The TameGenus package contains polynomial-time algorithms to decide isomorphism, construct automorphism groups, and build canonical labels of a class of p-groups of nilpotency class 2 and exponent p. We say such groups have tame genus, and we provide a function called IsTameGenusGroup to check whether the functions in TameGenus can be applied to the given group. This class includes d-generated groups of order pd + 2, exponent p, and class 2.For definitions, theorems, and descriptions of algorithms, see [BMW17]. When appropriate, intrinsics for groups also work for bilinear maps (TenSpcElt).
We have included intrinsics to allow for verbose printing. The string "TameGenus" is now accepted by SetVerbose with print levels between 0 and 2. Level 0 turns off all printing. Level 1 will print out the steps in the algorithms, and level 2 will print timings and extra structural information along the way.
> G := RandomGenus2Group(3, [4, 6, 10]); > #G eq 3^(4 + 6 + 10 + 2); true > Genus(G); 2
With the verbose printing, we can see steps of the algorithm.
> SetVerbose("TameGenus", 1); > A := TGAutomorphismGroup(G); Extracting the p-central tensor and computing pseudo-isometries. Checking the radicals. dim(Rad_V) = 0 dim(Rad_W) = 0 Writing tensor over its centroid. Tensor has genus 2. Computing the adjoint algebra. dim(Adj) = 40 Decomposing tensor into flat and sloped subtensors. Block dims = [ 4, 6, 10 ] Number of sloped blocks to lift: 3. Field is small enough, applying Pfaffian method. Number of flat blocks to lift: 0. Constructing the isometry group. G | Sp ( 2 , 3 ^ 2 ) * | Sp ( 2 , 3 ^ 3 ) * | Sp ( 2 , 3 ^ 5 ) * | 3 ^ 0 (unipotent radical) 1 Constructing automorphism group from pseudo-isometries.
We introduce functions to construct groups of genus ≤2.
Exponentp: BoolElt Default: true
Given q=pm, n>0, and g>0, returns a p-group G with genus ≤g and order qn + g. The algorithm is based on the Universal Coefficients Theorem, see [LGM02, Chapter 9] for the statement and proof. There is one optional parameter: Exponentp.
- (*)
- Exponentp: Set to false if the constructor should not force exponent p.
> G := TGRandomGroup(5^7, 8, 2); > #G eq 5^(7*(8 + 2)); true
We check that the genus of our group G is at most 2.
> Genus(G); 2
We verify that the commutator tensor has a centroid containing the field of order 57.
> t := pCentralTensor(G, 1, 1); > t; Tensor of valence 3, U2 x U1 >-> U0 U2 : Full Vector space of degree 56 over GF(5) U1 : Full Vector space of degree 56 over GF(5) U0 : Full Vector space of degree 28 over GF(5) > C := Centroid(t); > Dimension(C); 7 > IsSimple(C); true
Exponentp: BoolElt Default: true
Given q=pm and d=[d1, ..., dk], where di>0, return a genus 2 group G whose commutator tensor [, ] : V x V ↣ W which is GF(q)-bilinear and whose perp-decomposition has blocks of the prescribed dimensions d1, ..., dk over GF(q). Each di=1 increases the dimension of the radical of [, ] by 1 over GF(q). There is one optional parameter: Exponentp.
- (*)
- Exponentp: Set to false if the constructor should not force exponent p.
> G := RandomGenus2Group(7, [10, 7, 4, 4, 1]); > #G eq 7^(10 + 7 + 4 + 4 + 1 + 2); true
We can see these blocks with the genus 2 signature of the group. See TGSignature for a description of the output.
> S := TGSignature(G); > S; [* <7, 25, 2>, <1, 0>, [ 7 ], [* [ 1, 0, 1 ], [ 1, 4, 5 ], [ 1, 3, 3, 2, 6, 3 ] *] *]
Exponentp: BoolElt Default: true
Given q=pm, d>0, and r≥0, return a group G with genus 1 and of order q2d + r + 1. The center of G has order qr + 1, and G is m(2d + r)-generated. There is one optional parameter: Exponentp.
- (*)
- Exponentp: Set to false if the constructor should not force exponent p.
> G := RandomGenus1Group(81, 2, 0); > #G eq 81^(2*2 + 1); true
We verify that G has genus 1 and that the derived subgroup is isomorphic to (Z/(3))4.
> D := DerivedSubgroup(G); > IsElementaryAbelian(D); true > #D eq 3^4; true
Although the underlying field GF(81) is hidden within the relations of the polycyclic presentation, we can still construct it and compute the genus over this larger field.
