of cyclopentane

Figure 3. Reflected indexing with orientation reversal for an envelope conformation of cyclopentane: (a) reference CP image, mean plane, vertical reflection plane σvℓ=0\sigma_v^{\ell=0}, and oriented CP normal in green; (b) reflected CP indexing, which reverses the spatial cyclic order associated with the Cartesian representation.

ppsskk or ℓ\ellϕm′\phi'_mqN/2′q'_{N/2}label
0+1+10ϕm\phi_m+qN/2+q_{N/2}EE
0+1+1k=1,…,N−1k=1,\ldots,N-1ϕm+κmk\phi_m+\kappa_m k(−1)kqN/2(-1)^kq_{N/2}CNkC_N^k
1+1+1ℓ=0,2,…,N−2\ell=0,2,\ldots,N-2−ϕm−κmℓ+π-\phi_m-\kappa_m\ell+\pi−qN/2-q_{N/2}C2′,ℓC_2^{\prime,\ell}
1+1+1ℓ=1,3,…,N−1\ell=1,3,\ldots,N-1−ϕm−κmℓ+π-\phi_m-\kappa_m\ell+\pi+qN/2+q_{N/2}C2′′,ℓC_2^{\prime\prime,\ell}
0−1-10ϕm+π\phi_m+\pi−qN/2-q_{N/2}σh\sigma_h
0−1-1k=1,…,N−1k=1,\ldots,N-1ϕm+κmk+π\phi_m+\kappa_m k+\pi−(−1)kqN/2-(-1)^kq_{N/2}SNkS_N^k
1−1-1ℓ=0,2,…,N−2\ell=0,2,\ldots,N-2−ϕm−κmℓ-\phi_m-\kappa_m\ell+qN/2+q_{N/2}σvℓ\sigma_v^\ell
1−1-1ℓ=1,3,…,N−1\ell=1,3,\ldots,N-1−ϕm−κmℓ-\phi_m-\kappa_m\ell−qN/2-q_{N/2}σdℓ\sigma_d^\ell

Table II. Parameters (k,p,ℓ,s)(k,p,\ell,s), induced phase mapping ϕm′\phi'_m, transformation of qN/2q_{N/2}, and DNhD_{Nh} label for even NN, where κm=2πm/N\kappa_m=2\pi m/N. The notation SNkS_N^k follows the convention specified in Table I. Reflection rows use k=0k=0. The special-coordinate mappings include the effect of CP-normal reversal. All phase mappings are understood modulo 2π2\pi.

CP elevation indices, i.e. fixed positions in the ordered vector z\mathbf{z}, the reflected image is projected using the fixed CP-index convention and the recomputed CP coordinates coincide with the analytical transformation. In this convention, the original ordered vector of elevations (z1,z2,z3,z4,z5)(z_1,z_2,z_3,z_4,z_5) is transformed by σvℓ=0\sigma_v^{\ell=0} into (z1,z5,z4,z3,z2)(z_1,z_5,z_4,z_3,z_2).

By contrast, if the labels are followed as material atom labels in the Cartesian ring file, the same reflected geometry is read with the opposite cyclic orientation. This reindexing reverses the CP normal defined by the ordered atoms. Consequently, an additional 180° phase shift for the phase-bearing modes must be included in order to express the Cartesian-derived CP coordinates in the same convention as the analytical expression. For even-membered rings, the same change of normal also reverses the sign of the special coordinate.

This shift is purely conventional because once the reflected structure is expressed again in the fixed CP-index convention, or with the same positive CP normal as the reference structure, the Cartesian-derived and analytically transformed CP coordinates are consistent.

Rings with heteroatoms

When distinguishable atoms (nodes in the ring) are present, the allowed symmetry transformations are restricted to those that preserve the heteroatom pattern, either by fixing each heteroatom or by exchanging identical ones. As in the homogeneous case, the resulting equivalence is defined at the level of the CP image, not necessarily at the level of the full three-dimensional molecular geometry. Global inversion Is\mathcal{I}_s remains unrestricted but cyclic rotations are allowed only if they preserve the heteroatom pattern.

One heteroatom. If one node jaj_a differs from the rest, cyclic rotations Rk\mathcal{R}_k are restricted to the identity k=0k=0. A reflection Cℓ~\mathcal{C}_{\tilde{\ell}} is allowed only if it fixes the special node, which requires ℓ~≡2(ja−1)(modN)\tilde{\ell}\equiv2(j_a-1)\pmod{N}. The corresponding axis passes through the heteroatom. For even NN, its axis index is necessarily even. The allowed operations therefore form a four-element group of C2vC_{2v} type in puckering space, generated by {E,σv,σh,C2}\{E,\sigma_v,\sigma_h,C_2\}. All nontrivial cyclic rotations are forbidden because they move the heteroatom to a different ring position, producing a distinct labeled puckering conformation. Only the four operations that preserve the heteroatom site remain redundant. The allowed phase mappings are listed in Table III.

Two identical heteroatoms. If two nodes ja<jbj_a<j_b are identical to each other but distinct from the rest, the allowed symmetry operations depend on their relative placement.

Two types of reflection Cℓ~\mathcal{C}_{\tilde{\ell}} are possible:

  • Swapping axis (Cℓ~(ja)=jb, Cℓ~(jb)=ja)(\mathcal{C}_{\tilde{\ell}}(j_a)=j_b,\ \mathcal{C}_{\tilde{\ell}}(j_b)=j_a): always allowed, with

ℓ~=ja+jb−2.(24)\tilde{\ell}=j_a+j_b-2. \tag*{(24)}

For even NN, an even value of ℓ~\tilde{\ell} corresponds to an axis through two opposite atoms, whereas an odd value corresponds to an axis through the midpoints of two opposite bonds. For odd NN, every reflection axis passes through one atom and the midpoint of the opposite bond.

  • Fixing axis (C~ℓ(ja)=ja\widetilde{\mathcal{C}}_{\ell}(j_a)=j_a, C~ℓ(jb)=jb\widetilde{\mathcal{C}}_{\ell}(j_b)=j_b): requires jb=ja+N/2j_b=j_a+N/2 (antipodal placement, even NN only), with ℓ~1=2(ja−1)\widetilde{\ell}_1=2(j_a-1).

When the two heteroatoms are not antipodal (jb−ja≠N/2j_b-j_a\ne N/2), only the swapping axis is available and rotations are restricted to k=0k=0. The residual action on the CP image again has C2vC_{2v} type, with ℓ~=ja+jb−2\widetilde{\ell}=j_a+j_b-2; the phase mappings take the same form as in the one-heteroatom case and are given by Table III, with ℓ~\widetilde{\ell} now defined by Equation (24).

ppssϕm′\phi'_mC2vC_{2v}
0+1+1ϕm\phi_mEE
1+1+1−ϕm−κmℓ~+π-\phi_m-\kappa_m\widetilde{\ell}+\piC2C_2
0−1-1ϕm+π\phi_m+\piσh\sigma_h
1−1-1−ϕm−κmℓ~-\phi_m-\kappa_m\widetilde{\ell}σv\sigma_v

Table III. Generator parameters (p,s)(p,s), induced phase mapping ϕm′\phi'_m, and corresponding C2vC_{2v} operation for heteroatom-containing rings, with fixed reflection-axis index ℓ~\widetilde{\ell} and k=0k=0.

When the two heteroatoms are antipodal (jb=ja+N/2j_b=j_a+N/2, even NN only), both reflection axes are available and the rotation k=N/2k=N/2, which exchanges ja↔jbj_a\leftrightarrow j_b, is also allowed. The residual action on the CP image therefore extends to an eight-element group of D2hD_{2h} type. Here the rotation k=N/2k=N/2 is allowed because it exchanges the two identical heteroatoms. The full set of phase mappings is given in Table IV. Compared to the C2vC_{2v} case, the additional operations are the antipodal rotation C2C_2, the swapping reflection σv′\sigma'_v, inversion ii, and the second in-plane twofold operation C2′′C''_2.

ppssk,ℓ~k,\widetilde{\ell}ϕm′\phi'_mqN/2′q'_{N/2}Label
0+1+10,−−0,--ϕm\phi_m+qN/2+q_{N/2}EE
0+1+1N/2,−−N/2,--ϕm+πm\phi_m+\pi m(−1)N/2qN/2(-1)^{N/2}q_{N/2}C2C_2
1+1+10,ℓ~10,\widetilde{\ell}_1−ϕm−κmℓ~1+π-\phi_m-\kappa_m\widetilde{\ell}_1+\pi−qN/2-q_{N/2}C2′C'_2
1+1+10,ℓ~20,\widetilde{\ell}_2−ϕm−κmℓ~2+π-\phi_m-\kappa_m\widetilde{\ell}_2+\pi(−1)ℓ~2+1qN/2(-1)^{\widetilde{\ell}_2+1}q_{N/2}C2′′C''_2
0−1-10,−−0,--ϕm+π\phi_m+\pi−qN/2-q_{N/2}σh\sigma_h
0−1-1N/2,−−N/2,--ϕm+πm+π\phi_m+\pi m+\pi(−1)N/2+1qN/2(-1)^{N/2+1}q_{N/2}ii
1−1-10,ℓ~10,\widetilde{\ell}_1−ϕm−κmℓ~1-\phi_m-\kappa_m\widetilde{\ell}_1+qN/2+q_{N/2}σv\sigma_v
1−1-10,ℓ~20,\widetilde{\ell}_2−ϕm−κmℓ~2-\phi_m-\kappa_m\widetilde{\ell}_2(−1)ℓ~2qN/2(-1)^{\widetilde{\ell}_2}q_{N/2}σv′\sigma'_v

Table IV. Generator parameters, phase mapping ϕm′\phi'_m, transformation of the special coordinate qN/2q_{N/2}, and reference-subgroup labels for a ring with two identical heteroatoms at antipodal positions. Here κm=2πm/N\kappa_m=2\pi m/N, ℓ~1=2(ja−1)\widetilde{\ell}_1=2(j_a-1), and ℓ~2=ℓ~1+N/2\widetilde{\ell}_2=\widetilde{\ell}_1+N/2, with axis indices understood modulo NN. All phase mappings are understood modulo 2π2\pi.

The labels C2′C'_2 and C2′′C''_2 distinguish the two mutually perpendicular in-plane axes of this D2hD_{2h}-type subgroup: the first passes through the heteroatoms, whereas the second exchanges them. Likewise, σv\sigma_v and σv′\sigma'_v denote the vertical planes containing these axes. These are local subgroup labels and need not coincide with the prime and double-prime conventions of the parent DNhD_{Nh} group.

Redundancy, Orbit Reduction, and Puckering Symmetry

The transformations derived in Sec. II generate 4N4N formal CP images of a homogeneous ring. Some of these images may coincide because different operations produce the same ordered elevation pattern. The collection of distinct images is called the CP orbit.

