research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

Journal logoJOURNAL OF
APPLIED
CRYSTALLOGRAPHY
ISSN: 1600-5767

Orientation relationships, orientational variants and the embedding approach

crossmark logo

aSchool of Mathematics and Statistics, Victoria University of Wellington, PO Box 600, Wellington, New Zealand, bSchool of Mathematics and Statistics, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9SS, United Kingdom, and cGeophysics and Geoscience Informatics, TU Bergakademie Freiberg, Freiberg, Germany
*Correspondence e-mail: pej@st-andrews.ac.uk

Edited by A. R. Pearson, Universität Hamburg, Germany (Received 16 May 2022; accepted 6 April 2023; online 12 May 2023)

For phase transformations within polycrystalline materials, the connection between the crystal orientations of parent grains and those of child grains is usually expressed in terms of (theoretical or measured) orientation relationships. This paper introduces a new approach to various problems associated with orientation relationships: (i) estimation, (ii) whether or not a single orientation relationship fits the data adequately, (iii) whether or not a set of children comes from a common parent, and (iv) reconstruction of a parent or of grain boundaries. The approach is an extension to the crystallographic context of the well established embedding approach to directional statistics. It is inherently statistical, producing precise probabilistic statements. Explicit coordinate systems are not used and arbitrary thresholds are avoided.

1. Introduction

Phase transformations within polycrystalline materials often induce a transformation of the texture, i.e. of the statistical and spatial distribution of the crystallographic orientations of the crystallites. This usually involves the replacement of each parent crystal by several child crystals, possibly in a different symmetry group. Several mathematical models for describing these transformations have been constructed on theoretical grounds. Such models include the orientation relationships of Kurdjumow–Sachs (Kurdjumow & Sachs, 1930[Kurdjumow, G. & Sachs, G. (1930). Z. Phys. 64, 325-343.]), Nishiyama–Wassermann (Wassermann, 1933[Wassermann, G. (1933). Arch. Eisenhüttenwesen 6, 347-351.], 1935[Wassermann, G. (1935). Über den Mechanismus der α-γ Umwand­lung des Eisens. Mitteilungen aus dem Kaiser-Wilhelm-Institut für Eisenforschung zu Düsseldorf. Düsseldorf: Verlag Stahleisen.]; Nishiyama,1934[Nishiyama, Z. (1934). Sci. Rep. Tohoku Univ. 23, 637-664.]) and Pitsch (1959[Pitsch, W. (1959). Philos. Mag. 4, 577-584.], 1967[Pitsch, W. (1967). Arch. Eisenhüttenwesen 38, 853-864.]). The presence of two symmetry groups leads to the problem of multiplicity of variants, i.e. orientations of (unseen) parent crystals giving rise to orientations of given child crystals. (A precise mathematical definition is given in Section 2.3[link].) Much work has been carried out developing methods for fitting these models and examining their adequacy, notably by Humbert and co-authors (e.g. Humbert et al., 1994[Humbert, M., Moustahfid, H., Wagner, F. & Philippe, M. J. (1994). Scr. Metall. Mater. 30, 377-382.], 1995[Humbert, M., Wagner, F., Moustahfid, H. & Esling, C. (1995). J. Appl. Cryst. 28, 571-576.], 2015[Humbert, M., Germain, L., Gey, N. & Boucard, E. (2015). Acta Mater. 82, 137-144.]) and by Cayron and his collaborators (e.g. Cayron et al., 2006[Cayron, C., Artaud, B. & Briottet, L. (2006). Mater. Charact. 57, 386-401.]; Cayron, 2019[Cayron, C. (2019). Acta Cryst. A75, 411-437.]). Even within the topic of the austenite–martensite transformation there are very many papers in the materials science literature on problems involving variants, e.g. Nolze (2004b[Nolze, G. (2004b). Z. Metallkd. 95, 744-755.]), Kitahara et al. (2006[Kitahara, H., Ueji, R., Tsuji, N. & Minamino, Y. (2006). Acta Mater. 54, 1279-1288.]), Miyamoto et al. (2009[Miyamoto, G., Takayama, N. & Furuhara, T. (2009). Scr. Mater. 60, 1113-1116.]), Abbasi et al. (2012[Abbasi, M., Nelson, T. W., Sorensen, C. D. & Wei, L. (2012). Mater. Charact. 66, 1-8.], 2014[Abbasi, M., Kim, D.-I., Nelson, T. W. & Abbasi, M. (2014). Mater. Charact. 95, 219-231.]), Koumatos & Muehlemann (2017[Koumatos, K. & Muehlemann, A. (2017). Acta Cryst. A73, 115-123.]) and Nyyssönen et al. (2016[Nyyssönen, T., Isakov, M., Peura, P. & Kuokkala, V.-T. (2016). Metall. Mater. Trans. A, 47, 2587-2590.], 2018[Nyyssönen, T., Peura, P. & Kuokkala, V.-T. (2018). Metall. Mater. Trans. A, 49, 6426-6441.]). These problems have been addressed by Mainprice et al. (1990[Mainprice, D., Humbert, M. & Wagner, F. (1990). Tectonophysics, 180, 213-228.]) and Morales et al. (2018[Morales, L. F. G., Mainprice, D. & Kern, H. (2018). Tectonophysics, 724-725, 93-115.]), for instance, in a geological context.

As far as we are aware, previous work on fitting orientation relationships and identifying variants takes the traditional viewpoint of estimating the relationship and assessing its fit, either informally or by making use of an arbitrary threshold of some measure of goodness of fit. For example, Nolze (2004a[Nolze, G. (2004a). Proceedings of the CHANNEL Users Conference 2004, Ribe, Denmark, pp. 37-45. Berlin: BAM.],b[Nolze, G. (2004b). Z. Metallkd. 95, 744-755.], 2005[Nolze, G. (2005). Determining the f.c.c./b.c.c. Orientational Relationship in Plessite Regions of Iron Meteorites. HKL Users Meeting.]) measures the discrepancy between estimated and theoretical versions of a parameter [A]1,2 of an orientation relationship [see equation (1[link]) below] by the (minimal) misorientation angle between representative rotations. The discrepancy is considered significant if it is larger than the accuracy of the experimental apparatus. In this paper we establish a sound statistical viewpoint on various types of problem involving orientation relationships and variants. In the traditional viewpoint, the key idea is that of being `numerically close' and the typical question is `how far apart are the measured and theoretical quantities?'. From the statistical viewpoint, the key idea is that of being `statistically close' (in a precise probabilistic sense) and the typical question is `what is the probability (under some hypothesis) that the measured and theoretical quantities are (at least) this far apart?'. If this probability is small then the hypothesis is rejected. We implement the statistical viewpoint here by using the embedding approach to crystallographic orientations introduced by Arnold et al. (2018[Arnold, R., Jupp, P. E. & Schaeben, H. (2018). J. Multivariate Anal. 165, 73-85.]), which is an application of the embedding approach to directional statistics [see e.g. Sections 2.1 and 9.1 of Mardia & Jupp (2000[Mardia, K. V. & Jupp, P. E. (2000). Directional Statistics. Chichester: Wiley.])]. The necessary measures of distance are constructed using orientation representations which incorporate the relevant symmetries, reducing the need to search for optimal representations within equivalence classes. The use of explicit coordinate systems is avoided.

The key points of our approach are as follows:

(i) Regarding each object such as a crystallographic orientation as a single object (equivalence class) in a quotient space, rather than considering a single representative orientation.

(ii) Avoiding arbitrary thresholds.

(iii) Avoiding arbitrarily chosen explicit coordinate systems (such as are used in asymmetric domains).

The notation used in this article is more mathematical and general than that traditionally used for discussing variants. The reasons for this choice are (i) the concepts of orientation relationship and variant are appropriate in contexts wider than the crystallographic, and (ii) for the embedding approach our notation is particularly convenient. We use the word `random' in the statistical sense of `not deterministic', so that repeating measurement of e.g. rotations will not produce exactly the same values (rather than implying that these rotations are uniformly distributed).

Some of the material in this paper has been presented in a more mathematical form suitable for statisticians by Arnold & Jupp (2019[Arnold, R. & Jupp, P. E. (2019). Applied Directional Statistics: Modern Methods and Case Studies, edited by C. Ley & T. Verdebout, pp. 41-60. London: Chapman & Hall/CRC.]) and Arnold et al. (2018[Arnold, R., Jupp, P. E. & Schaeben, H. (2018). J. Multivariate Anal. 165, 73-85.], 2021[Arnold, R., Jupp, P. E. & Schaeben, H. (2021). Directional Statistics for Innovative Applications, edited by A. SenGupta & B. C. Arnold. Heidelberg: Springer.]).

Section 2[link] gives a mathematical description of orientation relationships and variants. The embedding approach to crystallographic orientations is recalled in Section 3[link] and extended to the context of variants. Sections 4[link], 5[link], 6[link] and 7[link] treat methods for estimation (sometimes referred to as `calculation', `determination' or `evaluation') of orientation relationships, assessment of the adequacy of a single orientation relationship, assessment of the common parentage of child crystals and reconstruction of parents, respectively. In each of these sections established methods are recalled and then methods based on the embedding approach are given. Some practical illustrations of these new methods are given in Section 8[link].

2. Orientation relationships and variants

2.1. Orientations of symmetrical objects

The orientation of a rigid object in [{\bb R}^{3}] can be described by a rotation that transforms it into some standard orientation. If the object is asymmetrical then this rotation is unique, so that the orientations of the object correspond to elements of the group SO(3) of rotations of [{\bb R}^{3}]. If the object has symmetry group K then a rotation U has the same effect as the rotation UH for any H in K. Then the orientations of the object correspond to elements of the space SO(3)/K, i.e. the set of equivalence classes of elements of SO(3) under the right action of K. For U in SO(3) we shall denote the corresponding left coset of K, i.e. the equivalence class {UH : HK} of U in SO(3)/K by [U].