> Genus(G); 1
Given either a univariate or homogeneous multivariate polynomial in two variables over GF(q), return a group G whose commutator GF(q)-tensor has a Pfaffian in the same Γ(L)(2, q)-orbit as f.
> P<x> := PolynomialRing(GF(9)); > f := x^3*(x-1)^2*(x-2); > G := Genus2Group(f); > #G eq 9^(2*6 + 2); true
The genus 2 signature reveals the block structure. All six perp-indecomposable subspaces are 2-dimensional.
> S := TGSignature(G); > S; [* <9, 12, 2>, <0, 0>, [], [* [ 1, 1 ], [ 1, 0, 0 ], [ 0, 0, 0, 1 ] *] *]
The output shows that the homogenization of f is equivalent to the polynomial g(x, y) = x2y3(x + y) in GF(q)[x, y].
Determining the genus of a group G requires knowing the direct indecomposable factors of G, so there are functions to decide if G is indecomposable.
Given a group G, decide if G is directly indecomposable. This algorithm is based on algorithms by Wilson in [Wil12].
Given a tensor t, decide if t is directly indecomposable. This algorithm is based on algorithms by Wilson in [Wil12].
> P<x, y> := PolynomialRing(GF(7), 2); > f := x*y; > G := Genus2Group(f); > G; GrpPC : G of order 117649 = 7^6 PC-Relations: G.3^G.1 = G.3 * G.5, G.4^G.2 = G.4 * G.6 > IsIndecomposable(G); false
This example also shows that TGRandomGroup may return a group that is directly decomposable.
> G := RandomGenus2Group(5, [3, 4]); > #G eq 5^(3 + 4 + 2); true > IsIndecomposable(G); true
We observe the central factors from the genus 2 signature of G. We see that G is a central product of genus 2 groups of order 55 and 56.
> TGSignature(G); [* <5, 7, 2>, <0, 0>, [ 3 ], [* [ 1, 0, 2 ] *] *]
Given a directly indecomposable p-group G, return the genus of G as defined in [BMW17].
Given a directly indecomposable tensor t: V x V ↣ W, return the rank of W as a C-module. Only implemented for algebras C such that C/J(C) is simple, where J(C) is the Jacobson radical.
> G := SmallGroup(3^6, 440); > #DerivedSubgroup(G) eq 3^2; true > IsElementaryAbelian(DerivedSubgroup(G)); true > Genus(G); 2
Let H be the group of order 36 with ID number 469. Its derived subgroup is isomorphic to (Z/(3))2 also, but H is genus 1.
> H := SmallGroup(3^6, 469); > #DerivedSubgroup(H) eq 3^2; true > IsElementaryAbelian(DerivedSubgroup(H)); true > Genus(H); 1
We can see this difference in genus in multiple ways. If we extract the commutator tensor from both groups, one is bilinear over a quadratic field extension where the other tensor is not. The algebra that records this is the centroid of a tensor, see Centroid.
> s := pCentralTensor(G, 1, 1); > t := pCentralTensor(H, 1, 1); > C_G := Centroid(s); > C_H := Centroid(t); > C_G; Matrix Algebra of degree 6 with 2 generators over GF(3) > C_H; Matrix Algebra of degree 6 with 2 generators over GF(3)
The centroid associated to the tensor s is 2-dimensional but with a nontrivial Jacobson radical. The centroid associated to the tensor t is 2-dimensional and a simple GF(3)-algebra.
> Dimension(C_G); 2 > Dimension(C_H); 2 > WedderburnDecomposition(C_G); Matrix Algebra of degree 6 with 1 generator over GF(3) Matrix Algebra of degree 6 with 1 generator over GF(3) > IsSimple(C_H); true
Decides if the functions in the TameGenus package (prefixed with TG) can be applied to the given group G.
Decides if the functions in the TameGenus package (prefixed with TG) can be applied to the given tensor t.
> H1 := SmallGroup(3^6, 440); > H2 := SmallGroup(3^6, 469); > G := DirectProduct(H1, H2); > G; GrpPC : G of order 531441 = 3^12 PC-Relations: G.2^G.1 = G.2 * G.5, G.3^G.1 = G.3 * G.6, G.3^G.2 = G.3 * G.5, G.4^G.1 = G.4 * G.5, G.8^G.7 = G.8 * G.11, G.9^G.7 = G.9 * G.12, G.9^G.8 = G.9 * G.11^2 * G.12, G.10^G.7 = G.10 * G.11 * G.12, G.10^G.8 = G.10 * G.11 > IsTameGenusGroup(G); false
> G2_SG := [*G : G in SmallGroups(3^6) | IsTameGenusGroup(G)*]; > #G2_SG; 7
We further filter these groups based on their genus, and we find that there are six groups of positive genus and one abelian group.