The orbit is determined by counting the operations that reproduce the reference elevation pattern, z=(z1,…,zN)\mathbf{z}=(z_1,\ldots,z_N). Such operations leave all active CP coordinates unchanged, including the signed coordinate qN/2q_{N/2} for even-membered rings. Phases associated with zero amplitudes are disregarded because they do not affect the elevation pattern.

We denote the number of operations that reproduce the reference pattern by mCPm_{\mathrm{CP}}. This number is also the multiplicity of each distinct CP image. Therefore, for a homogeneous ring the CP orbit is,

∣O({ϕm})∣=4NmCP.(25)\lvert\mathcal{O}(\{\phi_m\})\rvert=\frac{4N}{m_{\mathrm{CP}}}. \tag*{(25)}

If only the identity reproduces the reference pattern, mCP=1m_{\mathrm{CP}}=1 and all 4N4N images are distinct. Each additional operation that leaves the pattern unchanged is considered redundant and reduces the number of distinct images.

Redundancy types and orbit size

The redundancies can be classified into four types:

  • Type-R: two cyclic shifts produce the same CP image. These redundancies correspond to rotations about the axis normal to the reference plane.

  • Type-C: reflected indexing with normal reversal produces the same CP image as a cyclic shift. These redundancies correspond to twofold rotations about axes lying in the reference plane (C2,⊥C_{2,\perp}).

  • Type-I: global inversion of the elevations produces a CP image already obtained by a cyclic shift. These redundancies correspond to cyclic shifts combined with elevation inversion.

  • Type-P: reflected indexing combined with global inversion produces the same CP image as a cyclic shift. These redundancies correspond to mirror-plane operations.

The conditions for each type are derived in Appendix A. Type-R and Type-I depend exclusively on the active modes and are phase independent, whereas Type-P and Type-C also depend on the phase values.

We denote the rotational reduction factor by fRf_R. The other three factors are denoted as fIf_I, fPf_P, and fCf_C. They are two when the corresponding redundancy condition is satisfied and one otherwise. These contributions are not independent and combining operations from any two of the types P, I, and C produces an operation of the third type which is dependent on the other two. Therefore, when all three types are present, multiplying their factors counts one reduction twice. To correct this repeated counting, we introduce the overlap factor gg, defined as

g={2,fP=fI=fC=2,1,otherwise.(26)g = \begin{cases} 2, & f_P=f_I=f_C=2,\\ 1, & \text{otherwise}. \end{cases} \tag*{(26)}

The CP multiplicity is then

mCP=fRfPfIfCg,(27)m_{\mathrm{CP}}=\frac{f_R f_P f_I f_C}{g}, \tag*{(27)}

and the CP orbit becomes

∣O({ϕm})∣=g4NfRfPfIfC.(28)\lvert\mathcal{O}(\{\phi_m\})\rvert=g\frac{4N}{f_R f_P f_I f_C}. \tag*{(28)}

(28) applies to the full set of formal operations of a homogeneous ring. For rings containing heteroatoms, only operations that preserve the heteroatom pattern are allowed. The number of distinct CP images is then obtained by dividing the number of allowed operations by the number of those operations that reproduce the reference CP pattern.

CP multiplicities

There is a connection between CP redundancy and puckering symmetry. More generally, the operations that leave an elevation pattern unchanged determine its symmetry group, while their number gives the CP multiplicity and hence the orbit size. Table V summarizes this correspondence for representative homogeneous ring patterns. The point-group labels describe the elevation pattern on the regular reference polygon; they also apply to an ideal reconstructed ring when its in-plane geometry is compatible with these operations, but need not describe an arbitrary Cartesian molecular geometry.

PatternGroupfRf_RfCf_CfIf_IfPf_PggmCPm_{\mathrm{CP}}∣O∣\lvert\mathcal{O}\rvert
Five-membered, genericC1C_111111120
Five-membered, envelopeCsC_s11121210
Five-membered, twistC2C_212111210
Six-membered, chairD3dD_{3d}32222122
Six-membered, boatC2vC_{2v}2112146
Six-membered, skewD2D_22211146
Six-membered, half-chairC2C_212111212
Six-membered, half-boatCsC_s11121212
CNC_N puckered, S2nS_{2n}S2nS_{2n}nn12112n2n2N/n2N/n
CNC_N puckered, DnD_nDnD_nnn21112n2n2N/n2N/n
CNC_N puckered, DndD_{nd}DndD_{nd}nn22224n4nN/nN/n
NN-membered, planarDNhD_{Nh}NN22224N4N1

Table V. Representative homogeneous ring patterns, their symmetry groups, CP reduction factors, multiplicities, and orbit sizes. The symmetry labels refer to the elevation patterns on the regular reference polygon.

For example, cyclopentane has a single phase-bearing mode, m=2m=2, with neither Type-R nor Type-I redundancy. A generic phase therefore produces twenty distinct CP images. At an envelope phase, such as ϕ2=0∘\phi_2=0^\circ, Type-P reduces the orbit through a mirror-type operation (CsC_s symmetry). At a twist phase, such as ϕ2=18∘\phi_2=18^\circ, Type-C gives the same reduction through an in-plane twofold rotation (C2C_2 symmetry). In both cases, g=1g=1, mCP=2m_{\mathrm{CP}}=2, and ∣O∣=10\lvert\mathcal{O}\rvert=10.

Figure 4 illustrates these two orbits. The envelope images occur at 36∘36^\circ intervals starting from 0∘0^\circ, and the twist images at the same intervals starting from 18∘18^\circ. Thus, the two patterns have equal CP multiplicities and orbit sizes, but different symmetries.

Envelope and twist CP phase orbits of a homogeneous five-membered ring at fixed puckering amplitude

Figure 4. Envelope and twist CP phase orbits of a homogeneous five-membered ring at fixed puckering amplitude. Each orbit contains ten distinct phase points generated by twenty formal operations. Paired structures indicate operations producing the same CP coordinates. Red, green, and yellow denote ring elevations 1, 2, and 3, respectively.

Symmetry-Reduced Search Domains for CP Phases

The next step is to construct symmetry-reduced CP phase domains containing at least one representative of each orbit of symmetry-equivalent puckering patterns. For a single active phase, the intervals derived below are fundamental domains. For multiple active phases, we use covering domains that remove generic symmetry redundancy but may retain equivalent points on their boundaries.

For r≥1r\geq1, we take m1m_1 as the leading mode. The leading-phase interval is determined from the symmetry images of that isolated mode, allowing the remaining phases and, when present, the special-coordinate sign to change under the same operation.

The secondary phases can then be restricted sequentially. At each step, only symmetry operations that leave all previously restricted phases unchanged are used to reduce the next phase. Their intervals therefore depend on the ordered partial sets of active modes and cannot be chosen independently. When the special coordinate is active, its sign must also be included in the reduction. In general, imposing a positive sign from the outset generally requires a different phase domain.

Homogeneous rings

Reduced interval with one active mode

We first consider a single active phase-bearing mode, with all other amplitudes, including the special coordinate when present, set to zero. The symmetry reduction therefore depends only on the images of the phase ϕm1\phi_{m_1}.

Cyclic rotations and global inversion generate phase equivalences that hold throughout the phase interval, rather than only at particular geometries. Their associated redundancy factors are denoted by fR(1)f_{\mathrm{R}}^{(1)} and the global inversion, fI(1)f_{\mathrm{I}}^{(1)}, respectively. The superscript refers here to the first and only active mode. Reflection equivalences are considered separately below. For a homogeneous ring, the rotational redundancy factor is

fR(1)=gcd⁡(N,m1).(29)f_{\mathrm{R}}^{(1)} = \operatorname{gcd}(N,m_1). \tag*{(29)}

This factor gives the number of cyclic rotations that reproduce the original phase, including the identity. Since there are NN possible cyclic rotations and each distinct phase is generated fR(1)f_{\mathrm{R}}^{(1)} times, the rotational phase spacing satisfies

NΔR=2πfR(1)(30)N\Delta_{\mathrm{R}} = 2\pi f_{\mathrm{R}}^{(1)} \tag*{(30)}

and therefore

ΔR=2πfR(1)N.(31)\Delta_{\mathrm{R}} = \frac{2\pi f_{\mathrm{R}}^{(1)}}{N}. \tag*{(31)}

Global inversion adds a phase shift of π\pi. We set fI(1)=2f_{\mathrm{I}}^{(1)} = 2 if this shift is already generated by cyclic rotation of the isolated leading mode, and fI(1)=1f_{\mathrm{I}}^{(1)} = 1 otherwise. Thus,

fI(1)={2,N/fR(1) even,1,N/fR(1) odd.(32)f_{\mathrm{I}}^{(1)} = \begin{cases} 2, & N/f_{\mathrm{R}}^{(1)}\ \text{even},\\ 1, & N/f_{\mathrm{R}}^{(1)}\ \text{odd}. \end{cases} \tag*{(32)}

Including global inversion doubles the number of available operations from NN to $2Nsincecyclicrotationsaloneaddssince cyclic rotations alone addsN and cyclic rotations combined with inversion adds another NN. The combined redundancy factor is

fRI(1)=fR(1)fI(1)=gcd⁡(2m1,N)(33)f_{\mathrm{RI}}^{(1)}=f_{\mathrm{R}}^{(1)}f_{\mathrm{I}}^{(1)}=\gcd(2m_1,N) \tag*{(33)}

The phase spacing after including rotations and global inversion satisfies

2NΔRI=2πfR(1)fI(1)(34)2N\Delta_{\mathrm{RI}}=2\pi f_{\mathrm{R}}^{(1)}f_{\mathrm{I}}^{(1)} \tag*{(34)}

and therefore

ΔRI=πfRI(1)N.(35)\Delta_{\mathrm{RI}}=\frac{\pi f_{\mathrm{RI}}^{(1)}}{N}. \tag*{(35)}

Reflection equivalences fold this phase sector in half. The resulting interval for the leading phase is given by

Fm1=[0,πfRI(1)2N].(36)\mathcal{F}_{m_1}=\left[0,\frac{\pi f_{\mathrm{RI}}^{(1)}}{2N}\right]. \tag*{(36)}

A useful example is cyclopentane. Here N=5N=5 and the puckering is described by the single mode m=2m=2. Since NN is odd, the π\pi shift generated by the improper block cannot coincide with a rotational image. Thus Type-I does not hold and fI=1f_{\mathrm{I}}=1. Moreover, fR=1f_{\mathrm{R}}=1. The combined proper and improper blocks therefore give an effective period of 36∘36^\circ. This is not yet the fundamental domain, because reflection symmetry folds this sector in half:

F2=[0,π10]=[0∘,18∘].(37)\mathcal{F}_2=\left[0,\frac{\pi}{10}\right]=[0^\circ,18^\circ]. \tag*{(37)}

The two envelope forms at 0∘0^\circ and 36∘36^\circ are related by the symmetry of the homogeneous ring, while the twist form at 18∘18^\circ lies at the midpoint of the sector. Therefore the interval [0∘,18∘][0^\circ,18^\circ] already contains one representative of the single-mode symmetry-inequivalent conformations of cyclopentane.