2.2. Orientation relationships

In many contexts, e.g. transformations involving a phase change, interest lies in the relationship between two random orientations with possibly different symmetry groups. Let [U]1 and [V]2 be random orientations in SO(3)/K1 and SO(3)/K2, respectively, where K1 and K2 are the symmetry groups. Particularly simple models relating [U]1 and [V]2 are the orientation relationships. These have the form

[[{\bf V}]_{2} = [{\bf R}{\bf U}{\bf A}]_{2}, \eqno(1)]

where R and A are in SO(3). {Note that, in general, equation (1[link]) is not equivalent to [U]1 = [R−1VA−1]1.} The rotation R in (1[link]) can arise as the result of measurements being made in different coordinate systems. For example, this situation arises when the two samples are aligned at different angles to a common laboratory measurement frame. A further complication may occur if pairs of measurements ([U1]1, [V1]2),…, ([Un]1, [Vn]2) are made at widely differing locations, in which case realistic modelling may require the use of an R that depends on location. Equation (1[link]) determines the rotation R uniquely but it does not determine A. Indeed, because [U]1 = [UH]1 and [V]2 = [[{\bf V} {\widetilde {\bf H}}]_{2}] for any H in K1 and any [{\widetilde {\bf H}}] in K2, [{\bf H}{\bf A} {\widetilde {\bf H}}] and A give the same orientation relationship (1[link]). Thus (1[link]) does not determine A fully but determines only its image [A]1,2 in the double coset space [K_{1} \backslash SO(3)/K_{2}]. The space [K_{1} \backslash SO(3)/K_{2}] is the set of equivalence classes of elements W of SO(3) for which W and [{\bf H}{\bf W} {\widetilde {\bf H}}] are equivalent for any H in K1 and any [{\widetilde {\bf H}}] in K2. In crystallography it is usual to identify [K_{1} \backslash SO(3)/K_{2}] with an asymmetric domain, i.e. a connected subset of SO(3) that (apart from a set of measure zero) contains exactly one rotation in each equivalence class. Construction of asymmetric domains is considered in Section 6.3 of Morawiec (2004[Morawiec, A. (2004). Orientations and Rotations, 1st ed. Berlin: Springer-Verlag.]). Because, in any given context, there is no standard asymmetric domain, we prefer not to use such domains. An orientation relationship (1[link]) gives rise to four types of problem:

(i) The estimation problem of estimating the unknown R and [A]1,2 on the basis of observations ([U1]1, [V1]2),…, ([Un]1, [Vn]2).

(ii) The single orientation relationship problem of assessing whether or not a single orientation relationship can describe the data adequately.

(iii) The sibling problem of determining whether or not elements [V1]2,…, [Vn]2 of SO(3)/K2 arise from some unknown common parent [U]1 under a known orientation relationship (R, [A]1,2). If [V1]2,…, [Vn]2 are distinct then they arise from some common U if and only if they are part of a set of variants (see Section 2.3[link]).

(iv) The reconstruction problem of estimating the unknown [U]1 on the basis of a known (R, [A]1,2) and observations [V1]2,…, [Vn]2.

These problems are considered in Sections 4[link], 5[link], 6[link] and 7[link], respectively.

2.3. Variants

In general, an orientation relationship (1[link]) does not yield a unique map from SO(3)/K1 to SO(3)/K2. Instead, to each [U]1 in SO(3)/K1 it assigns s distinct elements

[[{\bf R} {\bf U} {\bf H}_{1} {\bf A}]_{2}, \ldots, [{\bf R} {\bf U} {\bf H}_{s} {\bf A}]_{2} \eqno(2)]

of SO(3)/K2, where H1,…, HsK1 and

[s = |K_{1}|/|K_{\bf A}| , \eqno(3)]

KA being the intersection group defined as

[K_{\bf A} = K_{1} \cap \left ( {\bf A} K_{2} {\bf A}^{-1} \right ) . \eqno(4)]

Distinctness of [RUH1A]2,…, [RUHsA]2 in (2[link]) is equivalent to [{\bf H}_{j}^{-1} {\bf H}_{i} \ \notin \ K_{\bf A}] for ij (see Appendix A[link]). The [RUH1A]2,…, [RUHsA]2 in (2[link]) are known as the (orientation) variants (or crystallographic variants) of [U]1 given by R and A. There is no distinguished variant among [RUH1A]2,…, [RUHsA]2, but the (arbitrary) choice of any one of these, [RUH0A]2, say, as a base point determines the function [\phi_{{\bf H}_{0}}] from K1 to the set of variants by [\phi_{{\bf H}_{0}} ([{\bf H}]_{1})] = [[{\bf R}{\bf U} {\bf H}_{0} {\bf H} {\bf A}]_{2}]. Since [\phi_{{\bf H}_{0}} ([{\bf H}_{i}]_{1})] = [\phi_{{\bf H}_{0}} ([{\bf H}_{j}]_{1})] if and only if [{\bf H}_{j}^{-1} {\bf H}_{i} \in K_{\bf A}], the set (2[link]) of variants can be identified with the coset space K1/KA.

As pointed out by Nolze (2008[Nolze, G. (2008). Cryst. Res. Technol. 43, 61-73.]), in contrast to the commonly used theoretical orientation relationships [A]1,2, measured orientation relationships [[{\widehat {\bf A}}]_{1,2}] are `irrational' in that their descriptions in terms of crystallographic planes and directions do not have low Miller indices. This is because it is not possible to make measurements with perfect accuracy, and so the orientations [Ui]1 and [Vi]2 are random. The argument in Proposition 1[link] in Appendix A[link] shows that the number of variants associated with a measured orientation relationship is |K1|. In real materials the variants are usually present in unequal quantities, a phenomenon known as variant selection, and some variants may even be absent. In many contexts it is of interest to estimate the overall proportions of the variants that are present. Whereas the orientational variants considered in this paper are cosets of KA in K1, the types of variant introduced by Cayron (2016[Cayron, C. (2016). Acta Mater. 111, 417-441.], 2019[Cayron, C. (2019). Acta Cryst. A75, 411-437.]) in connection with the physical mechanisms underlying martensitic crystallography are cosets of other subgroups of K1 that describe the distortion and stretch of crystal lattices.

As pointed out by Cayron (2006[Cayron, C. (2006). Acta Cryst. A62, 21-40.]), the important algebraic structure associated with the set (2[link]) of variants is that of a groupoid, i.e. a set of arrows endowed with a partially defined associative composition in which every arrow has an inverse. Denote the set (2[link]) of variants by V. Each triple ([[{\bf R} {\bf U} {\bf H}_{i} {\bf A}]_{2}], [K_{\bf A} {\bf H}_{i}^{-1} {\bf H}_{j} K_{\bf A}], [[{\bf R} {\bf U} {\bf H}_{j} {\bf A}]_{2}]) in [V \times (K_{\bf A} \backslash K_{1}/K_{\bf A}) \times V] can be regarded as an arrow from [RUHiA]2 to [RUHjA]2. The arrows ([[{\bf R} {\bf U} {\bf H}_{i} {\bf A}]_{2}], [K_{\bf A} {\bf H}_{i}^{-1} {\bf H}_{j} K_{\bf A}], [[{\bf R} {\bf U} {\bf H}_{j} {\bf A}]_{2}]) and ([[{\bf R} {\bf U} {\bf H}_{k} {\bf A}]_{2}], [K_{\bf A} {\bf H}_{k}^{-1} {\bf H}_{\ell} K_{\bf A}], [[{\bf R} {\bf U} {\bf H}_{\ell} {\bf A}]_{2}]) can be composed if and only if [RUHjA]2 = [RUHkA]2, in which case the composition is ([[{\bf R} {\bf U} {\bf H}_{i} {\bf A}]_{2}], [K_{\bf A} {\bf H}_{\ell}^{-1} {\bf H}_{k} K_{\bf A}], [[{\bf R} {\bf U} {\bf H}_{\ell} {\bf A}]_{2}]). It is useful to combine the arrows into equivalence classes called operators. Each operator can be written as [K_{\bf A} {\bf H}_{i}^{-1} {\bf H}_{j} K_{\bf A}] for some i, j (but, in general, i and j are not unique) and can be regarded as being a theoretical transformation that takes the variant [RUHiA]2 to [RUHjA]2. Thus, the set of operators can be identified with the double coset space [K_{\bf A} \backslash K_{1}/K_{\bf A}]. The operator [K_{\bf A} {\bf H}_{i}^{-1} {\bf H}_{j} K_{\bf A}] can be regarded as the misorientation between [RUHiA]2 and [RUHjA]2.

The composition table of the groupoid yields a multi-valued composition on the set of operators. This provides a way of attacking the sibling and reconstruction problems described in points (iii) and (iv) at the end of Section 2.2[link]. Gey & Humbert (2003[Gey, N. & Humbert, M. (2003). J. Mater. Sci. 38, 1289-1294.]) pointed out that, in the case of the Burgers orientation relationship (K1 = O, K2 = D6), the number of misorientations between the variants is less than the number of variants.

3. The embedding approach

3.1. Embedding SO(3)/K

Because the coset spaces SO(3)/K are not very easy to work with, Arnold et al. (2018[Arnold, R., Jupp, P. E. & Schaeben, H. (2018). J. Multivariate Anal. 165, 73-85.]) (see also Arnold & Jupp, 2019[Arnold, R. & Jupp, P. E. (2019). Applied Directional Statistics: Modern Methods and Case Studies, edited by C. Ley & T. Verdebout, pp. 41-60. London: Chapman & Hall/CRC.]) developed the embedding approach in which a function t : SO(3)/KE is used to send SO(3)/K into (but not onto) some inner-product space E. The function t is required to be (i) one-to-one, (ii) equivariant, i.e.t([VU]), t([VW])〉 = 〈t([U]), t([W])〉 for U, V, W in SO(3), where 〈·, ·〉 denotes the inner product, (iii) such that t([U]) has expectation 0 if [U] is uniformly distributed on SO(3)/K. For the crystallographic groups C1, C2, C3, C4, D2, D6, T and O some useful functions t are given explicitly in Table 1[link]. Together t and 〈·, ·〉 lead us to

[\| {\bf t} ([{\bf U}]) - {\bf t} ([{\bf V}]) \|^{2} = \langle {\bf t} ([{\bf U}]) - {\bf t} ([{\bf V}]), {\bf t} ([{\bf U}]) - {\bf t} ([{\bf V}]) \rangle \eqno(5)]

as a new measure of squared distance between elements [U] and [V] of SO(3)/K. By design it incorporates the symmetry group K and it replaces the need for misorientation angles. A useful summary of [U1],…, [Un] is their sample mean, [[{\overline {\bf U}}]], which is defined as the element of SO(3)/K that minimizes [\sum \nolimits_{i=1}^{n} \| {\bf t} ([{\overline {\bf U}}]) - {\bf t} ([{\bf U}_{i}]) \|^{2}] or, equivalently, maximizes [\langle {\bf t} ([{\overline {\bf U}}]), \sum \nolimits_{i=1}^{n} {\bf t} ([{\bf U}_{i}]) \rangle]. For K = Cr or Dr explicit approximations to sample means are given in Section 2.3 of Arnold & Jupp (2019[Arnold, R. & Jupp, P. E. (2019). Applied Directional Statistics: Modern Methods and Case Studies, edited by C. Ley & T. Verdebout, pp. 41-60. London: Chapman & Hall/CRC.]).