> G1 := [*G : G in G2_SG | Genus(G) eq 1*]; > G2 := [*G : G in G2_SG | Genus(G) eq 2*]; > #G1; 3 > #G2; 3
We determine the tame genus signatures to show that all these groups yield different signatures. We first consider the genus 1 groups.
> for G in G1 do > TGSignature(G); > end for; [* <9, 2, 1>, <0, 0>, [* *], [* *] *] [* <3, 2, 1>, <3, 0>, [* *], [* *] *] [* <3, 4, 1>, <1, 0>, [* *], [* *] *]
The genus 2 groups have the following signatures.
> for G in G2 do > TGSignature(G); > end for; [* <3, 3, 2>, <1, 0>, [ 3 ], [* *] *] [* <3, 4, 2>, <0, 0>, [], [* [ 0, 0, 1 ] *] *] [* <3, 4, 2>, <0, 0>, [], [* [ 1, 0 ], [ 0, 1 ] *] *]
Cent: BoolElt Default: true
Constructive: BoolElt Default: true
Method: RngIntElt Default: 0
Given class ≤2, exponent p, directly indecomposable odd-ordered p-groups G and H of genus ≤2, decide if G isomorphic to H. This will accept groups G of the form G isomorphic to A x K, where A is abelian and K is directly indecomposable. There are three optional parameters: Cent, Constructive, and Method.
- (*)
- Cent: If G and H are known to have genus ≤2 over the prime field, then set to false to save some time.
- (*)
- Constructive: Set to false if an explicit isomorphism is not needed.
- (*)
- Method: Input from { 0, 1, 2} is acceptable. This determines the method for handling the sloped part of the tensor. If the adjoint-tensor method should be used, set to 1. If the Pfaffian method should be used, set to 2. The default will try to find the optimal method based on the input; see [BMW17] for details.
> f := RandomIrreduciblePolynomial(GF(67), 3); > g := RandomIrreduciblePolynomial(GF(67), 3); > f, g; $.1^3 + 36*$.1^2 + 26*$.1 + 33 $.1^3 + 57*$.1^2 + 24*$.1 + 65 > G := Genus2Group(f); > H := Genus2Group(g);
In this case, the default method will run the adjoint-tensor method. We can see this if we turn on verbose printing.
> SetVerbose("TameGenus", 1); > isomorphic, phi := TGIsIsomorphic(G, H); Extracting p-central tensors and deciding pseudo-isometry. Checking the radicals. dim(Rad_V) = 0 dim(Rad_W) = 0 Checking the radicals. dim(Rad_V) = 0 dim(Rad_W) = 0 Writing tensor over its centroid. Writing tensor over its centroid. Computing the adjoint algebra. dim(Adj_s) = 12 dim(Adj_t) = 12 Computing the adjoint algebra. Genus 2 case. Decomposing tensors into flat and sloped subtensors. Block dims = [ 6 ] PGammaL is larger than symmetric group, applying adjoint-tensor method.
We see that the two groups are isomorphic, and φ is such an isomorphism.
> isomorphic; true > phi; Homomorphism of GrpPC : G into GrpPC : H induced by G.1 |--> H.1 * H.2^41 * H.3^30 * H.4^58 * H.5^43 * H.6^30 G.2 |--> H.1^53 * H.2^4 * H.3^34 * H.4^54 * H.5^6 * H.6^25 G.3 |--> H.1^5 * H.2^7 * H.3^49 * H.4^52 * H.5^21 * H.6^7 G.4 |--> H.1^30 * H.2^4 * H.3^17 * H.4^53 * H.5^40 * H.6^9 G.5 |--> H.1^14 * H.2^21 * H.3^30 * H.4^24 * H.5^38 * H.6^34 G.6 |--> H.1^47 * H.2^46 * H.3^66 * H.4^65 * H.5^30 * H.6^30 G.7 |--> H.7^10 * H.8^18 G.8 |--> H.7^54 * H.8^12
To run the Pfaffian test, set the Method to 2. This method is noticeably slower for these groups and should not be used for very large primes.