As a second example, consider cyclohexane with active mode m=2m=2. In this case fR=2f_{\mathrm{R}}=2 and fI=1f_{\mathrm{I}}=1, giving

F2=[0,π6]=[0∘,30∘].(38)\mathcal{F}_2=\left[0,\frac{\pi}{6}\right]=[0^\circ,30^\circ]. \tag*{(38)}

The corresponding one-mode intervals for rings with N=5N=5–20 are listed in Table VI. They are also useful as the base for the step size in multi-mode search as described in § 7.

Reduced interval with one active phase and the special coordinate

We now consider an even-membered ring with one active phase-bearing mode and a nonzero special coordinate, with all remaining amplitudes set to zero.

If both special-coordinate signs are retained in the sampling, the phase interval derived above remains applicable, since its construction allows the sign to change under the symmetry operations. Alternatively, we may impose qN/2>0q_{N/2}>0. Any configuration with a negative special coordinate can first be mapped into this sector by global inversion. Subsequent phase

reduction must then use only operations that preserve the positive sign of the special coordinate. Under a cyclic displacement kk, this coordinate acquires the factor (−1)k(-1)^k. Rotations alone therefore preserve its sign only for even displacements, leaving N/2N/2 available operations. Their redundancy factor is

fR(1),+=gcd⁡(N2,m1)(39)f_{\mathrm{R}}^{(1),+}=\gcd\left(\frac{N}{2},m_1\right) \tag*{(39)}

Accordingly, the rotational phase spacing within the positive sector satisfies

N2ΔR(+)=2πfR(1),+.(40)\frac{N}{2}\Delta_{\mathrm{R}}^{(+)}=2\pi f_{\mathrm{R}}^{(1),+}. \tag*{(40)}

Including global inversion doubles the number of sign-preserving operations from N/2N/2 to NN: even displacements without inversion and odd displacements combined with inversion. Defining the additional redundancy factor by

fRI(1),+=fR(1),+fI(1),+,(41)f_{\mathrm{RI}}^{(1),+}=f_{\mathrm{R}}^{(1),+}f_{\mathrm{I}}^{(1),+}, \tag*{(41)}

the combined phase spacing satisfies

NΔRI(+)=2πfR(1),+fI(1),+.(42)N\Delta_{\mathrm{RI}}^{(+)}=2\pi f_{\mathrm{R}}^{(1),+}f_{\mathrm{I}}^{(1),+}. \tag*{(42)}

Dividing by NN gives

ΔRI(+)=2πfRI(1),+N.(43)\Delta_{\mathrm{RI}}^{(+)}=\frac{2\pi f_{\mathrm{RI}}^{(1),+}}{N}. \tag*{(43)}

The numerator contains 2π2\pi, rather than π\pi, because only NN operations preserve the special-coordinate sign, compared with 2N2N operations in the phase-only case. The combined redundancy factor is

fRI(1),+=gcd⁡(N,m1+N2).(44)f_{\mathrm{RI}}^{(1),+}=\gcd\left(N,m_1+\frac{N}{2}\right). \tag*{(44)}

The reference vertical reflection with axis index zero preserves the special-coordinate sign and folds the phase sector in half. The resulting reduced interval is

Fm1(+)=[0,πNfRI(1),+].(45)\mathcal{F}_{m_1}^{(+)}=\left[0,\frac{\pi}{N}f_{\mathrm{RI}}^{(1),+}\right]. \tag*{(45)}

This interval provides coverage with the special coordinate restricted to positive values. It may be wider than the phase-only interval because operations that change the special-coordinate sign are no longer available within the positive sector.

For cyclohexane, the ordinary mode has index two and the special coordinate has index three. With the special coordinate restricted to positive values,

gcd⁡(6,2+62)=gcd⁡(6,5)=1,(46)\gcd\left(6,2+\frac{6}{2}\right)=\gcd(6,5)=1, \tag*{(46)}

so that

F2(+)=[0,π6]=[0∘,30∘].(47)\mathcal{F}_{2}^{(+)}=\left[0,\frac{\pi}{6}\right]=[0^\circ,30^\circ]. \tag*{(47)}

Thus the phase interval remains unchanged from the case in which the special coordinate is zero.

For cyclooctane, consider only the ordinary coordinate of index two and the special coordinate of index four, with the mode of index three set to zero. Restricting the special coordinate to positive values gives

gcd⁡(8,2+82)=gcd⁡(8,6)=2,(48)\gcd\left(8,2+\frac{8}{2}\right)=\gcd(8,6)=2, \tag*{(48)}

and therefore

F2(+)=[0,π4]=[0∘,45∘].(49)\mathcal{F}_{2}^{(+)}=\left[0,\frac{\pi}{4}\right]=[0^\circ,45^\circ]. \tag*{(49)}

In both examples, including the special coordinate with its sign fixed to be positive leaves the reduced phase interval unchanged, although its magnitude introduces an additional sampling coordinate.

The methodology extends to the active-mode sets M\mathcal{M} and M+\mathcal{M}_{+} defined in Sec. II. The corresponding redundancy factors and phase intervals are summarized in Table VII.

Active modesRedundancy factorsFirst-phase intervalAdditional-phase intervals
r=1r=1fRI(1)=gcd⁡(N,2m1)f_{\mathrm{RI}}^{(1)}=\gcd(N,2m_1)ϕm1∈[0,πfRI(1)2N]\phi_{m_1}\in\left[0,\frac{\pi f_{\mathrm{RI}}^{(1)}}{2N}\right]—
r+=2r_+=2fRI(1),+=gcd⁡(N,m1+N2)f_{\mathrm{RI}}^{(1),+}=\gcd\left(N,m_1+\frac{N}{2}\right)ϕm1∈[0,πfRI(1),+N]\phi_{m_1}\in\left[0,\frac{\pi f_{\mathrm{RI}}^{(1),+}}{N}\right]—
r=2r=2fRI(2)=gcd⁡(N,2m1,m2−m1)f_{\mathrm{RI}}^{(2)}=\gcd(N,2m_1,m_2-m_1)ϕm1∈[0,πfRI(1)2N]\phi_{m_1}\in\left[0,\frac{\pi f_{\mathrm{RI}}^{(1)}}{2N}\right]ϕm2∈[0,2πfRI(2)fRI(1)]\phi_{m_2}\in\left[0,\frac{2\pi f_{\mathrm{RI}}^{(2)}}{f_{\mathrm{RI}}^{(1)}}\right]
r+=3r_+=3fRI(2),+=gcd⁡(N,m1+N2,m2+N2)f_{\mathrm{RI}}^{(2),+}=\gcd\left(N,m_1+\frac{N}{2},m_2+\frac{N}{2}\right)ϕm1∈[0,πfRI(1),+N]\phi_{m_1}\in\left[0,\frac{\pi f_{\mathrm{RI}}^{(1),+}}{N}\right]ϕm2∈[0,2πfRI(2),+fRI(1),+]\phi_{m_2}\in\left[0,\frac{2\pi f_{\mathrm{RI}}^{(2),+}}{f_{\mathrm{RI}}^{(1),+}}\right]
r≥1r\geq1fRI(i−1)=gcd⁡(N,2m1,{mk−m1}k=2i−1)f_{\mathrm{RI}}^{(i-1)}=\gcd\left(N,2m_1,\{m_k-m_1\}_{k=2}^{i-1}\right)<br>fRI(i)=gcd⁡(fRI(i−1),mi−m1)f_{\mathrm{RI}}^{(i)}=\gcd\left(f_{\mathrm{RI}}^{(i-1)},m_i-m_1\right)ϕm1∈[0,πfRI(1)2N]\phi_{m_1}\in\left[0,\frac{\pi f_{\mathrm{RI}}^{(1)}}{2N}\right]ϕmi∈[0,2πfRI(i)fRI(i−1)]\phi_{m_i}\in\left[0,\frac{2\pi f_{\mathrm{RI}}^{(i)}}{f_{\mathrm{RI}}^{(i-1)}}\right], i=2,…,ri=2,\ldots,r
r+=r+1,r≥1r_+=r+1,\quad r\geq1fRI(i−1),+=gcd⁡(N,{mk+N2}k=1i−1)f_{\mathrm{RI}}^{(i-1),+}=\gcd\left(N,\left\{m_k+\frac{N}{2}\right\}_{k=1}^{i-1}\right)<br>fRI(i),+=gcd⁡(fRI(i−1),+,mi+N2)f_{\mathrm{RI}}^{(i),+}=\gcd\left(f_{\mathrm{RI}}^{(i-1),+},m_i+\frac{N}{2}\right)ϕm1∈[0,πfRI(1),+N]\phi_{m_1}\in\left[0,\frac{\pi f_{\mathrm{RI}}^{(1),+}}{N}\right]ϕmi∈[0,2πfRI(i),+fRI(i−1),+]\phi_{m_i}\in\left[0,\frac{2\pi f_{\mathrm{RI}}^{(i),+}}{f_{\mathrm{RI}}^{(i-1),+}}\right], i=2,…,ri=2,\ldots,r

Table VII. Redundancy factors and reduced phase intervals for homogeneous rings. Rows labeled by rr correspond to an inactive special mode. Rows labeled by r+r_+ include an active special mode, require even NN, and refer to the sector qN/2>0q_{N/2}>0.

There are additional redundancies at the boundaries, for instance, when the first phase is fixed at ϕm1=0\phi_{m_1}=0, the residual reflection symmetry allows the second-phase interval listed in Table VII to be halved, with the upper endpoint of the reduced interval included. The intervals of all subsequent phases remain unchanged. This reduction does not modify the tabulated redundancy factors. Additional reductions can be applied on the boundaries of the phase domain. However, their implementation is cumbersome and lies beyond the scope of the present discussion.

Rings with heteroatoms

The phase-reduction argument used for homogeneous rings also applies to rings containing heteroatoms, provided that only operations preserving the atom-type pattern are counted. In particular, NN must be replaced by the number KK of allowed cyclic rotations, and the redundancy factors must be evaluated within this restricted set.