Table 1
Some embeddings tK : SO(3)/KE

For Cr with r ≥ 3, u0 = [sin(2π/r)]−1 u1 × u2. For D2, u3 = ±u1 × u2. The symmetric arrays ⊗rui and Nr are defined in equations (37[link]) and (38[link]), respectively.

Group, K tK
C1 [{\bf t}_{C_{1}} ({\bf u}_{1}, {\bf u}_{2}, {\bf u}_{3}) = ({ {\bf u}}_{1}, {\bf u}_{2}, {\bf u}_{3})]
C2 [{\bf t}_{C_{2}} ({\bf u}_{0}, \pm{\bf u}_{1}) = ({\bf u}_ {0}, {\bf u}_{1}{\bf u}_{1}^{\rm T} - {\textstyle {1 \over 3}} {\bf I}_{3})]
C3 [{\bf t}_{C_{3}} ([{\bf u}_{1}, {\bf u}_{2}, {\bf u}_{3}]) = \left ( {\bf u}_{0}, \sum_{i = 1}^{3} \otimes^{3} {\bf u}_{i} \right )]
C4 [{\bf t}_{C_{4}} ([{\bf u}_{1}, \ldots, {\bf u}_{4}]) = \left ( {\bf u}_{0}, \sum_{i = 1}^{4} \otimes^{4} {\bf u}_{i} - {\textstyle {4 \over 5}} {\bf N}_{4} \right )]
D2 [{\bf t}_{D_{2}} (\pm{\bf u}_{1}, \pm{\bf u}_{2}) = ({\bf u}_{1} {\bf u}_{1}^{\rm T} - {\textstyle {1 \over 3}} {\bf I}_{3}, {\bf u}_{2} {\bf u}_{2}^{\rm T} - {\textstyle {1 \over 3}} {\bf I}_{3}, {\bf u}_{3} {\bf u}_{3}^{\rm T} - {\textstyle {1 \over 3}} {\bf I}_{3})]
D4 [{\bf t}_{D_{4}} ([{\bf u}_{1}, \ldots, {\bf u}_{4}]) =\sum_{i = 1}^{4} \otimes^{4} {\bf u}_{i} - {\textstyle {4 \over 5}} \, {\bf N}_{4}]
D6 [{\bf t}_{D_{6}} ([{\bf u}_{1}, \ldots, {\bf u}_{6}]) = \sum_{i = 1}^{6} \otimes^{6} {\bf u}_{i} - {\textstyle {6 \over 7}} {\bf N}_{6}]
T [{\bf t}_T (\{ {u}_1, {\bf u}_2, {\bf u}_3, {\bf u}_4 \}) = \otimes^3 {\bf u}_1 + \otimes^3 {\bf u}_2 + \otimes ^3 {\bf u}_3 + \otimes^3 {\bf u}_4]
O [{\bf t}_O (\{ \pm {\bf u}_1, \pm {\bf u}_2, \pm {\bf u}_3 \}) = \otimes^4 {\bf u}_1 + \otimes^4 {\bf u}_2 + \otimes^4 {\bf u}_3 - {\textstyle {3 \over 5}} {\bf N}_4]

Embeddings are discussed further in Appendix B[link].

3.2. Embedding K1\SO(3)/K2

The double coset spaces [K_{1} \backslash SO(3)/K_{2}] are even more complicated than the coset spaces SO(3)/K, but the embedding approach can be used for them also.

Any t : SO(3)/K2E, where E is a vector space, can be averaged over K1 to give a corresponding [{\overline {\bf t}}:K_{1} \backslash SO(3)/K_{2}\to E], defined by

[{\overline {\bf t}} ([{\bf U}]_{1,2}) = |K_{1}|^{-1} \sum_{{\bf H}\in K_{1}} {\bf t} ([{\bf H} {\bf U}]_{2}) . \eqno(6)]

We shall exploit such [{\overline {\bf t}}] in order to carry out inference on orientation relationships. If t has properties (ii) and (iii) of Section 3.1[link] then so does [{\overline {\bf t}}]. On the other hand, t having property (i) (being one-to-one) does not imply that [{\overline {\bf t}}] has this property also. In a few very special cases (e.g. K1 and K2 isomorphic to C2 and C3 and with a common axis) [{\overline {\bf t}}] is identically zero. In the context of phase transitions it seems that such symmetries are not of practical interest.

4. Estimation

4.1. Estimation based on orientations of parents and children: established methods

Suppose that we are given paired observations ([U1]1, [V1]2),…, ([Un]1, [Vn]2) with [Ui]1 in SO(3)/K1 and [Vi]2 in SO(3)/K2 for 1 ≤ in. It is assumed that the pairs ([U1]1, [V1]2),…, ([Un]1, [Vn]2) are observations of a pair ([U]1, [V]2) of random orientations that satisfy the orientation relationship (1[link]). The problem is that of estimating the unknown [A]1,2 (and R if it is not known). In the case K1 = K2 = C1 (i.e. no symmetry, so that variants do not occur), an explicit estimate of A was given by Mackenzie (1957[Mackenzie, J. K. (1957). Acta Cryst. 10, 61-62.]). In the general case, it follows from (1[link]) that [[{\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{1,2}] is close to [A]1,2 for i = 1,…, n. Then [[{\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{1,2}] can be considered as the `local' estimate of [A]1,2 given by the observed pair ([Ui]1, [Vi]2). Here `local' is used in the sense that the estimate uses only quantities with index i, which in many cases means that they are measured at the ith location.

Using this approach Nolze found that, for transformation from face-centred cubic to body-centred cubic lattices, the standard theoretical models did not provide a good fit to the experimental data that he was investigating

4.2. Estimation based on orientations of parents and children: embedding approach

The embedding approach uses a suitable embedding t2 : SO(3)/K2E. If ([U1]1, [V1]2),…, ([Un]1, [Vn]2) are observations on a pair ([U]1, [V]2) of random orientations that satisfy the orientation relationship (1[link]) then [[{\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{1,2}] is close to [A]1,2 for i = 1,…, n. Therefore, it is sensible to estimate R and [A]1,2 by [{\widehat {\bf R}}] and [[{\widehat {\bf A}}]_{1,2}], which are the R and [A]1,2 that minimize the squared distance [\sum \nolimits_{i=1}^{n} \| {\overline {\bf t}}_{2} ([{\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{1,2}) - {\overline {\bf t}}_{2} ([{\bf A}]_{1,2}) \|^{2}], or equivalently that maximize [S_{2} ({\bf R}, [{\bf A}]_{1,2} \semi \{ ([{\bf U}_{i}]_{1}, [{\bf V}_{i}]_{2}) \}_{i=1}^{n})], defined by

[\eqalignno{& S_{2} \left ( {\bf R}, [{\bf A}]_{1,2} \semi \left \{ ([{\bf U}_{i}]_{1}, [{\bf V}_{i}]_{2}) \right \}_{i=1}^{n} \right ) \cr & \quad = \sum \limits_{i=1}^{n} \left \langle {\overline {\bf t}}_{2} ([{\bf A}]_{1,2}), {\overline {\bf t}}_{2} ([{\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{1,2}) \right \rangle \cr & \quad = n \left \langle {\bf t}_{2} ([{\bf A}]_{2}), {{1} \over {n|K_{1}|}} \sum \limits_{i=1}^{n} \sum \limits_{{\bf H} \in K _{1}} {\bf t}_{2} ([{\bf H} {\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{2}) \right \rangle . &(7)}]

Thus, [A]1,2 can be taken as the sample mean of the images by [{\overline {\bf t}}_{2}] [defined by (6[link])] of [\{ [{\bf H} {\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{2} : {\bf H} \in K_{1}, i = 1, \ldots, n \}]. An alternative method of estimation uses the R and [A]1,2 that maximize [S_{\infty} ({\bf R}, [{\bf A}]_{1,2} \semi \{ ([{\bf U}_{i}]_{1}, [{\bf V}_{i}]_{2}) \}_{i = 1}^{n})], which is defined by

[\eqalignno{ & S_{\infty} ({\bf R}, [{\bf A}]_{1,2} \semi \{ ([{\bf U}_{i}]_{1}, [{\bf V}_{i}]_{2}) \}_{i = 1}^{n}) \cr & \quad = \sum_{i = 1}^{n} \max_{{\bf H} \in K _{1}} \, \langle {\bf t}_{2} ([{\bf H} {\bf A}]_{2}), {\bf t}_{2} ([{\bf U}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{2}) \rangle &(8)}]

[see Section 6.1 of Arnold et al. (2021[Arnold, R., Jupp, P. E. & Schaeben, H. (2021). Directional Statistics for Innovative Applications, edited by A. SenGupta & B. C. Arnold. Heidelberg: Springer.])].