> isomorphic, phi2 := TGIsIsomorphic(G, H : Method := 2); Extracting p-central tensors and deciding pseudo-isometry. Checking the radicals. dim(Rad_V) = 0 dim(Rad_W) = 0 Checking the radicals. dim(Rad_V) = 0 dim(Rad_W) = 0 Writing tensor over its centroid. Writing tensor over its centroid. Computing the adjoint algebra. dim(Adj_s) = 12 dim(Adj_t) = 12 Computing the adjoint algebra. Genus 2 case. Decomposing tensors into flat and sloped subtensors. Block dims = [ 6 ] Method set to Pfaffian. > phi2; Homomorphism of GrpPC : G into GrpPC : H induced by G.1 |--> H.1^47 * H.2^54 * H.3^54 G.2 |--> H.1^63 * H.2^33 * H.3^15 G.3 |--> H.1^26 * H.2^3 * H.3^60 G.4 |--> H.4^44 * H.5^61 * H.6^16 G.5 |--> H.4^66 * H.5^26 * H.6^36 G.6 |--> H.4^57 * H.5^6 * H.6^30 G.7 |--> H.7^62 * H.8^42 G.8 |--> H.7^18 * H.8^58
Cent: BoolElt Default: true
Constructive: BoolElt Default: true
Method: RngIntElt Default: 0
Given directly indecomposable, alternating tensors s, t : V x V ↣ W, where V and W are GF(q)-vector spaces of odd order, decide if s is pseudo-isometric to t. This will accept tensors with nontrivial radicals and coradicals. There are three optional parameters: Cent, Constructive, and Method.
- (*)
- Cent: If G and H are known to have genus ≤2 over the prime field, then set to false to save some time.
- (*)
- Constructive: Set to false if an explicit isomorphism is not needed.
- (*)
- Method: Input from { 0, 1, 2} is acceptable. This determines the method for handling the sloped part of the tensor. If the adjoint-tensor method should be used, set to 1. If the Pfaffian method should be used, set to 2. The default will try to find the optimal method based on the input; see [BMW17] for details.
> M := RandomMatrix(GF(3), 50, 50); > N := RandomMatrix(GF(3), 50, 50); > Forms1 := [M - Transpose(M), N - Transpose(N)]; > s := Tensor(Forms1, 2, 1); > s; Tensor of valence 3, U2 x U1 >-> U0 U2 : Full Vector space of degree 50 over GF(3) U1 : Full Vector space of degree 50 over GF(3) U0 : Full Vector space of degree 2 over GF(3) > IsAlternating(s); true
We construct a random invertible co-homotopism varphi = (varphi1, varphi0) such that varphi1∈GL(50, 3) and varphi0∈GL(2, 3). We define a new and pseudo-isometric tensor t=svarphi in this way.
> X := Random(GL(50, 3)); > Y := Random(GL(2, 3)); > Maps := [*X, X, Y*]; > H := Homotopism(Maps, CohomotopismCategory(3)); > t := s @ H; > t; Tensor of valence 3, U2 x U1 >-> U0 U2 : Full Vector space of degree 50 over GF(3) U1 : Full Vector space of degree 50 over GF(3) U0 : Full Vector space of degree 2 over GF(3) > s eq t; false
We verify that both s and t are pseudo-isometric, and since we do not need to write these tensors over their centroid, we avoid that step by setting Cent to false.
> pisometric, Phi := TGIsPseudoIsometric(s, t : Cent :=false); > pisometric; true > Phi.0; [0 1] [2 2]
Cent: BoolElt Default: true
Method: RngIntElt Default: 0
Mat: BoolElt Default: false
Given a directly indecomposable odd-ordered p-group G, of class ≤2, exponent p, and genus ≤2, return Aut(G). This will accept groups G of the form G isomorphic to A x K, where A is abelian and K is directly indecomposable. This intrinsic provides three optional parameters: Cent, Method, and Mat.
- (*)
- Cent: If G and H are known to have genus ≤2 over the prime field, then set to false to save some time.
- (*)
- Method: Input from { 0, 1, 2} is acceptable. This determines the method for handling the sloped part of the tensor. If the adjoint-tensor method should be used, set to 1. If the Pfaffian method should be used, set to 2. The default will try to find the optimal method based on the input; see [BMW17] for details.
- (*)
- Mat: Set to true for a linear representation of the automorphism group. The row vectors of the matrix correspond to the given polycyclic vectors from the polycyclic presentation.
> p := 541; > G := RandomGenus2Group(p, [29]); > #G eq p^(29 + 2); true > A := TGAutomorphismGroup(G); > IsDivisibleBy(#A, #GL(2, p)); true
The isometry group of the corresponding commutator tensor has an elementary abelian unipotent radical of rank 28, and so is isomorphic to an extension of GL(1, p) by (Z/(p))28, as seen in [BW12]. Therefore, we verify that we have constructed all automorphisms as
|Aut(G)| = (p - 1)p28 .(p2 - 1)(p2 - p) .p29.2.