If KK cyclic rotations are allowed, including global inversion gives 2K2K rotation–inversion operations. Of these, fRI(1)f_{\mathrm{RI}}^{(1)} leave the active phase unchanged modulo 2π2\pi. The orbit therefore contains 2K/fRI(1)2K/f_{\mathrm{RI}}^{(1)} distinct phase images, and their spacing satisfies

2KΔRI=2πfRI(1).(50)2K\Delta_{\mathrm{RI}}=2\pi f_{\mathrm{RI}}^{(1)}. \tag*{(50)}

Hence,

ΔRI=πfRI(1)K.(51)\Delta_{\mathrm{RI}}=\frac{\pi f_{\mathrm{RI}}^{(1)}}{K}. \tag*{(51)}

An allowed reflection folds this interval in half. Up to an appropriate choice of phase origin, the reduced interval is

Fm=[0,πfRI(1)2K].(52)\mathcal{F}_{m}=\left[0,\frac{\pi f_{\mathrm{RI}}^{(1)}}{2K}\right]. \tag*{(52)}

If the special mode is also active, we first use global inversion to choose the sector qN/2>0q_{N/2}>0. Since a cyclic displacement kk changes the special coordinate by the factor (−1)k(-1)^k, each allowed rotation has exactly one sign-preserving version: without inversion for even kk, and combined with inversion for odd kk. There are therefore KK, rather than 2K2K, available rotation–inversion operations. Writing their redundancy factor as fRI(1),+f_{\mathrm{RI}}^{(1),+}, the spacing satisfies

KΔRI(+)=2πfRI(1),+.(53)K\Delta_{\mathrm{RI}}^{(+)}=2\pi f_{\mathrm{RI}}^{(1),+}. \tag*{(53)}

A reflection preserving the positive sector, either alone or combined with inversion, then gives

Fm(+)=[0,πfRI(1),+K].(54)\mathcal{F}_{m}^{(+)}=\left[0,\frac{\pi f_{\mathrm{RI}}^{(1),+}}{K}\right]. \tag*{(54)}

a. One heteroatom or two identical non-antipodal heteroatoms. In these cases, no nontrivial cyclic rotation preserves the atom-type pattern, so K=1K=1. Rotations alone therefore give fR(1)=1f_{\mathrm{R}}^{(1)}=1. Without an active special mode, inversion generates the additional phase image ϕm+π\phi_m+\pi. Since this image differs from the original phase, inversion does not increase the redundancy, i.e., fI(1)=1f_{\mathrm{I}}^{(1)}=1. The two available operations thus generate two distinct phase images, giving ΔRI=π\Delta_{\mathrm{RI}}=\pi. A reflection preserving the atom-type pattern folds this interval in half, yielding

Fm=[0,π2].(55)\mathcal{F}_{m}=\left[0,\frac{\pi}{2}\right]. \tag*{(55)}

This is consistent with the residual CP symmetry of C2vC_{2v} type (see Table III).

With qN/2>0q_{N/2}>0, inversion alone is no longer available within the chosen sector. The only sign-preserving rotation–inversion operation is the identity, so fRI(1),+=1f_{\mathrm{RI}}^{(1),+}=1, and ΔRI(+)=2π\Delta_{\mathrm{RI}}^{(+)}=2\pi.

In this case, the reduced interval becomes

Gm(+)=[0,π].(56)\mathcal{G}_m^{(+)}=[0,\pi]. \tag*{(56)}

b. Two identical antipodal heteroatoms. Antipodal positions are possible only for even NN. There are now K=2K=2 allowed cyclic rotations, i.e., the identity and the half-ring rotation, which exchanges the identical heteroatoms. The rotational redundancy factor is consequently fR(1)=gcd⁡(2,m)f_{\mathrm{R}}^{(1)}=\gcd(2,m).

Without an active special mode, the homogeneous-ring expression for the combined redundancy applies with two allowed rotations, and fRI(1)=gcd⁡(2,2m)=2f_{\mathrm{RI}}^{(1)}=\gcd(2,2m)=2. The four available rotation–inversion operations (in this case K=2K=2) generate only two distinct phase images:

4ΔRI=2πfRI(1)=4π.(57)4\Delta_{\mathrm{RI}}=2\pi f_{\mathrm{RI}}^{(1)}=4\pi. \tag*{(57)}

Thus ΔRI=π\Delta_{\mathrm{RI}}=\pi, and reflection gives

Gm=[0,π2].(58)\mathcal{G}_m=\left[0,\frac{\pi}{2}\right]. \tag*{(58)}

Although the residual CP symmetry is enlarged to D2hD_{2h} type, its additional operations do not provide further reduction of an isolated active phase.

With qN/2>0q_{N/2}>0, only two rotation–inversion operations preserve the chosen sector, and in analogy with the case of one active mode plus the qN/2q_{N/2} mode, the redundancy factor is therefore

fRI(1),+=gcd⁡(2,m+N2).(59)f_{\mathrm{RI}}^{(1),+}=\gcd\left(2,m+\frac{N}{2}\right). \tag*{(59)}

As in the homogeneous positive-sector reduction, this factor counts the available operations that reproduce the original phase. Here the available set contains only two operations, so

2ΔRI(+)=2πfRI(1),+.(60)2\Delta_{\mathrm{RI}}^{(+)}=2\pi f_{\mathrm{RI}}^{(1),+}. \tag*{(60)}

Applying a reflection compatible with the positive sector gives

Gm(+)=[0,π2gcd⁡(2,m+N2)].(61)\mathcal{G}_m^{(+)}=\left[0,\frac{\pi}{2}\gcd\left(2,m+\frac{N}{2}\right)\right]. \tag*{(61)}

Conformational Sampling and Applications

Sampling strategy

The symmetry-reduced covering domains defined in § 4 provide a convenient basis for a systematic conformational search of homogeneous rings. The practical grid includes the lower-boundary reflection reduction described above, but retains residual symmetry redundancy on other boundary sets.

We sample the active phase-bearing modes in M\mathcal{M}. When the special mode is active, we use M+\mathcal{M}_{+} and sample qN/2q_{N/2} separately in the positive sector. For each of these modes, we choose the phase step as

Δϕmi=sup⁡Fmii=1,…,r,(62)\Delta\phi_{m_i}=\sup\mathcal{F}_{m_i}\qquad i=1,\ldots,r, \tag*{(62)}

Here Fmi\mathcal{F}_{m_i} is the isolated-mode fundamental interval listed in Table 6, not the sequential interval of the multimode domain. Thus, when two modes are active, the phase steps are (Δϕm1,Δϕm2)(\Delta\phi_{m_1},\Delta\phi_{m_2}). An initial preconditioned search may include configurations with one, two, or three active phase-bearing modes, i.e., {m1}\{m_1\}, {m1,m2}\{m_1,m_2\}, and {m1,m2,m3}\{m_1,m_2,m_3\}, respectively.

NNmmfRf_{\mathrm{R}}fIf_{\mathrm{I}}Fm\mathcal{F}_mNNmmfRf_{\mathrm{R}}fIf_{\mathrm{I}}Fm\mathcal{F}_m
5any11[0∘,18∘][0^\circ,18^\circ]152,4,711[0∘,6∘][0^\circ,6^\circ]
6221[0∘,30∘][0^\circ,30^\circ]153,631[0∘,18∘][0^\circ,18^\circ]
7any11[0∘,12.857∘][0^\circ,12.857^\circ]15551[0∘,30∘][0^\circ,30^\circ]
8222[0∘,45∘][0^\circ,45^\circ]162,622[0∘,22.5∘][0^\circ,22.5^\circ]
8312[0∘,22.5∘][0^\circ,22.5^\circ]163,5,712[0∘,11.25∘][0^\circ,11.25^\circ]
92,411[0∘,10∘][0^\circ,10^\circ]16442[0∘,45∘][0^\circ,45^\circ]
9331[0∘,30∘][0^\circ,30^\circ]17any11[0∘,5.294∘][0^\circ,5.294^\circ]
102,411[0∘,18∘][0^\circ,18^\circ]182,4,821[0∘,10∘][0^\circ,10^\circ]
10312[0∘,18∘][0^\circ,18^\circ]18332[0∘,30∘][0^\circ,30^\circ]
11any11[0∘,8.182∘][0^\circ,8.182^\circ]185,712[0∘,10∘][0^\circ,10^\circ]
12222[0∘,30∘][0^\circ,30^\circ]18661[0∘,30∘][0^\circ,30^\circ]
12332[0∘,45∘][0^\circ,45^\circ]19any11[0∘,4.737∘][0^\circ,4.737^\circ]
12441[0∘,30∘][0^\circ,30^\circ]202,622[0∘,18∘][0^\circ,18^\circ]
12512[0∘,15∘][0^\circ,15^\circ]203,7,912[0∘,9∘][0^\circ,9^\circ]
13any11[0∘,6.923∘][0^\circ,6.923^\circ]204,841[0∘,18∘][0^\circ,18^\circ]
142,4,621[0∘,12.857∘][0^\circ,12.857^\circ]20552[0∘,45∘][0^\circ,45^\circ]
143,512[0∘,12.857∘][0^\circ,12.857^\circ]

Table 6. Fundamental phase intervals for homogeneous rings with one active CP phase, reported in degrees. Only modes with an associated independent phase variable are included. Modes with identical reduction factors and identical fundamental intervals are grouped in the same row. The intervals correspond to the fully reduced single-phase domains, including the final folding by reflection symmetry.

To determine the number of sampled configurations, let LiL_i denote the length of the sequential interval for ϕmi\phi_{m_i} in Table VII. When the special coordinate is active and restricted to positive values, the corresponding lengths are denoted by Li,+L_{i,+}. The first phase is sampled over a closed interval, whereas the remaining phases are sampled over half-open intervals:

ϕm1∈[0,L1],ϕmi∈[0,Li),i=2,…,r.(63)\phi_{m_1}\in[0,L_1],\qquad\phi_{m_i}\in[0,L_i),\qquad i=2,\ldots,r. \tag*{(63)}

When the first phase is zero, the residual reflection symmetry reduces the second-phase interval to ϕm2∈[0,L2/2]\phi_{m_2}\in[0,L_2/2], with the upper endpoint included. The intervals of subsequent phases remain unchanged. The chosen steps divide the corresponding interval lengths exactly, including the halved second-phase interval.

For two active modes, the number of sampled configurations is

Ssym(2)(m1,m2)=L1Δϕm1L2Δϕm2+(L22Δϕm2+1).(64)S_{\mathrm{sym}}^{(2)}(m_1,m_2)=\frac{L_1}{\Delta\phi_{m_1}}\frac{L_2}{\Delta\phi_{m_2}}+\left(\frac{L_2}{2\Delta\phi_{m_2}}+1\right). \tag*{(64)}

The first term counts configurations with a nonzero first phase, while the second counts those on the reflection-reduced boundary. For even NN and considering qN/2>0q_{N/2}>0, there are additional configurations which are given by

Ssym,+(2)(m1,N2)=L1,+Δϕm1+1(65)S_{\mathrm{sym},+}^{(2)}\left(m_1,\frac{N}{2}\right)=\frac{L_{1,+}}{\Delta\phi_{m_1}}+1 \tag*{(65)}

where L1,+L_{1,+} is the length of the interval for mode m1m_1 when qN/2q_{N/2} is active.