Not only can a point estimate [[{\widehat {\bf A}}]_{1,2}] of [A]1,2 be obtained, it is also possible to get confidence regions for [A]1,2. Bootstrap confidence regions for [A]1,2 can be calculated by resampling the data as follows. For a suitable B and m, for b = 1,…, B, sample (with replacement) m pairs (mn) ([Ub1]1, [Vb1]2),…, ([Ubm]1, [Vbm]2) from ([U1]1, [V1]2),…, ([Un]1, [Vn]2). Denote the estimate of [A]1,2 based on ([Ub1]1, [Vb1]2),…, ([Ubm]1, [Vbm]2) by [[{\widehat {\bf A}}_{b}]_{1,2}]. Define the similarity measure

[\delta \left ( [{\bf A}]_{1,2}, [{\bf B}]_{1,2} \right ) = {\max \limits _{{\bf H}\in K_{1}}} \ \left \langle {\bf t}_{2} ([{\bf HA}]_{1,2}), {\bf t}_{2} ([{\bf B}]_{1,2}) \right \rangle . \eqno(9)]

For 0 < α < 1, define cα by

[c_{\alpha} = 100 \times (1 - \alpha)\% \ {\rm largest \ of} \ \left \{ \delta ([{\widehat {\bf A}}_{b}]_{1,2}, [{\widehat {\bf A}}]_{1,2}) \right \}_{b = 1}^{B} . \eqno(10)]

Then a 100 × (1 − α)% bootstrap confidence region for [A]1,2 is

[\left \{ [{\bf A}]_{1,2} : \delta ([{\bf A}]_{1,2}, [{\widehat {\bf A}}]_{1,2}) \ \gt \ c_{\alpha} \right \} . \eqno(11)]

For the corresponding test of H0 : [A]1,2 = [A0]1,2 (where [A0]1,2 is a specified theoretical [A]1,2), the p-value (i.e. the probability under H0 of another sample producing a value of [[{\widehat {\bf A}}]_{1,2}] at least as extreme as that observed) is

[p = B^{-1} \ {\rm No.\ of \ } \left \{ b : \delta ([{\widehat {\bf A}}_{b}]_{1,2}, [{\widehat {\bf A}}]_{1,2}) \ \gt \ \delta ([{\bf A}_{0}]_{1,2}, [{\widehat {\bf A}}]_{1,2}) \right \} . \eqno(12)]

The hypothesis H0 is rejected at significance level α if pα or, equivalently, if the estimated [[{\widehat {\bf A}}]_{1,2}] lies outside the 100 × (1 − α)% bootstrap confidence region (11[link]) for [A]1,2.

A flow chart for implementation of the techniques described in this subsection is given in Fig. 1[link].

[Figure 1]
Figure 1
A flow chart for (i) estimation of (R, [A1,2]) based on (U1, V1),…, (Un, Vn), (ii) confidence regions for [A1,2] and (iii) testing of [A1,2] = [A01,2]. An alternative method minimizes S2 in (7)[link] instead of S in (8)[link].

4.3. Estimation based on orientations of children alone: established methods

In some settings, no observations on the orientations of parent grains are available. Nevertheless, in many cases it is possible to estimate [A]1,2 from orientations [V1]2,…, [Vn]2 of child grains.

The estimation method of Humbert et al. (2015[Humbert, M., Germain, L., Gey, N. & Boucard, E. (2015). Acta Mater. 82, 137-144.]) starts from observed child crystallographic orientations [V1]2,…, [V3]2 that are the visible part of ([U1]1, [V1]2), ([U2]1, [V2]2), ([U3]1, [V3]2). These pairs are taken to obey (at least approximately) a form of the orientation relationship (1[link]), so that

[[{\bf V}_{r}]_{2} \simeq [{\bf R} {\bf U}_{r} {\bf A}_{r}]_{2}, \quad r = 1,2,3 \eqno(13)]

for some U1, U2, U3 and A1, A2, A3. It is assumed that U1U2 ≃ U3U and A1A2A3A for some U and A in SO(3) (assumptions that are reasonable if the orientations are measured at points near a triple junction and the local orientation relationships vary only slowly with position). It then follows from (13[link]) that for i, j in {1, 2, 3} it is possible to choose Hi, Hj in K1 and [{\widetilde {\bf H}}_{i}], [{\widetilde {\bf H}}_{j}] in K2 such that UiUHi, UjUHj and

[{\bf V}_{i} \simeq {\bf R} {\bf U} {\bf H}_{i} {\bf A} {\widetilde {\bf H}}_{i}, \quad {\bf V}_{j} \simeq {\bf R} {\bf U} {\bf H}_{j} {\bf A} {\widetilde {\bf H}}_{j} . \eqno(14)]

Then

[{\bf V}_{j}^{-1} {\bf V}_{i} \simeq {\widetilde {\bf H}}_{j}^{-1} {\bf A}^{-1} {\bf H}_{j}^{-1} {\bf H}_{i} {\bf A} {\widetilde {\bf H}}_{i} \eqno(15)]

or, equivalently,

[{\bf A} {\widetilde {\bf H}}_{j} {\bf V}_{i}^{-1} {\bf V}_{j} \simeq {\bf H}_{m} {\bf A} {\widetilde {\bf H}}_{i} , \eqno(16)]

where [{\bf H}_{m} = {\bf H}_{j}^{-1} {\bf H}_{ i}]. For given i and j, solutions for A of the equation obtained from (16[link]) by replacing approximate equality by exact equality are far from unique. On the other hand, for general V1, V2, V3 (outside some set of measure 0), the three corresponding equations as i, j run through distinct (unordered) pairs in {1, 2, 3} do have a unique solution. In view of this, Humbert et al. (2015[Humbert, M., Germain, L., Gey, N. & Boucard, E. (2015). Acta Mater. 82, 137-144.]) suggested estimating the rotation A by any element of SO(3) that minimizes

[\sum_{(i,j)} \min_{{\bf H}_{m}\in K_{1}} \, \min_{{\widetilde {\bf H}}_{i}, {\widetilde {\bf H}}_{j} \in K_{2}} \| {\bf A} {\widetilde {\bf H }}_{j} {\bf V}_{j}^{-1} {\bf V}_{i} - {\bf H}_{m} {\bf A} {\widetilde {\bf H}}_{i} \|^{2} , \eqno(17)]

the sum being over (i, j) = (1, 2), (1, 3), (2, 3) and the norm being the Frobenius norm [= Hilbert–Schmidt norm, defined by ∥B2 = trace(BBT)].

The estimation stage of the iterative reconstruction method of Nyyssönen et al. (2016[Nyyssönen, T., Isakov, M., Peura, P. & Kuokkala, V.-T. (2016). Metall. Mater. Trans. A, 47, 2587-2590.], 2018[Nyyssönen, T., Peura, P. & Kuokkala, V.-T. (2018). Metall. Mater. Trans. A, 49, 6426-6441.]) is also motivated by (16[link]). Given (i) orientations [Vi]2 and [Vj]2 of distinct variants and (ii) Hm in K1 and [{\widetilde {\bf H}}_{i}], [{\widetilde {\bf H}}_{j}] in K2, [A]1,2 is estimated by [[{\widehat {\bf A}}]_{1,2}], where [{\widehat {\bf A}}] is found by iterative solution of

[{\bf A} = {\bf H}_{m} {\bf A} {\widetilde {\bf H}}_{i} {\bf V}_{i}^{-1} {\bf V}_{j} {\widetilde {\bf H}}_{j}^{-1} . \eqno(18)]

Since Hm, [{\widetilde {\bf H}}_{i}] and [{\widetilde {\bf H}}_{j}] are not known, Nyyssönen and co-workers recommend averaging the above [{\widehat {\bf A}}] as Hm, [{\widetilde {\bf H}}_{i}], [{\widetilde {\bf H}}_{j}] run through K1, K2, K2. Note that averaging of rotations is not well defined.

4.4. Estimation based on orientations of children alone: embedding approach

Given an embedding t2 : SO(3)/K2E, define [\check{\bf t}_{2} : K_{2} \backslash SO(3)/K_{2} \to E] by

[{\check {\bf t}}_{2} ([{\bf V}]_{2,2}) = |K_{2}|^{-1} \sum \limits_{{\widetilde {\bf H}} \in K_{2}} {\bf t}_{2} ([{\widetilde {\bf H}} {\bf V}]_{2}) , \eqno(19)]

where [V]2,2 denotes the image of [V]2 in the double coset space [K_{2} \backslash SO(3)/K_{2}].

It follows from (15[link]) that

[{\check {\bf t}}_{2} ({\bf V}_{i}^{-1} {\bf V}_{j}) \simeq {\check {\bf t}}_{2} ({\bf A}^{-1} {\bf H}_{m} {\bf A}) \eqno(20)]

for some Hm in K1. It is therefore reasonable to estimate [A]1,2 as any element of [K_{1} \backslash SO(3)/K_{2}] that minimizes

[\sum_{(i,j)} \, \min_{{\bf H}_{m} \in K_{1}} \| {\check {\bf t}}_{2} ({\bf V}_{i}^{-1} {\bf V}_{j}) - {\check {\bf t}}_{2} ({\bf A}^{-1} {\bf H}_{m} {\bf A}) \|^{2} , \eqno(21)]

the sum being over ordered pairs (i, j) of distinct elements of {1,…, n}.

A flow chart for implementation of the technique described in this subsection is given in Fig. 2[link].

[Figure 2]
Figure 2
A flow chart for estimation of [A1,2] based on V1,…Vn.

5. Assessing the adequacy of a single orientation relationship

Whereas some data sets can be fitted well by the orientation relationship (1[link]), for others such a single orientation relationship is inadequate and several orientation relationships are required. For data sets in which a single orientation relationship does not suffice, it is possible to identify clusters of observations within each of which the data share a single orientation relationship.

5.1. Established methods

We are unaware of anything in the crystallographic literature that considers exactly this problem.