For searches involving three active ordinary modes, the second and third phases use twice their isolated-mode phase steps only when the doubled step divides the corresponding interval length exactly. Otherwise, the original step is retained. We define the effective step as

h(L,Δ)={2Δ,L/(2Δ)∈Z>0,Δ,otherwise.(66)h(L,\Delta)= \begin{cases} 2\Delta, & L/(2\Delta)\in\mathbb{Z}_{>0},\\ \Delta, & \text{otherwise}. \end{cases} \tag*{(66)}

The second phase is sampled over its full interval when the first phase is nonzero, but over an interval of half that length when the first phase is zero. We therefore choose its step separately in these two regions. The doubled step may divide the full interval exactly without dividing the halved interval exactly. We use h2h_2 for the full interval and h2,bh_{2,\mathrm{b}} for the reflection-reduced boundary interval, where b denotes the boundary. The third-phase step is denoted by h3h_3. Therefore,

h2=h(L2,Δϕm2),h2,b=h(L2/2,Δϕm2),h3=h(L3,Δϕm3).(67)\begin{aligned} h_2&=h(L_2,\Delta\phi_{m_2}),\\ h_{2,\mathrm{b}}&=h(L_2/2,\Delta\phi_{m_2}),\\ h_3&=h(L_3,\Delta\phi_{m_3}). \tag*{(67)} \end{aligned}

For example, if the full second-phase interval has a length of sixty degrees and the original step is ten degrees, we use a step of twenty degrees in that interval. On the boundary, the interval has a length of thirty degrees, so we retain the original ten-degree step.

The number of sampled configurations for three active ordinary modes is therefore

Ssym(3)(m1,m2,m3)=⌊L1Δϕm1⌋⌊L2h2⌋+(⌊L22h2,b⌋+1)⌊L3h3⌋.(68)S_{\mathrm{sym}}^{(3)}(m_1,m_2,m_3)=\left\lfloor\frac{L_1}{\Delta\phi_{m_1}}\right\rfloor\left\lfloor\frac{L_2}{h_2}\right\rfloor+\left(\left\lfloor\frac{L_2}{2h_{2,\mathrm{b}}}\right\rfloor+1\right)\left\lfloor\frac{L_3}{h_3}\right\rfloor. \tag*{(68)}

The first term counts points with a nonzero leading phase, and the second counts points on its zero-phase boundary. Floor and ceiling functions account for closed and half-open intervals, respectively, without adding endpoints outside the regular grid.

For even NN and considering qN/2>0q_{N/2}>0, the count with three modes should include

Ssym,+(3)(m1,m2,N2)=L1,+Δϕm1L2,+Δϕm2+(L2,+2Δϕm2+1).(69)S_{\mathrm{sym},+}^{(3)}\left(m_1,m_2,\frac{N}{2}\right)=\frac{L_{1,+}}{\Delta\phi_{m_1}}\frac{L_{2,+}}{\Delta\phi_{m_2}}+\left(\frac{L_{2,+}}{2\Delta\phi_{m_2}}+1\right). \tag*{(69)}

There are M−1M-1 available ordinary modes, since their indices run from 2 to MM. A combination of rr active ordinary modes can therefore be selected in

C(r)=(M−1r)(70)C^{(r)}=\binom{M-1}{r} \tag*{(70)}

ways. For even-membered rings, the total number of combinations with r+r_{+} active amplitudes is obtained by adding combinations of r+r_{+} ordinary amplitudes and combinations of r=r+−1r=r_{+}-1 ordinary amplitudes plus the special coordinate:

Ctotal(r+)=(M−1r+)+(M−1r)=(Mr+).(71)C_{\mathrm{total}}^{(r_{+})}=\binom{M-1}{r_{+}}+\binom{M-1}{r}=\binom{M}{r_{+}}. \tag*{(71)}

In addition to the phases, initial mode amplitudes must be specified for conformational sampling. The total puckering amplitude Q≥0Q\geq0 is defined by

Q2={∑m=2Mqm2,N odd,∑m=2Mqm2+qN/22,N even.(72)Q^2= \begin{cases} \sum_{m=2}^{M}q_m^2, & N\ \text{odd},\\ \sum_{m=2}^{M}q_m^2+q_{N/2}^2, & N\ \text{even}. \end{cases} \tag*{(72)}

An effective strategy is to obtain QQ from an optimized reference geometry and distribute its squared magnitude equally among the active modes. For a set M\mathcal{M} of r=∣M∣r = |\mathcal{M}| active ordinary modes, with the special coordinate absent or inactive, the initial mode amplitudes are therefore

qm=Qr,m∈M,(73)q_m = \frac{Q}{\sqrt{r}}, \qquad m \in\mathcal{M}, \tag*{(73)}

which preserves the total puckering amplitude. When the special mode is also active, r+=∣M+∣=r+1r_{+} = |\mathcal{M}_{+}| = r + 1, and the equal-amplitude initialization in the positive sector is then

qm=Qr+,m∈M+.(74)q_m = \frac{Q}{\sqrt{r_{+}}}, \qquad m \in\mathcal{M}_{+}. \tag*{(74)}

As new conformations become available, the initial guess for each subsequent structure should be constructed from the closest available geometry, following the procedure described in Ref. 16.

Local amplitude balance nomenclature

To support conformational searches, we introduce the semidiscrete Local Amplitude Balance (LAB) nomenclature, which groups ring geometries according to the relative magnitudes of consecutive puckering amplitudes. By distinguishing absent, equal-amplitude, dominant, and subordinate modes, this scheme provides a compact framework for organizing sampled structures and exploring phase variations within a given amplitude pattern.

Unlike cumulative hyperspherical descriptions, each comparison involves only two neighboring modes, so changing one amplitude affects at most two symbols in the classification. For six- and seven-membered rings, there are only two puckering amplitudes and hence a single amplitude-balance angle. In these cases, the local and hyperspherical descriptions coincide when the same amplitude ordering and angular convention are used. For six-membered rings, the magnitude of the special coordinate enters this comparison, while its sign is recorded separately. The distinction between local and cumulative amplitude comparisons arises only for rings with three or more puckering amplitudes.

The resulting LAB labels provide a qualitative classification of amplitude patterns, facilitating comparisons between conformational families while retaining the full CP coordinates for their quantitative description.

We describe local amplitude balance through the angles

ψm−2=atan2⁡(qm−1,qm),m=3,…,M.(75)\psi_{m-2} = \operatorname{atan2}(q_{m-1},q_m), \qquad m = 3,\ldots,M. \tag*{(75)}

For even-membered rings, an additional angle compares the last ordinary amplitude with the magnitude of the special coordinate:

ψM−1=atan2⁡(qM,∣qN/2∣).(76)\psi_{M-1} = \operatorname{atan2}(q_M,|q_{N/2}|). \tag*{(76)}

Thus, all local angles lie between 0 and 90 degrees. The special coordinate is taken in absolute value for angle calculation and symbol assignment, while its sign is recorded separately in an external subscript. If both amplitudes vanish, the angle is undefined and the pair is assigned the symbol 0.

Each consecutive pair is assigned one symbol according to Table VIII. For LAB assignments of optimized geometries, amplitudes with magnitudes below 0.001 Å are treated as zero. The phases associated with these amplitudes are omitted from the labels.

SymbolConditionMeaning
Zqm−1=0q_{m-1}=0, qm>0q_m>0zenith (pole)
d0<qm−1<qm0<q_{m-1}<q_mcurrent mode dominant
Mqm−1=qm>0q_{m-1}=q_m>0equal magnitudes
sqm−1>qm>0q_{m-1}>q_m>0current mode subordinate
Eqm−1>0q_{m-1}>0, qm=0q_m=0equator
0qm−1=qm=0q_{m-1}=q_m=0both modes absent

Table VIII. Symbols for the LAB nomenclature, based on the relative magnitudes of consecutive puckering amplitudes. For the terminal pair in even-membered rings, the same conditions apply to (qM,∣qN/2∣)(q_M,|q_{N/2}|).

Z, M, and E correspond to local angles of 0, 45, and 90 degrees, respectively, and d and s denote open sectors. Dominance (d) and subordination (s) refer only to the current mode relative to its immediate predecessor.

The conformational label consists of a symbol string enclosed in square brackets, followed by the ring size and the phases of the active ordinary modes:

[Λ](σ)(N;ϕm1,…,ϕmr),(77)[\Lambda]_{(\sigma)}(N;\phi_{m_1},\ldots,\phi_{m_r}), \tag*{(77)}

with

Λ={λ1⋯λM−2,N odd,λ1⋯λM−1,N even.(78)\Lambda= \begin{cases} \lambda_1\cdots\lambda_{M-2}, & N\ \text{odd},\\ \lambda_1\cdots\lambda_{M-1}, & N\ \text{even}. \end{cases} \tag*{(78)}

The symbols correspond to consecutive ordinary-mode pairs,

λm−2⟷(qm−1,qm),m=3,…,M,(79)\lambda_{m-2} \longleftrightarrow(q_{m-1},q_m), \qquad m = 3,\ldots,M, \tag*{(79)}

with the additional terminal symbol

λM−1⟷(qM,∣qN/2∣)(80)\lambda_{M-1} \longleftrightarrow(q_M,|q_{N/2}|) \tag*{(80)}

for even-membered rings. When the special coordinate is active, its sign is indicated by the external subscript:

σ={+,qN/2>0,−,qN/2<0.(81)\sigma= \begin{cases} +, & q_{N/2}>0,\\ -, & q_{N/2}<0. \end{cases} \tag*{(81)}

When the special coordinate vanishes, the terminal symbol is E or 0 and the subscript is omitted. When it is active, the external subscript indicates its sign. The two poles correspond to opposite signs of the special coordinate, with all ordinary amplitudes equal to zero. No sign subscript is used for odd-membered rings.

Only phases of active ordinary modes, i.e., those with nonzero amplitudes, are included in the label. The retained phases are listed in increasing mode-index order, with their indices inferred from the amplitude pattern. The special coordinate has no associated phase. Empty symbol strings are omitted together with their brackets, and empty phase lists are omitted together with the separating semicolon.

The bracketed symbol string alone may be used as an abbreviated label for an amplitude-pattern family, retaining the sign subscript when the special coordinate is active. The ring size and phase list provide the additional information needed to distinguish its members at the LAB level. The label describes qualitative amplitude balance but does not encode the total puckering amplitude or replace the complete CP coordinates. Examples are given in Appendix B.

For medium-sized and large homogeneous rings, Dale employed a geometric nomenclature based on the arrangement of ring segments between corners.[17] For example, the square and rectangular conformations of cyclohexadecane are designated [4444] and [3535], respectively. The numerical labels of Dale describe the geometric organization of the ring, whereas LAB symbols encode local relationships between consecutive CP amplitudes. The two descriptions are therefore complementary rather than interchangeable. LAB labels provide a systematic classification of amplitude patterns, including those of low-symmetry and mixed-mode conformations that do not fit a simple geometric designation, and the same scheme applies across different ring sizes.

Application to cyclohexadecane

The conformational search strategy outlined in § 7 A was applied to cyclohexadecane (hereafter, C16C_{16}), a sixteen-membered cycloalkane that provides a useful test of the present framework. Its conformational preferences have been investigated through strain-energy calculations,[17] molecular mechanics calculations,[18] NMR spectroscopy,[19] and vibrational spectroscopy,[20] providing a reference for both the lowest-energy structure and the diversity of conformations in this flexible ring.