5.2. Embedding approach

One way of exploring the adequacy of (1[link]) is to use cluster analysis to divide the n pairs of observations ([U1]1, [V1]2),…, ([Un]1, [Vn]2) into subsets, each of which consists of pairs that give similar estimates of [A]1,2. Cluster analysis is described in detail by Everitt et al. (2011[Everitt, B., Landau, S., Leese, M. & Stahl, D. (2011). Cluster Analysis, 5th ed. Wiley Series in Probability and Statistics. Chichester: John Wiley and Sons.]), who give several algorithms. We find it convenient to use the following simple divisive algorithm. For an estimate [[{\widehat {\bf A}}]_{1,2}] found using the method of Section 4.2[link] and for i = 1,…, n, define di by

[d_{i} = \delta ([{\bf U}_{i}^{-1} {\bf V}_{i}]_{1,2}, [{\widehat {\bf A}}]_{1,2}) , \eqno(22)]

where δ is defined in (9[link]). The maximum possible value of di is

[\rho^{2} = \left \langle {\bf t}_{2} ([{\bf U}]_{2}), {\bf t}_{2} ([{ {\bf U}}]_{2}) \right \rangle \eqno(23)]

[which does not depend on U in SO(3)]. A value of di near ρ2 indicates that the `local estimate' [[{\bf U}_{i}^{-1} {\bf V}_{i}]_{1,2}] of [A]1,2 is close to the `global estimate' [[{\widehat {\bf A}}]_{1,2}]. The values of ρ2 corresponding to various symmetry groups are given in Table 2[link]. If the single orientation relationship [[{\widehat {\bf A}}]_{1,2}] gives a good fit to the data ([U1]1, [V1]2),…, ([Un]1, [Vn]2) then d1,…, dn are all large. Placing all observations with di close to ρ2 into a cluster, re-estimate [[{\widehat {\bf A}}]_{1,2}] for that cluster. Repeat this process on the remaining observations, thus grouping them sequentially into clusters having similar local orientation relationships. If only one cluster is found then the single orientation relationship [[{\widehat {\bf A}}]_{1,2}] will describe the data well. Whereas the clustering of locations used by Johnstone et al. (2020[Johnstone, D. N., Martineau, B. H., Crout, P., Midgley, P. A. & Eggeman, A. S. (2020). J. Appl. Cryst. 53, 1293-1298.]) and Ostapovich & Trusov (2021[Ostapovich, K. V. & Trusov, P. (2021). Crystals, 11, 336.]) in the construction of crystal orientation maps takes place on SO(3)/K and is based on misorientation angles, the clustering used here takes place on [K_{1} \backslash SO(3)/K_{2}] and is based on the di of (22[link]). The clustering of locations used by Gomes de Araujo et al. (2021[Gomes de Araujo, E., Pirgazi, H., Sanjari, M., Mohammadi, M. & Kestens, L. A. I. (2021). J. Appl. Cryst. 54, 569-579.]) in the reconstruction of parent microstructure is based on misorientation angles between adjacent grains.

Table 2
Values of the squared radius ρ2

ρ2 is defined in equation (23[link]).

Group, K ρ2
C1 3
C2 5/3
Cr (r ≥ 3)  
r odd 1 + 21−rr2
r even [1 + r^{2} 2^{1-r} \{ 1 + 2^{-1} {\textstyle {r\choose r/2}} \} - r^{2}/(r+1)]
D2 2
Dr (r ≥ 3)  
r odd 21−rr2
r even [r^{2} 2^{1-r} \{ 1 + 2^{-1} {\textstyle {r\choose r/2}} \} - r^{2}/(r+1)]
T 32/9
O 6/5

A flow chart for implementation of the technique described in this subsection is given in Fig. 3[link].

[Figure 3]
Figure 3
A flow chart for assessing the adequacy of a single orientation relationship (OR).

6. Common parentage

In some contexts, such as locating the boundaries of parent grains, it is useful to be able to assess whether or not a set of child crystals are from the same parent. Observed elements [V1]2,…, [Vn]2 of SO(3)/K2 can be considered as arising from some common parent if and only if they are close to variants corresponding to some element of SO(3)/K1.

6.1. Established methods

The method of Gey & Humbert (2003[Gey, N. & Humbert, M. (2003). J. Mater. Sci. 38, 1289-1294.]) and Karthikeyan et al. (2006[Karthikeyan, T., Saroja, S. & Vijayalakshmi, M. (2006). Scr. Mater. 55, 771-774.]) assesses [V1]2,…, [Vn]2 to be from the same parent if the observed misorientation angles between [Vi]2 and [Vj]2 for 1 ≤ i, jn are within some given threshold of the theoretical misorientation angles. These observed misorientation angles are

[\arccos \,[ \{ \max {\rm tr} ({\widetilde {\bf V}}_{i}^{-1} {\widetilde {\bf V}}_{j}) - 1 \}/2] , \eqno(24)]

where the maximum is over [{\widetilde {\bf V}}_{i}] in SO(3) with [[{\widetilde {\bf V}}_{i}]_{1,2}] = [[{\bf V}_{i}]_{1,2}] for i = 1,…, n.

Cayron et al. (2006[Cayron, C., Artaud, B. & Briottet, L. (2006). Mater. Charact. 57, 386-401.]) considered the problem of finding maximal sets [V1]2,…, [Vn]2 of children from a common parent. Their method is based on considering triples [Vi]2, [Vj]2 and [Vk]2 to be from the same parent if they are coherent, i.e. there are operators Oij, Oik, Ojk taking [Vi]2 to [Vj]2, [Vi]2 to [Vk]2, [Vj]2 to [Vk]2, respectively, (at least approximately) and with Oik as some value of the composition of Ojk with Oij. Such a `nucleus' triple is `grown' by adding progressively further grains, to obtain a set of grains in which each triple is coherent. This is continued until a maximal such set is obtained.

6.2. Embedding approach

The embedding approach uses a suitable embedding t2 : SO(3)/K2E. Divisive cluster analysis is then applied to t2([V1]2),…, t2([Vn]2) using

[2 \left ( \rho^{2} - \langle {\bf t}_{2} ([{\bf V}_{i}]_{2}), {\bf t}_{2}([{ {\bf V}}_{c}]_{2}) \rangle \right ) \eqno(25)]

as a measure of squared distance [see (5[link])] between [Vi]2 and the putative centre [Vc]2 of a cluster. The [Vc]2 for a cluster is chosen to minimize the sum of squared distances from that centre to the members of the cluster (i.e. it is set to be the sample mean defined in Section 3.1[link]). If only one cluster is found then it is considered that [V1]2,…, [Vn]2 arise from the same parent.

A flow chart for implementation of the technique described in this subsection is given in Fig. 4[link].

[Figure 4]
Figure 4
A flow chart for assessing common parentage.

7. Reconstruction

In some contexts the orientations [V1]2,…, [Vn]2 of child crystals are observed but the parent crystals are not visible. The reconstruction problem is that of estimating the orientation [U]1 of the parent crystal, assuming a given orientational relationship of the form (1[link]) with R and [A]1,2 known. Thus, the problem is to estimate [U]1 such that [Vi]2 is close to [RWiA]2 (i = 1,…, n) for some W1,…, Wn with [W1]1 = ⋯ = [Wn]1 = [U]1.

7.1. Established methods

The reconstruction method introduced by Humbert et al. (1994[Humbert, M., Moustahfid, H., Wagner, F. & Philippe, M. J. (1994). Scr. Metall. Mater. 30, 377-382.]) is based on the fact that, if the pairs ([U]1, [Vi]2) and ([U]1, [Vj]2) each satisfy the orientation relationship (1[link]) (at least approximately) then (14[link]) holds. It follows that

[[{\bf U}]_{1} \simeq [{\bf R}^{-1} {\bf V}_{i} {\widetilde {\bf H}}_{i}^{-1} {\bf A}^{-1}]_{1} \simeq [{\bf R}^{-1} {\bf V}_{j} {\widetilde {\bf H}}_{j}^{-1} {\bf A}^{-1}]_{1} . \eqno(26)]

If R and A are known then applying this method to the three pairs (i, j) obtained from three variants yields a unique value for [U]1. If observations on n variants are available then Humbert & Gey (2002[Humbert, M. & Gey, N. (2002). J. Appl. Cryst. 35, 401-405.]) and Gey & Humbert (2003[Gey, N. & Humbert, M. (2003). J. Mater. Sci. 38, 1289-1294.]) recommend that the values of [U]1 given by the ((n) || 3) triples of variants be averaged. Note that averaging of rotations is not well defined.

7.2. Embedding approach

One reconstruction method using the embedding approach starts with a suitable embedding t2 of SO(3)/K2 and known (or estimated) values of R and [A]2. It then calculates W1,…, Wn by the [{\widehat {\bf W}}_{1}, \ldots, {\widehat {\bf W}}_{n}] that maximize [\sum \nolimits_{i = 1}^{n} \langle {\bf t}_{2} ([{\bf R} {\bf W}_{i} {\bf A}]_{2}), {\bf t}_{2} ([{\bf V}_{i}]_{2}) \rangle], i.e. that maximize

[\sum_{i = 1}^{n} \left \langle {\bf t}_{2} ([{\bf W}_{i}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{2}), {\bf t}_{2} ([{\bf A}]_{2}) \right \rangle \eqno(27)]

over W1,…, Wn satisfying [W1]1 = ⋯ = [Wn]1, and then estimates [U]1 by [[{\widehat {\bf W}}_{1}]_{1}]. In other words, the estimate of [U]1 is the element of SO(3)/K1 corresponding to the U in SO(3) that maximizes

[\sum_{i = 1}^{n} \max_{{\bf H}_{i} \in K_{1}} \ \left \langle {\bf t}_{2} ([{\bf H}_{i}^{-1} {\bf U}^{-1} {\bf R}^{-1} {\bf V}_{i}]_{2}), {\bf t}_{2} ([{\bf A}]_{2}) \right \rangle , \eqno(28)]

where A is any representative in SO(3) of [A]1,2. In (28[link]) it can be assumed without loss of generality that H1 = I3, the identity.

Locating the maximum of (28[link]) involves maximization over SO(3) ×K1n, so it may be useful to consider an alternative estimator which is easier to compute. The left-hand approximate equality in (14[link]) gives

[{\bf R}^{-1} {\bf V}_{i} {\widetilde {\bf H}}_{i}^{-1} {\bf A}^{-1} \simeq {\bf U} {\bf H}_{i} , \eqno(29)]

and so

[{\bf t}_{1} ([{\bf R}^{-1} {\bf V}_{i} {\widetilde {\bf H}}_{i}^{-1} {\bf A}^{-1}]_{1}) \simeq {\bf t}_{1} ([{\bf U}]_{1}) \eqno(30)]

for any embedding t1 of SO(3)/K1. Define [{\overline {\bf A}}^{-1}] by

[{\overline {\bf A}}^{-1} = |K_{2}|^{-1} \sum \limits_{{\tilde {\bf H}} \in K_{2}} {\tilde {\bf H}} {\bf A}^{-1} . \eqno(31)]

Then (30[link]) gives

[{\bf t}_{1} ([{\bf U}]_{1}) \simeq {\bf t}_{1} ([{\bf R}^{-1} {\bf V}_{i} {\overline {\bf A}}^{-1}]_{1}) . \eqno(32)]

Thus it is reasonable to estimate [U]1 by the sample mean (based on t1) of [[{\bf R}^{-1} {\bf V}_{1} {\overline {\bf A}}^{-1}]_{1}, \ldots, [{\bf R}^{-1} {\bf V}_{n} {\overline {\bf A}}^{-1}]_{1}], i.e. the [U]1 that maximizes

[\left \langle {\bf t}_{1} ([{\bf U}]_{1}), \sum \limits_{i = 1}^{n} {\bf t}_{1} ([{\bf R}^{-1} {\bf V}_{i} {\overline {\bf A}}^{-1}]_{1}) \right \rangle . \eqno(33)]

A flow chart for implementation of the technique described in this subsection is given in Fig. 5[link].