For the conformational search of cyclohexadecane, we used the phase step sizes specified in Table VI and calculated the numbers of sampled CP configurations from Eqs. (64)–(69). Symmetry reduction, combined with selective phase-grid coarsening for three ordinary modes, decreases these numbers from 7696 to 297 for two active amplitudes and from 221696 to 2023 for three active amplitudes, including combinations involving the special coordinate. Phase-grid coarsening provides an additional reduction of approximately a factor of three beyond symmetry alone in the combined sampling count. These counts refer to sampled CP points rather than to Cartesian starting geometries because some points may fail to yield a valid reconstruction, whereas others may produce several distinct geometries. Sampling counts and total reduction factors for homogeneous rings containing six to twenty atoms, together with a discussion of the factors governing the savings, are provided in Appendix C.

Cartesian geometries were generated using the ring reconstruction algorithm developed in our previous work.[16] CP coordinates specify the out-of-plane elevations but do not uniquely determine the in-plane geometry. Zou and co-workers[21] addressed this limitation by introducing additional coordinates to describe in-plane deformations. Our algorithm instead determines the projected geometry from the target elevations and reference bond lengths by enforcing ring closure and minimizing bond-angle distortions. The three-dimensional structure is recovered by incorporating the target elevations, which remain fixed along with the bond lengths during the angular adjustment. For each sampled CP point, we explored different arrangements of convex and concave projected bond angles, allowing between zero and seven concave angles (projected interior angles greater than 180∘180^\circ). All successfully reconstructed geometries passing the geometric checks were retained, and redundant structures were removed before electronic-structure optimization.

The resulting candidates were optimized at the HF/3-21G level.[22] Converged structures within 3.50 kcal/mol of the lowest-energy structure found at this level were reoptimized using the PBE0 functional[23] with the def2-TZVP basis set.[24] Single-point energies were subsequently calculated at the coupled-cluster (CC) DLPNO-CCSD(T) level[25] with the cc-pVTZ basis set.[26] All HF and DFT calculations were performed with Gaussian16[27] and all single-point CC calculations with ORCA[28]. All this information is available on Zenodo[29].

Although the starting geometries were generated from CP points with only two or three active amplitudes, the subsequent electronic-structure optimizations were not restricted to these modes. The search yielded 780 conformers at the HF/3-21G level. Table IX lists the ten lowest-energy conformers after DFT reoptimization and vibrational frequency calculations, ranked by their DLPNO-CCSD(T) energies including zero-point energy (ZPE) corrections at the DFT level. The four lowest-energy structures are shown in Figure 5 as projections onto their CP mean planes.

LAB labelPGΔ(EDFT)\Delta(E_{\mathrm{DFT}})Δ(ECC)\Delta(E_{\mathrm{CC}})
[E00ZE0](16; 0, 180)D2dD_{2d}0.000.00
[sssdss](+)_{(+)}(16; 4, 24, 224, 70, 190, 155)C1C_12.052.03
[ssdsss](+)_{(+)}(16; 19, 81, 154, 10, 66, 308)C1C_11.752.03
[dsddss](+)_{(+)}(16; 22.5, 79, 315, 191, 67.5, 124)C2C_21.902.07
[dssdsE](16; 0, 45, 270, 135, 180, 225)C2C_22.242.59
[sddssd](+)_{(+)}(16; 20, 162, 318, 280, 68, 125)C1C_12.512.62
[EZEZEZ](+)_{(+)}(16; 22.5, 315, 67.5)D2D_21.862.68
[ssdsss](+)_{(+)}(16; 14, 4, 211, 136, 294, 28)C1C_12.712.83
[ddsdsd](+)_{(+)}(16; 0, 90, 180, 90, 180, 90)CsC_s2.843.09
[ssddss](+)_{(+)}(16; 0, 0, 180, 180, 180, 180)CsC_s3.053.16

Table IX. The ten lowest-energy conformers of cyclohexadecane, ranked by their relative CC energies including zero-point energy corrections. Relative DFT and CC energies, both including ZPE, are reported in kcal/mol with respect to the lowest-energy conformer at the CC+ZPE level. LAB labels and molecular point groups (PG) refer to the optimized geometries. Phase values in the labels are expressed in degrees, with the degree symbol omitted. Generic phases are rounded to the nearest degree, but symmetry-constrained values are retained at their exact angular values.

The four lowest-energy conformers of C16

Figure 5. The four lowest-energy conformers of C16_{16}, shown as projections onto their CP mean planes with elevations obtained at the DFT level. Top left: [E00ZE0]; top right: [sssdss](+)_{(+)}; bottom left: [ssdsss](+)_{(+)}; bottom right: [dsddss](+)_{(+)}.

The lowest-energy conformer found in our search is the square structure, labeled [E00ZE0](16;0,180)(16;0,180), with molecular point-group symmetry D2dD_{2d}. This assignment agrees with the early strain-energy calculations of Dale,[17] which favored the square [4444] conformation over the rectangular [3535] form. Anet and Cheng[19] subsequently supported the predominance of [4444] at low temperature through carbon-13 NMR spectroscopy, and the three resonances observed in a 1:2:1 intensity ratio were consistent with its carbon environments. Their iterative strain-energy calculations placed [3535] 1.9 kcal/mol above [4444], whereas later MM2 calculations by Allinger and co-workers[18] gave a separation of approximately 3.4 kcal/mol.

In our calculations, the rectangular [3535] conformer, labeled [EZEZEZ](+)(16;22.5,315,67.5)_{(+)}(16;22.5,315,67.5), lies 2.68 kcal/mol above the [4444] minimum. However, five other conformers occur between these two structures in the CC+ZPE ranking. The two lowest of these have C1C_{1} symmetry and both lie 2.03 kcal/mol above the minimum at the precision reported in Table IX. The search recovers the square structure as the lowest-energy form and identifies several competing minima beyond the square and rectangular conformations. This is qualitatively consistent with the study of Shannon and co-workers,[20] who associated the ordered low-temperature crystal with [4] and suggested contributions from low-symmetry conformers in the liquid and high-temperature crystalline phases. The change in energy ordering between the DFT and CC results further illustrates the sensitivity of the relative conformational energies to the computational method.

The low-symmetry minima found here show that symmetry reduction in CP sampling does not restrict the search to symmetric conformations. Instead, it avoids sampling equivalent descriptions of the same ring elevation pattern, such as those obtained by changing the starting atom or reversing the numbering direction in a ring of equivalent atoms. This equivalence does not imply that the corresponding molecular geometry possesses any nontrivial point-group symmetry, and the subsequent electronic-structure optimizations impose no symmetry constraints.

The square conformer provides a concrete illustration. For its reference CP point, with q2=1.540q_2 = 1.540 Å, q6=1.641q_6 = 1.641 Å, a total puckering amplitude of Q2=2.250Q_2 = 2.250 Å and phases (ϕ2,ϕ6)=(0,180)(\phi_2,\phi_6)=(0,180), the 64 operations considered produce eight distinct orbit points, each obtained by eight operations. The operations yielding the same CP coordinates as the identity are EE, C168C_{16}^{8}, S164S_{16}^{4}, S1612S_{16}^{12}, C24C_2^{4}, C212C_2^{12}, σv\sigma_v, σv\sigma_v. These labels describe operations on the CP coordinates and ring indexing, rather than directly naming molecular point-group operations. For this particular geometry, the eight operations leaving the reference CP point unchanged were verified to correspond to the eight operations of the independently determined molecular point group D2dD_{2d}.

Representatives generating the eight distinct orbit points are EE, C16kC_{16}^{k} with k=1,…,7k=1,\ldots,7. They preserve the amplitudes and yield, respectively, the phase pairs (0,180)(0,180), (45,315)(45,315), (90,90)(90,90), (135,225)(135,225), (180,0)(180,0), (225,135)(225,135), (270,270)(270,270), (315,45)(315,45) with phase values expressed in degrees. Cyclic shifts separated by eight atom positions give identical CP coordinates, while numbering reversal and elevation-sign reversal add no further orbit points. Thus, the 64 results comprise eight distinct CP descriptions and 56 repetitions. All eight descriptions belong to the same equivalence class, not to different conformers, and only one orbit representative is needed in the symmetric CP sampling.

Cyclohexadecane with only modes two and six active illustrates the two-phase reduction in Table VII. The special coordinate is zero, so the row for two active ordinary modes applies. Taking mode two as the leading mode, the sequential redundancy factors are fRI(1)=gcd⁡(16,4)=4f_{\mathrm{RI}}^{(1)}=\gcd(16,4)=4, and fRI(2)=gcd⁡(16,4,6−2)=4f_{\mathrm{RI}}^{(2)}=\gcd(16,4,6-2)=4. Substitution into the tabulated expressions gives

ϕ2∈[0,π8]=[0∘,22.5∘],\phi_2 \in\left[0,\frac{\pi}{8}\right]=[0^\circ,22.5^\circ],
ϕ6∈[0,2π)=[0∘,360∘).(82)\phi_6 \in[0,2\pi)=[0^\circ,360^\circ). \tag*{(82)}

The leading phase therefore has the same reduced interval as the isolated mode listed in Table VI. The second phase, however, must retain its full range. On the lower boundary, where the leading phase is zero, the residual reflection maps the second phase to its negative, so the second interval reduces to ϕ6∈[0∘,180∘]\phi_6 \in[0^\circ,180^\circ] with both endpoints included.

The square minimum is not the only conformation with the LAB amplitude pattern [E00ZE0]. Two additional conformers with this pattern were located and are shown together with the minimum in Figure 6. The conformer labeled [E00ZE0](16; 0, 0) lies 10.49 kcal/mol above the square minimum at the CC+ZPE level. The other conformer, labeled [E00ZE0](16; 12, 61), has molecular S4S_4 symmetry and lies 9.22 kcal/mol above the minimum.

Three cyclohexadecane conformers sharing the LAB amplitude pattern

Figure 6. Three cyclohexadecane conformers sharing the LAB amplitude pattern [E00ZE0], shown as projections onto their CP mean planes. From left to right: [E00ZE0](16; 0, 180), [E00ZE0](16; 0, 0), and [E00ZE0](16; 12, 61). Only modes two and six are active, but the distinct phase combinations produce different elevation patterns and projected ring shapes.

These structures illustrate the conformational complexity of large homogeneous rings. Although their puckering involves the same two active modes, their projected ring shapes differ substantially. This diversity emphasizes the importance of exploring different arrangements of convex and concave projected bond angles during reconstruction, rather than restricting the search to convex ring projections, where deformation coordinates introduced by Zou et al. can play a role.[21]

CP coordinates describe the out-of-plane elevations, but do not uniquely determine the in-plane arrangement of the atoms. Consequently, even structures with similar CP coordinates can have markedly different Cartesian geometries. Symmetry makes this exploration more tractable by restricting the phase search to a reduced domain containing representatives of all symmetry-equivalent classes, thereby avoiding redundant sampling without excluding distinct CP patterns. A comprehensive conformational search therefore combines symmetry-reduced phase sampling with exploration of the different projected ring geometries compatible with each sampled CP point.