[Figure 5]
Figure 5
A flow chart for reconstruction of a parent.

8. Practical applications

8.1. Spatially varying orientation relationship: austenite–martensite transformation

The data set considered by Wendler et al. (2017[Wendler, M., Ullrich, C., Hauser, M., Krüger, L., Volkova, O., Weiß, A. & Mola, J. (2017). Acta Mater. 133, 346-355.]) consists of 9707 pairs of orientations ([Ui]1, [Vi]2), i = 1,…, 9707, of austenite and martensite, respectively. For each pair the orientations are measured at sites that are close but separated by a boundary between grains. The locations of the pairs on the surface of a steel sample are shown in Fig. 6[link]. Fig. 7[link] displays these orientations on two stereonets (stereographic projections), one for the (face-centred cubic) austenite phase and the other for the (body-centred cubic) martensite phase. The symmetry groups K1 and K2 are both equal to the octahedral group O. In Fig. 7[link], each disc represents the upper half of the unit sphere, and each orientation [Ui]1 or [Vi]2 is represented by the three points at which the three (unordered) orthogonal axes determined by [Ui]1 or [Vi]2 intersect the upper half of the unit sphere. In each diagram there is considerable variation, which is due partly to the differing crystal orientations [Ui]1 of the austenite phase prior to the transformation. However, there is a degree of clustering present, which can be explained by differing orientation relationships [A]1,2 among the pairs of observations.

[Figure 6]
Figure 6
Locations of sites coloured by cluster.
[Figure 7]
Figure 7
(a) A stereonet diagram showing the n = 9707 orientations [Ui]1, i = 1,…, n, in the austenite phase from Wendler et al. (2017[Wendler, M., Ullrich, C., Hauser, M., Krüger, L., Volkova, O., Weiß, A. & Mola, J. (2017). Acta Mater. 133, 346-355.]). (b) The corresponding martensite orientations [Vi]2, i = 1,…, n.

A cluster analysis, as outlined in Section 5[link], using S from (8[link]) [rather than S2 from (7[link])] was carried out. It reveals six clusters in [O \backslash SO(3)/O] of values [[{\widehat {\bf A}}_{1}]_{1,2}, \ldots, [{\widehat {\bf A}}_{6}]_{1,2}] of [[{\widehat {\bf A}}]_{1,2}]. Most observations belong to cluster 1 (n1 = 7214) or cluster 2 (n2 = 1844). Note that this is not a spatial clustering but rather a clustering of common orientation relationships. The locations of the clusters are shown by the colouring of the symbols in Fig. 6[link], and all six clusters occur at locations spread all over the surface of the steel sample.

Fig. 8[link](a) shows the estimated orientations [[{\widehat {\bf A}}_{c}]_{1,2}] for each of the six clusters, c = 1,…, 6. These same orientations are shown together with the [[{\bf U}_{i}^{\rm T} {\bf V}_{i}]_{1,2}] in Fig. 8[link](b). In drawing diagrams of this type we have had to make the choice to display each [[{\widehat {\bf A}}_{c}]_{1,2} \in O \backslash SO(3)/O] as an element [[{\widehat {\bf A}}_{c}]_{2}] of SO(3)/O, (arbitrarily) selecting the representation closest to the first observation [[{\bf U}_{c1}^{\rm T} {\bf V}_{c1}]_{2}] in each cluster.

[Figure 8]
Figure 8
(a) Orientation relationships [[{\widehat {\bf A}}_{c}]_{1,2}] in the six fitted clusters c = 1,…, 6. The orientations are plotted as members of SO(3)/K2 by (arbitrarily) selecting the element of SO(3)/O for which [[{\widehat {\bf A}}_{c}]_{2}] is closest to the first observation in each cluster: [[{\bf U}_{c1}^{\rm T} {\bf V}_{c1}]_{2}]. (b) Under the same convention as in panel (a), the data [[{\bf U}_{i}^{\rm T} {\bf V}_{i}]_{1,2}] are shown, coloured by cluster. The fitted orientations [[{\widehat {\bf A}}]_{1,2}] are shown as red filled circles.

Fig. 7[link](b) appears to be a left–right reflection of the centre plot [labelled `{100}') in Fig. 5(d) of Wendler et al. (2017[Wendler, M., Ullrich, C., Hauser, M., Krüger, L., Volkova, O., Weiß, A. & Mola, J. (2017). Acta Mater. 133, 346-355.])]. (That the figures are not identical may be due to a difference in convention regarding Euler angles.) The colour codings in the two figures are completely different; that in Fig. 7[link](b) is derived from having taken full account of all ambiguities in rotation.

8.2. Testing whether [A]1,2 has a given value

In the analysis in Section 8.1[link] the estimated orientation relationship [[{\widehat {\bf A}}]_{1,2}] for cluster 2 is very close to the identity I3. We can test the hypothesis H0 : [A]1,2 = [I3]1,2 by forming a bootstrap confidence region for [[{\widehat {\bf A}}]_{1,2}] using the methods of Section 4[link]. For demonstration purposes, we took a random subset of 50 observations from cluster 2. We then generated B = 100 bootstrap samples from this subset, and re-estimated [A]1,2 in each case. The distribution of values of [\{ \delta ([{\widehat {\bf A}}_{b}]_{1,2}, [{\widehat {\bf A}}]_{1,2}) \}_{b = 1}^{B}] is shown in Fig. 9[link](a), and the similarity measure δ0 for the hypothesized value falls well outside the distribution, a convincing rejection of H0. Fig. 9[link](b) displays the original estimate [[{\widehat {\bf A}}]_{1,2}] and the 100 replicate estimates, all lying well away from the hypothesized value [I3]1,2.

[Figure 9]
Figure 9
(a) A histogram of bootstrap [\{ \delta ([{\widehat {\bf A}}_{b}]_{1,2}, [{\widehat {\bf A}}]_{1,2 }) \}_{b = 1}^{B}] values. The limit cα of the 95% confidence region is indicated by the red dashed line. The estimated δ value for the hypothesized value is the vertical solid blue line on the left-hand side. (b) Locations of bootstrap estimates on an enlarged stereonet. The estimate [[{\widehat {\bf A}}]_{1,2}] is shown as the large open circle, the hypothesized value [I3]1,2 as a filled circle at the origin, and the bootstrap estimates as small open circles. The large cross in (b) is the same size as the crosses at the centres of the stereonets in Figs. 7[link] and 8[link].

9. Software

The analyses in Section 8[link] were performed using the freely available statistical software R (https://www.R-project.org/). A general and flexible MATLAB tool kit for the analysis of crystallographic data is provided by the open-source crystallographic toolbox MTEX (Bachmann et al., 2010[Bachmann, F., Hielscher, R. & Schaeben, H. (2010). Solid State Phenom. 160, 63-68. ]).

10. Conclusions

The intrinsic symmetries of crystal structures provide a challenge to statistical analysis due to the ambiguities of their representations. The challenge becomes greater in a setting where the orientations of crystals with different symmetries are to be compared.

In this paper we have applied the embedding approach from directional statistics to the representation of crystallographic orientation data. This approach enables us to reformulate standard problems of estimation and inference in the crystallographic setting, eliminating the ambiguities which arise from the crystallographic symmetries, while retaining the genuine multiplicity of crystallographic variants.

Although the embedded objects we work with may be unfamiliar, they are straightforward to implement in software, and should provide a practical tool for researchers seeking to characterize crystallographic structures and their transformations.

APPENDIX A

The number of variants

In the context of Section 2.3[link], elements [UHi]1 and [UHj]1 of SO(3)/K1 give rise to the same element of SO(3)/K2 under orientation relationship (1[link]) if and only if [RUHiA]2 = [RUHjA]2. Since

[\eqalignno{& [{\bf R} {\bf U} {\bf H}_{i} {\bf A}]_{2} = [{\bf R} {\bf U} {\bf H}_{j} {\bf A}]_{2} \Leftrightarrow ({\bf R} {\bf U} {\bf H}_{i} {\bf A})^{-1} ({\bf R} {\bf U} {\bf H}_{j} {\bf A}) \in K_{2} \cr & \Leftrightarrow {\bf A}^{-1} {\bf H}_{i}^{-1} {\bf H}_{j} {\bf A} \in K_{2} \Leftrightarrow {\bf H}_{i}^{-1} {\bf H}_{j} \in K_{1} \cap ({\bf A} K_{2} {\bf A}^{-1}) = K_{\bf A} , \cr &&(34)}]

the number of variants is s = |K1|/|KA|.

Proposition 1

There is a subset [{\cal B}] of SO(3) that has measure 0 and such that if A is outside [{\cal B}] then s = |K1|.

Proof

For given Hi in K1 and [{\tilde {\bf H}}_{j}] in K2, define the set [{\cal B}_{ij}] by

[{\cal B}_{ij} = \{ {\bf W} \in SO(3) : {\bf H}_{i} {\bf W} = {\bf W} {\tilde {\bf H}}_{j} \} . \eqno(35)]

Then either (i) [{\cal B}_{ij} = SO(3)] or (ii) [{\cal B}_{ij}] is contained in some proper subspace of the vector space of 3 × 3 matrices. In case (i), [{\bf H}_{i} {\bf W}] = [{\bf W} {\tilde {\bf H}}_{j}] for all W in SO(3), so that [{\bf H}_{i}] = [{\tilde {\bf H}}_{j}] and therefore HiW = WHi for all W (including W = I3), implying that Hi = I3. In case (ii), there is some C in SO(3) such that tr(CW) = 0 for all W in [{\cal B}_{ij}]. Thus [{\cal B}_{ij}] has measure 0. Now let Hi and [{\tilde {\bf H}}_{j}] run through in K1 and K2, respectively and define [{\cal B}] as

[{\cal B} = \cup_{{\cal B}_{ij}\neq SO(3)} {\cal B}_{ij} . \eqno(36)]

Since this is a finite union, [{\cal B}] has measure 0. If HKA then H = Hi and H = [{\bf A} {\tilde {\bf H}}_{j} {\bf A}^{-1}] for some Hi in K1 and [{\tilde {\bf H}}_{j}] in K2, so that [{\bf A} \in {\cal B}_{ij}]. Thus, if [{\bf A} \,\notin\, {\cal B}] then H = I3, so that |KA| = 1, and therefore s = |K1|.

It follows from Proposition 1[link] that if A is chosen randomly from some continuous distribution then, with probability 1, s = |K1|.

APPENDIX B

Some embeddings of SO(3)/K

The embedding approach is based on embeddings, i.e. well defined equivariant one-to-one functions tK : SO(3)/KE, where E is an inner-product space on which SO(3) acts, such that tK([U]) has expectation 0 if [U] is uniformly distributed on SO(3)/K. Here, we focus on a simple choice of embedding tK of SO(3)/K into an appropriate space of symmetric arrays, where K is one of the crystallographic groups C1, C2, C3, C4, D2, D6 and O or the tetrahedral group T (in Schoenflies notation). Appropriate embeddings tK for general point groups K of the first kind are considered by Arnold & Jupp (2019[Arnold, R. & Jupp, P. E. (2019). Applied Directional Statistics: Modern Methods and Case Studies, edited by C. Ley & T. Verdebout, pp. 41-60. London: Chapman & Hall/CRC.]) and Arnold et al. (2018[Arnold, R., Jupp, P. E. & Schaeben, H. (2018). J. Multivariate Anal. 165, 73-85.], 2021[Arnold, R., Jupp, P. E. & Schaeben, H. (2021). Directional Statistics for Innovative Applications, edited by A. SenGupta & B. C. Arnold. Heidelberg: Springer.]).

It is convenient to describe the orientation of a crystal by a frame, meaning a set of vectors or axes that are fixed in the crystal. The presence of symmetry under group K means that such a frame is equivalent to one obtained from it by a rotation in K. The equivalence classes are known as K-frames. For K = C1 the frame is taken to be (u1, u2, u3) with u1, u2, u3 orthogonal unit vectors; for K = C2 the frame is (u0, ±u1) with u0 a unit vector and ±u1 an axis orthogonal to u0; for K = D2 it is a pair of orthogonal axes (±u1, ±u2); for K = Cr with r ≥ 3 or K = D6 with r ≥ 3, the vectors (u1,…, ur) are unit normals to the sides of a regular r-gon or hexagon; for O and T, the axes (±u1, ±u2, ±u3) and (±u1, ±u2, ±u3, ±u4) are unit normals to the sides of the cubic crystal and tetrahedral crystal, respectively. The K-frames will be denoted by square brackets, e.g. for K = Cr [u1,…, ur] denotes the K-frame arising from (u1,…, ur). The frames used are listed in Table 3[link]. Note that the frames are specified entirely by vectors and axes; they do not depend on any coordinate system. For notational simplicity, we shall sometimes denote a K-frame by [U].

Table 3
Symmetry groups and frames

The ui are unit vectors. |K| is the number of elements in K.

Group, K Name |K| Frame Notes
C1 Trivial 1 (u1, u2, u3) u1, u2, u3 orthonormal, u3 = u1 × u2
C2 Cyclic 2 (u0, ± u1) u0, u1 orthonormal
Cr (r ≥ 3) Cyclic r (u1,…, ur) u1,…, ur coplanar, known up to cyclic order, [{\bf u}_{i}^{\rm T} {\bf u}_{i-1}] = cos(2π/r) for i = 2,…, r
D2 Dihedral 4 u1, ± u2) Orthogonal axes
D6 Dihedral 12 (u1,…, u6) u1,…, u6 coplanar, known up to cyclic order and reversal, [{\bf u}_{i}^{\rm T} {\bf u}_{i-1}] = cos(π/3) for i = 2,…, 6
T Tetrahedral 12 {u1,…, u4} [{\bf u}_{i}^{\rm T} {\bf u}_{j}] = −1/3 for ij
O Octahedral = cubic 24 u1, ± u2, ± u3} Orthogonal axes

The embeddings tK are given explicitly in Table 1[link]. They are based on symmetric r-way arrays, ⊗rui, which can be thought of as rth powers of vectors. In mathematical terms, they are coordinate representations of r-fold tensor products. For vectors u1,…, ur with ui = (ui,1, ui,2, ui,3)T for i = 1,…, r, the r-way array ⊗rui has (j1,…, jr)th entry

[(\otimes^{r} {\bf u}_{i})_{j_{1}, \ldots, j_{r}} = \prod_{k = 1}^{r} u _{i,j_{k}}, \ 1 \leq j_{1}, \ldots, j_{r} \leq 3 , \eqno(37)]

e.g. [\otimes^{2} {\bf u}_{i} = {\bf u}_{i} {\bf u}_{i}^{\rm T}]. Some of the tK also involve the arrays Nr for r even. These symmetric r tensors are defined by

[\langle {\bf N}_{r}, \otimes^{r} {\bf v} \rangle = \| {\bf v} \|^{r} \quad {\bf v} \in {\bb R}^{3} , \eqno(38)]

where v denotes the length of v. In coordinate terms they have entries

[({\bf N}_{r})_{j_{1}, \ldots, j_{r}} = {{1} \over {r!}} \sum \limits_{\sigma \in \Sigma_{r}} \prod_{i = 1}^{r/2} \delta_{\sigma (j_{2i-1}) \sigma(j_{2i})} , \eqno(39)]

where Σr is the group of permutations of {1,…, r} and δij = 1 for i = j, δij = 0 otherwise. Thus, e.g. [({\bf N}_{4})_{j_{1}, j_{2}, j_{3}, j_{4}}] = ([\delta_{j_{1} j_{2}} \delta_{j_{3} j_{4}}] + [\delta_{j_{1} j_{3}} \delta_{j_{2} j_{4}}] + [\delta_{j_{1} j_{4}} \delta_{j_{2} j_{3}}])/3. Some of the embeddings involve subtraction of constant terms to ensure that E{tK([U])} = 0 under uniformity. In the case of Cr, the embeddings contain oriented vectors u0 to specify the orientation of the directed axis of symmetry.

For our purposes it is not necessary to construct the representations tK([U]) explicitly; it is sufficient to compute inner products 〈tK([U]), tK([V])〉 as required, where 〈·, ·〉 is the standard inner product on the relevant space of symmetric arrays. The inner products are listed in Table 4[link].

Table 4
Inner products of transforms of symmetric frames

For Cr with r ≥ 3, u0 = {sin(2π/r)}−1 u1 × u2. For D2, u3 = ±u1 × u2.

Group, K Inner product
C1 [\langle {\bf t}_{C_{1}} ({\bf u}_{1}, {\bf u}_{2}, {\bf u}_{3}), {\bf t}_{C_{1}} ({\bf v}_{1}, {\bf v}_{2}, {\bf v}_{3}) \rangle = {\bf u}_{1}^{\rm T} {\bf v}_{1} + {\bf u}_{2}^{\rm T} {\bf v}_{2} + {\bf u}_{3}^{\rm T} {\bf v}_{3}]
C2 [\langle {\bf t}_{C_{2}} ({\bf u}_{0}, \pm {\bf u}_{1}), {\bf t}_{C_{2}} ({\bf v}_{0}, \pm {\bf v}_{1}) \rangle = {\bf u}_{0}^{\rm T} {\bf v}_{0} + ({\bf u}_{1}^{\rm T} {\bf v}_{1})^{2} - 1/3]
C3 [\langle {\bf t}_{C_{3}} ([{\bf u}_{1}, \ldots, {\bf u}_{3}]), {\bf t}_{C_{3}} ([{\bf v}_{1}, \ldots, {\bf v}_{3}]) \rangle = {\bf u}_{0}^{\rm T} {\bf v}_{0} + \sum_{i = 1}^{3} \sum_{j = 1}^{3} ({\bf u}_{i}^{\rm T} {\bf v}_{j})^{3}]
C4 [\langle {\bf t}_{C_{4}} ([{\bf u}_{1}, \ldots, {\bf u}_{4}]), {\bf t}_{C_{4}} ([{\bf v}_{1}, \ldots, {\bf v}_{4}]) \rangle = {\bf u}_{0}^{\rm T} {\bf v}_{0} + \sum_{i = 1}^{4} \sum_{j = 1}^{4} ({\bf u}_{i}^{\rm T} {\bf v}_{j})^{4} - 16/5]
D2 [\langle {\bf t}_{D_{2}} (\pm {\bf u}_{1}, \pm {\bf u}_{2}), {\bf t}_{D_{2}} (\pm {\bf v}_{1}, \pm {\bf v}_{2}) \rangle = ({\bf u}_{1}^{\rm T} {\bf v}_{1})^{2} + ({\bf u}_{2}^{\rm T} {\bf v}_{2})^{2} + ({\bf u}_{3}^{\rm T} {\bf v}_{3})^{2} - 1]
D4 [\langle {\bf t}_{D_{4}} ([{\bf u}_{1}, \ldots, {\bf u}_{4}]), {\bf t}_{D_{4}} ([{\bf v}_{1}, \ldots, {\bf v}_{4}]) \rangle = \sum_{i = 1}^{4} \sum_{j = 1}^{4} ({\bf u}_{i}^{\rm T} {\bf v}_{j})^{4} - 16/5]
D6 [\langle {\bf t}_{D_{6}} ([{\bf u}_{1}, \ldots, {\bf u}_{6}]), {\bf t}_{D_{6}} ([{\bf v}_{1}, \ldots, {\bf v}_{6}]) \rangle = \sum_{i = 1}^{6} \sum_{j = 1}^{6} ({\bf u}_{i}^{\rm T} {\bf v}_{j})^{6} - 36/7]
T [\langle {\bf t}_{T} ( \{ {\bf u}_{1}, {\bf u}_{2}, {\bf u}_{3}, {\bf u}_{4} \} ), {\bf t}_{T} (\{ {\bf v}_{1}, {\bf v}_{2}, {\bf v}_{3}, {\bf v}_{4} \}) \rangle = \sum_{i = 1}^{4} \sum_{j = 1}^{4} ({\bf u}_{i}^{\rm T} {\bf v}_{j})^{3}]
O [\langle {\bf t}_{O} (\{ \pm {\bf u}_{1}, \pm {\bf u}_{2}, \pm {\bf u}_{3} \}), {\bf t}_{O} (\{ \pm {\bf v}_{1}, \pm {\bf v}_{2}, \pm {\bf v}_{3} \}) \rangle = \sum_{i = 1}^{3} \sum_{j = 1}^{3} ({\bf u}_{i}^{\rm T} {\bf v}_{j})^{4} - 9/5]

Some much more general classes of embeddings than the tK given in Table 1[link] are considered by Hielscher & Lippert (2021[Hielscher, R. & Lippert, L. (2021). J. Multivariate Anal. 185, 104764.]) and Arnold et al. (2021[Arnold, R., Jupp, P. E. & Schaeben, H. (2021). Directional Statistics for Innovative Applications, edited by A. SenGupta & B. C. Arnold. Heidelberg: Springer.]).