Conclusions

We have derived general analytical rules that generate all symmetry-equivalent Cremer–Pople representations of a ring puckering pattern. These rules act directly on the phases and, for even-membered rings, on the signed special coordinate. They allow equivalent representations to be identified before Cartesian reconstruction and geometry optimization. These rules connect CP multiplicity with puckering symmetry and provide exact single-phase intervals and reduced multiphase search domains, with residual boundary redundancy. The analysis also covers rings containing one heteroatom or

two identical heteroatoms. The semidiscrete Local Amplitude Balance (LAB) nomenclature complements this framework by organizing conformations according to their amplitude patterns and active phases.

Symmetry makes conformational searches more efficient by removing equivalent descriptions without excluding distinct puckering patterns. This reduction does not impose molecular symmetry, and the subsequent optimizations remain unrestricted. For cyclohexadecane, symmetry reduction alone decreases the total number of sampled CP points by approximately a factor of thirty. When combined with selective phase-grid coarsening, the reduction factor approaches one hundred. The search recovers the square conformer as the lowest-energy structure and identifies several competing low-symmetry minima between the square and rectangular forms.

The conformational diversity of large rings also requires exploration of the in-plane geometries compatible with each CP point. Because CP coordinates specify the elevations but do not uniquely determine the Cartesian structure, different arrangements of convex and concave projected bond angles must be considered during reconstruction. Combining this exploration with symmetry-reduced phase sampling provides a systematic approach to ring conformational searches while avoiding redundant computational effort.

Autor Declarations

Conflict of interest

The authors have no conflicts to disclose.

Author contributions

A.F.-R. and A.L.-S. designed the method. A.L.-S. implemented the algorithm and drafted the initial manuscript, A.F.-R. performed the electronic-structure calculations and prepared the final version of the manuscript, and both authors revised it.

Data availability

The data are available on Zenodo repository publicly accessible at https://doi.org/10.5281/zenodo.23985089

Classification of CP redundancies and additional examples

The formal operations of a homogeneous ring form four blocks, each containing NN operations:

(p,s)=(0,+1): CNk,(p,s)=(1,+1): C2k,ℓ,(p,s)=(0,−1): SNk,(p,s)=(1,−1): σvℓ.(83)\begin{aligned} (p,s)=(0,+1)&:\ C_N^k,\\ (p,s)=(1,+1)&:\ C_2^{k,\ell},\\ (p,s)=(0,-1)&:\ S_N^k,\\ (p,s)=(1,-1)&:\ \sigma_v^\ell. \tag*{(83)} \end{aligned}

Here C2⊥,ℓC_2^{\perp,\ell} denotes a twofold rotation about an axis in the reference plane, perpendicular to its normal. For even-membered rings, atom and bond axes can be distinguished by the conventional labels C2′C_2' and C2′′C_2''; the corresponding mirror planes are labeled σv\sigma_v and σd\sigma_d.

Ring-index shifts are understood modulo NN. Phase coincidences are understood modulo 2π2\pi, although the resulting conditions on the original phases are conveniently written modulo π\pi. All conditions must hold simultaneously for every active mode. All phases are modulo 2π2\pi and m∈Mm \in\mathcal{M}.

The symmetry interpretations below refer to the elevation pattern on the regular reference polygon. They need not be symmetries of a reconstructed molecule unless its in-plane geometry is also compatible.

Type-R: rotational redundancies

Type-R occurs within a single block when two cyclic shifts generate the same CP image. A cyclic displacement kk transforms the phase of each active ordinary mode according to

ϕm′=ϕm+2πmkN.(84)\phi'_m=\phi_m+\frac{2\pi m k}{N}. \tag*{(84)}

Thus, two displacements kk and k′k' produce a phase difference

Δϕm=2πmΔkN,Δk=k−k′.(85)\Delta\phi_m=\frac{2\pi m\Delta k}{N},\qquad\Delta k=k-k'. \tag*{(85)}

The resulting CP images coincide only if this difference is an integer number of complete turns for every active mode:

2πmΔkN=2πnm,nm∈Z.(86)\frac{2\pi m\Delta k}{N}=2\pi n_m,\qquad n_m\in\mathbb{Z}. \tag*{(86)}

Dividing both sides by 2π2\pi gives the equivalent condition

mΔkN∈Z.(87)\frac{m\Delta k}{N}\in\mathbb{Z}. \tag*{(87)}

For each active ordinary mode, this requires Δk\Delta k to be a multiple of N/gcd⁡(N,m)N/\gcd(N,m). When ordinary modes are active, their simultaneous preservation therefore gives the common displacement step

dR=lcm⁡(Ngcd⁡(N,m))=Ngcd⁡(N,M).(88)d_R=\operatorname{lcm}\left(\frac{N}{\gcd(N,m)}\right)=\frac{N}{\gcd(N,\mathcal{M})}. \tag*{(88)}

where lcm is the least common multiple. There are N/dRN/d_R allowed displacements modulo NN, including the identity.

If the special coordinate is active, its transformation qN/2′=(−1)ΔkqN/2q'_{N/2}=(-1)^{\Delta k}q_{N/2} also requires an even displacement. This is equivalent to imposing the same divisibility condition for m=N/2m=N/2. Hence, for a nonplanar pattern,

fR={gcd⁡(N,M+),N even and qN/2≠0,gcd⁡(N,M),otherwise.(89)f_R= \begin{cases} \gcd(N,\mathcal{M}_+), & N\text{ even and }q_{N/2}\ne0,\\ \gcd(N,\mathcal{M}), & \text{otherwise}. \end{cases} \tag*{(89)}

The gcd is taken over NN and all indices in the indicated set. If only the special coordinate is active, this gives fR=N/2f_R=N/2. For the planar pattern, every cyclic shift preserves the elevations and fR=Nf_R=N. Thus fRf_R counts the cyclic shifts preserving the complete elevation pattern, including the identity.

Type-C: reflected indexing with normal reversal

Type-C connects blocks with the same ss and opposite pp:

CNk↔C2k,ℓ,SNk↔σvℓ.(90)C_N^k\leftrightarrow C_2^{k,\ell},\qquad S_N^k\leftrightarrow\sigma_v^\ell. \tag*{(90)}

Equating the cyclic and reflected-index phases gives

ϕm+2πmkN=−ϕm−2πmℓN+π(91)\phi_m+\frac{2\pi m k}{N}=-\phi_m-\frac{2\pi m\ell}{N}+\pi \tag*{(91)}

With kC=−(k+ℓ)k_C=-(k+\ell), this becomes

ϕm=π2+πmkCN(modπ).(92)\phi_m=\frac{\pi}{2}+\frac{\pi m k_C}{N}\pmod{\pi}. \tag*{(92)}

The same integer kCk_C must satisfy this condition for every active ordinary mode; separate modewise solutions are insufficient.

Reflected indexing reverses the CP normal. Using the reference normal, the ordered elevations and special coordinate transform as

EℓZj=−zN−j+ℓ+2,qN/2′=(−1)ℓ+1qN/2.(93)\mathcal{E}_{\ell Z_j}=-z_{N-j+\ell+2},\qquad q'_{N/2}=(-1)^{\ell+1}q_{N/2}. \tag*{(93)}

If the special coordinate is active, coincidence with the cyclic image requires (−1)k=(−1)ℓ+1(-1)^k=(-1)^{\ell+1}. Thus k+ℓk+\ell, and therefore kCk_C, must be odd.

When the complete condition holds, an in-plane twofold rotation preserves the elevation pattern. Composing any one such operation with the fRf_R preserving cyclic shifts generates all preserving Type-C operations. Type-C therefore doubles the multiplicity supplied by Type-R, giving fC=2f_C=2; otherwise, fC=1f_C=1.

Type-I: inversion-related redundancies

Type-I connects blocks with the same pp and opposite ss:

CNk↔SNk′,C2ℓ,k↔σvℓ′.(94)C_N^k \leftrightarrow S_N^{k'},\qquad C_2^{\ell,k} \leftrightarrow\sigma_v^{\ell'}. \tag*{(94)}

Global elevation inversion adds a half-turn to every ordinary-mode phase. It is equivalent to a cyclic shift if a single compensating displacement k∗k^* satisfies

mk∗N=nm+12(95)\frac{m k^*}{N}=n_m+\frac{1}{2} \tag*{(95)}

where nmn_m is an integer number. If the special coordinate is active, the same condition must hold for m=N/2m=N/2, or equivalently for every index in M+\mathcal{M}_+. For this coordinate, the condition expresses sign compensation and requires odd k∗k^*.

Without an active special coordinate, a cyclic shift with optional elevation inversion preserves all ordinary phases when their rotational contributions are either all integers or all half-integers. Choosing an active mode m1m_1 as a reference, the necessary and sufficient conditions are

2m1kN∈Z,(m−m1)kN∈Z.(96)\frac{2m_1k}{N}\in\mathbb{Z},\qquad\frac{(m-m_1)k}{N}\in\mathbb{Z}. \tag*{(96)}

The first condition permits either fractional part; the difference conditions enforce the same inversion choice for every mode. The divisibility argument used for Type-R gives

fRI=gcd⁡(N,2m1,{m−m1}).(97)f_{\mathrm{RI}}=\gcd\left(N,2m_1,\{m-m_1\}\right). \tag*{(97)}

This expression is independent of the chosen reference mode.

If the special coordinate is active, its preservation requires s(−1)k=1s(-1)^k=1, hence s=(−1)ks=(-1)^k. The inversion contribution is then equivalent to k/2k/2 modulo integers, and the ordinary-phase conditions become

(m+N/2)kN∈Z.(98)\frac{(m+N/2)k}{N}\in\mathbb{Z}. \tag*{(98)}

Therefore, when ordinary modes are also active,

fRI=gcd⁡(N,{m+N/2}).(99)f_{\mathrm{RI}}=\gcd\left(N,\{m+N/2\}\right). \tag*{(99)}

If only the special coordinate is active, every displacement admits exactly one preserving inversion choice, giving fRI=Nf_{\mathrm{RI}}=N.

In either nonplanar case,

fI=fRIfR.(100)f_{\mathrm{I}}=\frac{f_{\mathrm{RI}}}{f_{\mathrm{R}}}. \tag*{(100)}

A compensating inversion–shift operation, when present, generates as many additional preserving operations as there are preserving rotations. Thus fI=2f_{\mathrm{I}}=2 when Type-I holds, and fI=1f_{\mathrm{I}}=1 otherwise. Type-I depends only on the active-mode set, not on the phases. For the planar pattern, elevation inversion is trivial and fI=2f_{\mathrm{I}}=2.