In general, it is simpler and computationally faster to use inner products than to use misorientation angles. This is illustrated neatly in the cubic case. Consider [U] and [V] in SO(3)/O with u1, u2, u3 and v1, v2, v3 as the columns of U and V, respectively. Then

[\langle {\bf t}_O ([{\bf U}]), {\bf t}_O ([{\bf V}]) \rangle = \sum \limits_{i = 1}^3 \sum \limits_{j = 1}^3 ({\bf u}_i^{\rm T} {\bf v}_j)^4 - 9/5 , \eqno(40)]

whereas the misorientation angle is

[\min {\rm arccos} \{ (\epsilon_1 {\bf u}_1^{\rm T} {\bf v}_j + \epsilon_2 {\bf u}_2^{\rm T} {\bf v}_k + \epsilon_3 {\bf u}_3^{\rm T} {\bf v}_{\ell} - 1)/2 \} , \eqno(41)]

the minimum being over all ε1, ε2, ε3 = ±1 and all permutations j, k, ℓ of 1, 2, 3. Evaluating (41[link]) involves about six times as many operations as evaluating (40[link]).

Acknowledgements

We gratefully acknowledge invaluable discussions with and advice from Christiane Ullrich, Institute of Materials Science, and Javad Mola, Institute of Iron and Steel Technology, both at TU Bergakademie Freiberg, Germany, and also from Nathalie Gey, Laboratoire d'Etude des Microstructures et de Mécanique des Matériaux, UMR CNRS 7239, Université de Lorraine, Metz, France. Moreover, we are very thankful to Christiane Ullrich for providing the data used in Section 8.1. This paper is partly an extended version of oral contributions to the International Conference THERMEC'2018 held in Paris, France, 8–13 July 2018, and the International Materials Research Congress (IMRC) held in Cancún, Mexico, 19–24 August 2018.

References

First citationAbbasi, M., Kim, D.-I., Nelson, T. W. & Abbasi, M. (2014). Mater. Charact. 95, 219–231.  Web of Science CrossRef CAS Google Scholar
First citationAbbasi, M., Nelson, T. W., Sorensen, C. D. & Wei, L. (2012). Mater. Charact. 66, 1–8.  Web of Science CrossRef CAS Google Scholar
First citationArnold, R. & Jupp, P. E. (2019). Applied Directional Statistics: Modern Methods and Case Studies, edited by C. Ley & T. Verdebout, pp. 41–60. London: Chapman & Hall/CRC.  Google Scholar
First citationArnold, R., Jupp, P. E. & Schaeben, H. (2018). J. Multivariate Anal. 165, 73–85.  Web of Science CrossRef Google Scholar
First citationArnold, R., Jupp, P. E. & Schaeben, H. (2021). Directional Statistics for Innovative Applications, edited by A. SenGupta & B. C. Arnold. Heidelberg: Springer.  Google Scholar
First citationBachmann, F., Hielscher, R. & Schaeben, H. (2010). Solid State Phenom. 160, 63–68.   CrossRef CAS Google Scholar
First citationCayron, C. (2006). Acta Cryst. A62, 21–40.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationCayron, C. (2016). Acta Mater. 111, 417–441.  Web of Science CrossRef CAS Google Scholar
First citationCayron, C. (2019). Acta Cryst. A75, 411–437.  Web of Science CrossRef IUCr Journals Google Scholar
First citationCayron, C., Artaud, B. & Briottet, L. (2006). Mater. Charact. 57, 386–401.  Web of Science CrossRef CAS Google Scholar
First citationEveritt, B., Landau, S., Leese, M. & Stahl, D. (2011). Cluster Analysis, 5th ed. Wiley Series in Probability and Statistics. Chichester: John Wiley and Sons.  Google Scholar
First citationGey, N. & Humbert, M. (2003). J. Mater. Sci. 38, 1289–1294.  Web of Science CrossRef CAS Google Scholar
First citationGomes de Araujo, E., Pirgazi, H., Sanjari, M., Mohammadi, M. & Kestens, L. A. I. (2021). J. Appl. Cryst. 54, 569–579.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationHielscher, R. & Lippert, L. (2021). J. Multivariate Anal. 185, 104764.  Web of Science CrossRef Google Scholar
First citationHumbert, M., Germain, L., Gey, N. & Boucard, E. (2015). Acta Mater. 82, 137–144.  Web of Science CrossRef CAS Google Scholar
First citationHumbert, M. & Gey, N. (2002). J. Appl. Cryst. 35, 401–405.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationHumbert, M., Moustahfid, H., Wagner, F. & Philippe, M. J. (1994). Scr. Metall. Mater. 30, 377–382.  CrossRef CAS Web of Science Google Scholar
First citationHumbert, M., Wagner, F., Moustahfid, H. & Esling, C. (1995). J. Appl. Cryst. 28, 571–576.  CrossRef CAS Web of Science IUCr Journals Google Scholar
First citationJohnstone, D. N., Martineau, B. H., Crout, P., Midgley, P. A. & Eggeman, A. S. (2020). J. Appl. Cryst. 53, 1293–1298.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKarthikeyan, T., Saroja, S. & Vijayalakshmi, M. (2006). Scr. Mater. 55, 771–774.  Web of Science CrossRef CAS Google Scholar
First citationKitahara, H., Ueji, R., Tsuji, N. & Minamino, Y. (2006). Acta Mater. 54, 1279–1288.  Web of Science CrossRef CAS Google Scholar
First citationKoumatos, K. & Muehlemann, A. (2017). Acta Cryst. A73, 115–123.  Web of Science CrossRef IUCr Journals Google Scholar
First citationKurdjumow, G. & Sachs, G. (1930). Z. Phys. 64, 325–343.  CrossRef Google Scholar
First citationMackenzie, J. K. (1957). Acta Cryst. 10, 61–62.  CrossRef IUCr Journals Web of Science Google Scholar
First citationMainprice, D., Humbert, M. & Wagner, F. (1990). Tectonophysics, 180, 213–228.  CrossRef Web of Science Google Scholar
First citationMardia, K. V. & Jupp, P. E. (2000). Directional Statistics. Chichester: Wiley.  Google Scholar
First citationMiyamoto, G., Takayama, N. & Furuhara, T. (2009). Scr. Mater. 60, 1113–1116.  Web of Science CrossRef CAS Google Scholar
First citationMorales, L. F. G., Mainprice, D. & Kern, H. (2018). Tectonophysics, 724–725, 93–115.  Web of Science CrossRef Google Scholar
First citationMorawiec, A. (2004). Orientations and Rotations, 1st ed. Berlin: Springer-Verlag.  Google Scholar
First citationNishiyama, Z. (1934). Sci. Rep. Tohoku Univ. 23, 637–664.  CAS Google Scholar
First citationNolze, G. (2004a). Proceedings of the CHANNEL Users Conference 2004, Ribe, Denmark, pp. 37–45. Berlin: BAM.  Google Scholar
First citationNolze, G. (2004b). Z. Metallkd. 95, 744–755.  Web of Science CrossRef CAS Google Scholar
First citationNolze, G. (2005). Determining the f.c.c./b.c.c. Orientational Relationship in Plessite Regions of Iron Meteorites. HKL Users Meeting.  Google Scholar
First citationNolze, G. (2008). Cryst. Res. Technol. 43, 61–73.  Web of Science CrossRef CAS Google Scholar
First citationNyyssönen, T., Isakov, M., Peura, P. & Kuokkala, V.-T. (2016). Metall. Mater. Trans. A, 47, 2587–2590.  Google Scholar
First citationNyyssönen, T., Peura, P. & Kuokkala, V.-T. (2018). Metall. Mater. Trans. A, 49, 6426–6441.  Google Scholar
First citationOstapovich, K. V. & Trusov, P. (2021). Crystals, 11, 336.  Web of Science CrossRef Google Scholar
First citationPitsch, W. (1959). Philos. Mag. 4, 577–584.  CrossRef Web of Science Google Scholar
First citationPitsch, W. (1967). Arch. Eisenhüttenwesen 38, 853–864.  CAS Google Scholar
First citationWassermann, G. (1933). Arch. Eisenhüttenwesen 6, 347–351.  CAS Google Scholar
First citationWassermann, G. (1935). Über den Mechanismus der α–γ Umwand­lung des Eisens. Mitteilungen aus dem Kaiser-Wilhelm-Institut für Eisenforschung zu Düsseldorf. Düsseldorf: Verlag Stahleisen.  Google Scholar
First citationWendler, M., Ullrich, C., Hauser, M., Krüger, L., Volkova, O., Weiß, A. & Mola, J. (2017). Acta Mater. 133, 346–355.  Web of Science CrossRef CAS Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.

Journal logoJOURNAL OF
APPLIED
CRYSTALLOGRAPHY
ISSN: 1600-5767
Follow J. Appl. Cryst.
Sign up for e-alerts
Follow J. Appl. Cryst. on Twitter
Follow us on facebook
Sign up for RSS feeds