Type-P: mirror-type redundancies

Type-P connects blocks with opposite values of both labels:

CNk↔σvℓ,C2ℓ,k↔SNk.(101)C_N^k \leftrightarrow\sigma_v^\ell,\qquad C_2^{\ell,k}\leftrightarrow S_N^k. \tag*{(101)}

Elevation inversion cancels the normal reversal associated with reflected indexing. Equating the phases modulo 2π2\pi gives

ϕm+2πmkN≡−ϕm−2πmℓN(mod2π).(102)\phi_m+\frac{2\pi m k}{N}\equiv-\phi_m-\frac{2\pi m\ell}{N}\pmod{2\pi}. \tag*{(102)}

With kP=−(k+ℓ)k_{\mathrm{P}}=-(k+\ell), this reduces to

ϕm≡πmkPN(modπ).(103)\phi_m\equiv\frac{\pi m k_{\mathrm{P}}}{N}\pmod{\pi}. \tag*{(103)}

Again, a single integer kPk_{\mathrm{P}} must satisfy every active-phase condition. The special coordinate transforms as qN/2′=(−1)ℓqN/2q_{N/2}'=(-1)^\ell q_{N/2}, so coincidence with the cyclic image requires (−1)k=(−1)ℓ(-1)^k=(-1)^\ell, and hence even kPk_{\mathrm{P}}, when this coordinate is active.

The complete condition identifies a mirror operation preserving the reference elevation pattern. Specifically, applying the inverse cyclic rotation to the two coincident images gives a preserving mirror with axis index a=k+ℓ=−kP(modN)a=k+\ell=-k_{\mathrm{P}}\pmod{N}; this need not be the axis ℓ\ell used in the initial comparison. As for Type-C, one preserving operation generates fRf_{\mathrm{R}} such operations by composition with the preserving rotations. Hence fP=2f_{\mathrm{P}}=2 when Type-P is present, and fP=1f_{\mathrm{P}}=1 otherwise.

Examples of the semidiscrete conformational nomenclature

tbl:X illustrates the notation for representative conformations of six- to eight-membered rings, showing how their local amplitude-balance angles and nonvanishing phases are encoded in the semidiscrete labels.

For instance, the two chair conformations of cyclohexane are naturally denoted as [Z(+)(6)[\mathbb{Z}_{(+)}(6) and Z(−)(6)\mathbb{Z}_{(-)}(6), whereas an equatorial six-membered-ring conformation may be labeled [E](6;ϕ2)[E](6;\phi_2). Thus, the boat conformation is [E](6;0)[E](6;0) and the skew conformation is [E](6;30)[E](6;30). In this case, the symbol E indicates that q2>0q_2>0 and q3=0q_3=0. No sign subscript is used because the special coordinate vanishes, and the only active phase, ϕ2\phi_2, is retained. The special coordinate q3q_3 has no associated phase, regardless of whether it is active.

More generally, phases are omitted whenever their associated ordinary-mode amplitudes vanish. For example, in a twelve-membered ring, ψ1=90∘\psi_1=90^\circ corresponds to q2>0q_2>0 and q3=0q_3=0, so the first LAB symbol is E and ϕ3\phi_3 is omitted from the conformational label. The remaining symbols and phases depend on the other active modes.

For N=8N=8, if q2=q3=0q_2=q_3=0 and q4≠0q_4\ne0, the LAB symbol string is 0Z. The first symbol is 0 because both ordinary amplitudes vanish, whereas the terminal symbol is Z because q3=0q_3=0 and ∣q4∣>0|q_4|>0. Both ordinary phases, ϕ2\phi_2 and ϕ3\phi_3, are omitted, and the conformational label reduces to [0Z](+)(8)[0Z]_{(+)}(8) or [0Z](−)(8)[0Z]_{(-)}(8), depending on the sign of the surviving special coordinate q4q_4.

NNψ1\psi_1ψ2\psi_2DescriptionNomenclature
690°—boat (B)E
690°—skew (S)E
60°—chair (C)Z(6)
790°—boat (B)E
790°—twist-boat (TB)E
70°—chair (C)Z
70°—twist-chair (TC)Z
890°—boat-boat (BB)E0
890°—twist-boat (TB)E0
80°90°long-chair (L)ZE
80°90°chair (C)ZE
8—0°crown (W)0Z(8)

Table X. Representative conformations of six- to eight-membered rings expressed in terms of local amplitude-balance angles and the semidiscrete nomenclature. Phase values included in the labels are given in degrees, with the degree symbol omitted. A dash indicates that the corresponding local angle is undefined or does not exist. The subscript (±) denotes the two possible signs of the active special coordinate.

Reduction of phase sampling

Table XI compares the numbers of sampled CP points in full and symmetry-reduced grids for homogeneous rings containing six to twenty atoms. The counts include all combinations of two or three active amplitudes and use the original isolated-mode phase steps in both grids. Their difference therefore measures the reduction due only to symmetry. For even-membered rings, combinations involving the special coordinate are also included. The full grids sample both signs, whereas the symmetry-reduced grids use only the positive sector. Grid points follow the closed or half-open character of each interval, without adding endpoints outside the regular grid.

NNTwo active amplitudesThree active amplitudesFactor
FullSymmetryFullSymmetry
6242——12/12
778443——18/18
8176112561318/18
92 1609315 55268423/65
101 3207010 40040825/44
1111 616402340 73611 79229/104
121 2166011 42443725/47
1327 0407901 406 08041 08034/128
148 120316235 2007 15733/86
1521 040561940 80024 39239/141
167 696297221 6966 59133/99
1797 1042 16311 005 120245 14045/173
1817 832560838 36820 73340/117
19161 7283 22024 582 656489 44050/195
2018 816674939 20021 76743/111

Table XI. Full and symmetry-reduced CP grid counts for searches with two and three active amplitudes. The counts shown use the original isolated-mode phase steps in both grids, so their reduction is due only to symmetry. Both signs of the special coordinate are included in the full grids, whereas reduced grids use only its positive sector. Factor lists two rounded reduction factors for the combined searches, separated by a slash. The first includes only symmetry reduction, whereas the second also includes selective doubling of the second and third phase steps for three ordinary modes.

The last column of Table XI lists two reduction factors, rounded to the nearest integer and separated by a slash. The first is obtained by dividing the combined full-grid count for the two searches by the corresponding symmetry-reduced count shown in the table. The second also includes the selective increase in phase steps described in Sec. V A. For three active ordinary modes, the second and third phase steps are doubled only when they divide the corresponding interval lengths exactly. This rule is applied separately to the second-phase interval and its reflection-reduced interval on the zero-phase boundary. Combinations involving the special coordinate retain the original steps.

Both factors measure reductions in sampled CP points, not in reconstructed Cartesian geometries, optimized conformers, or computational time. Unlike symmetry reduction, increasing the phase steps changes the sampling resolution and provides an additional saving rather than removing equivalent representations.

The symmetry reduction depends on ring size, the active-mode combinations, and the treatment of domain boundaries.

Common divisors of the ring size and mode-index combinations determine the redundancy factors and sequential phase intervals. Some active-mode subsets admit nontrivial operations that leave every pattern in the subset unchanged. These operations reduce the number of distinct CP images per pattern and therefore the reduction available within that subset. For prime ring sizes, such phase-independent redundancies are absent for subsets of ordinary modes, giving a more uniform reduction across combinations. For even-membered rings, the special coordinate introduces an additional dependence on the operations that preserve its sign. The finite-grid reduction also reflects the relative contribution of boundary points.

A larger reduction factor does not necessarily imply a smaller search. For example, symmetry alone gives rounded combined factors of 50 and 33 for the nineteen- and sixteen-membered rings, respectively. Nevertheless, their symmetry-reduced grids contain 492 660 and 6 888 points. The larger ring has more available mode combinations and finer isolated-mode phase steps, resulting in a substantially larger grid.

Selective phase-grid coarsening increases these factors to approximately 195 and 99 and reduces the corresponding counts to 126 644 and 2 320 points. Thus, the two factors distinguish the saving obtained by removing symmetry-equivalent descriptions from the additional saving obtained by sampling less finely. The reduced point count indicates the size of the remaining sampling task.

Acknowledgments

This work has received financial support from MICIU/AEI/PID2025-173245NB-I00, Xunta de Galicia (Centro singular de Investigación de Galicia, accreditation 2023-2027, ED431G 2023/03 and the European Regional Development Fund (ERDF), and Grupo de Referencia Competitiva, ED431C 2025/06) and María de Maeztu Excellence Accreditation CEX2025-001612-M funded by MICIU/AEI /10.13039/501100011033. The authors thank Galician Supercomputing Center (CESGA) for providing access to their computing facilities.

References

  1. [1]J. E. Kilpatrick, K. S. Pitzer, and R. Spitzer, J. Am. Chem. Soc. 69, 2483 (1947).
  2. [2]D. Cremer and J. A. Pople, J. Am. Chem. Soc. 97, 1354 (1975).
  3. [3]H. M. Pickett and H. L. Strauss, J. Chem. Phys. 55, 324 (1971).
  4. [4]J. C. A. Boeyens and D. G. Evans, Acta Cryst. B45, 577 (1989).
  5. [16]A. Lema-Saavedra and A. Fernández-Ramos, J. Chem. Phys. 164, 074109 (2026).
  6. [17]J. Dale, Acta Chem. Scand. 27, 1115 (1973).
  7. [18]N. L. Allinger and S. Profeta Jr., Tetrahedron 36, 859 (1980).
  8. [19]F. A. L. Anet and A. K. Cheng, J. Am. Chem. Soc. 97, 2420 (1975).
  9. [20]V. L. Shannon, H. L. Strauss, R. G. Snyder, C. A. Elliger, and W. L. Mattice, J. Am. Chem. Soc. 111, 1947 (1989).
  10. [21]W. Zou, Y. Tao, and E. Kraka, J. Chem. Phys. 152, 154107 (2020).
  11. [22]J. S. Binkley, J. A. Pople, and W. J. Hehre, J. Am. Chem. Soc. 102, 939 (1980).
  12. [23]C. Adamo and V. Barone, J. Chem. Phys. 110, 6158 (1999).
  13. [24]F. Weigand and R. Ahlrichs, Phys. Chem. Chem. Phys. 7, 3297 (2005).
  14. [25]B. Rplinger, B. Sandhoefer, A. Hansen, and F. Neese, J. Chem. Phys. 139, 134101 (2013).
  15. [26]T. H. Dunning, Jr., J. Chem. Phys. 90, 1007 (1989).
  16. [27]M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. V. Marenich, J. Bloino, B. G. Janesko, R. G. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. R. Nascimento, V. G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, Q. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas, J. B. Foresman, and D. J. Fox, “Gaussian 16 Revision C.01,” (2016), gaussian Inc. Wallingford CT.
  17. [28]F. Neese, Wiley Interdiscip. Rev. Comput. Mol. Sci. 15, e70019 (2025).
  18. [29]A. Lema-Saavedra and A. Fernández-Ramos, “Symmetry and redundancy in Cremer–Pople phases for ring puckering,” (2026), DOI: 10.5281/zenodo.23985089.

Paper details

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

views · 9

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