> #A eq (p - 1)*p^28 * (p^2 - 1)*(p^2 - p) * p^(29*2); true
Cent: BoolElt Default: true
Method: RngIntElt Default: 0
Given a directly indecomposable, alternating tensor t : V x V ↣ W, where V and W are GF(q)-vector spaces for odd q, return the pseudo-isometry group of t as a subgroup of GL(V) x GL(W). This accepts tensors with nontrivial radicals and coradicals. This intrinsic provides two optional parameters: Cent and Method.
- (*)
- Cent: If G and H are known to have genus ≤2 over the prime field, then set to false to save some time.
- (*)
- Method: Input from { 0, 1, 2} is acceptable. This determines the method for handling the sloped part of the tensor. If the adjoint-tensor method should be used, set to 1. If the Pfaffian method should be used, set to 2. The default will try to find the optimal method based on the input; see [BMW17] for details.
> G := RandomGenus2Group(3^2, [6]); > t := pCentralTensor(G, 1, 1); > t; Tensor of valence 3, U2 x U1 >-> U0 U2 : Full Vector space of degree 12 over GF(3) U1 : Full Vector space of degree 12 over GF(3) U0 : Full Vector space of degree 4 over GF(3) > Genus(t); 2
The Pfaffian of t is an irreducible cubic over GF(9), so computing its pseudo-isometry group may contain elements of (Gal)(GF(9)).
> PI := TGPseudoIsometryGroup(t); > Random(PI); [0 1 0 0 0 2 1 1 0 2 1 0 0 0 0 0] [2 0 0 0 1 0 0 2 1 0 1 2 0 0 0 0] [1 1 0 0 0 1 1 1 0 0 2 1 0 0 0 0] [0 2 0 0 2 0 0 2 0 0 1 1 0 0 0 0] [1 2 2 0 0 2 2 2 1 0 0 0 0 0 0 0] [2 2 2 1 1 0 0 1 1 2 0 0 0 0 0 0] [2 2 2 0 1 0 1 0 2 2 1 2 0 0 0 0] [0 1 2 1 1 2 1 2 0 1 2 2 0 0 0 0] [2 0 1 0 2 2 2 1 1 1 0 0 0 0 0 0] [2 1 1 2 0 1 1 1 0 2 0 0 0 0 0 0] [0 2 0 1 2 2 1 1 0 1 1 0 0 0 0 0] [1 0 2 0 0 1 0 2 2 0 1 2 0 0 0 0] [0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0] [0 0 0 0 0 0 0 0 0 0 0 0 2 0 1 2] [0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0] [0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0] > Factorization(#PI); [ <2, 8>, <3, 7>, <5, 1>, <7, 1>, <13, 1>, <73, 1> ]
Indeed, the pseudo-isometry group may not be linear over the centroid but semi-linear. To see the discrepancy, we rewrite t over its centroid and determine its pseudo-isometry group as a subgroup of GLE(V) x GLE(W). Note that this group is smaller by a factor of 2. We see that the pseudo-isometry group of t is strictly semi-linear over GF(9), as there exists a Galois involution.
> s := TensorOverCentroid(t); > s; Tensor of valence 3, U2 x U1 >-> U0 U2 : Full Vector space of degree 6 over GF(3^2) U1 : Full Vector space of degree 6 over GF(3^2) U0 : Full Vector space of degree 2 over GF(3^2) > PI_C := TGPseudoIsometryGroup(s); > Random(PI_C); [ 1 $.1^7 1 $.1^2 $.1^2 $.1^5 0 0] [ $.1 2 $.1^7 $.1^3 $.1^6 $.1^5 0 0] [$.1^7 $.1^2 $.1^7 1 $.1^6 $.1^2 0 0] [$.1^7 $.1^7 $.1^7 $.1^3 $.1^7 1 0 0] [ 2 $.1^6 $.1^3 $.1^2 $.1^3 0 0 0] [$.1^7 $.1 $.1^7 $.1^6 $.1^6 $.1^7 0 0] [ 0 0 0 0 0 0 2 $.1^7] [ 0 0 0 0 0 0 $.1^3 $.1^5] > Factorization(#PI_C); [ <2, 7>, <3, 7>, <5, 1>, <7, 1>, <13, 1>, <73, 1> ]
Given an odd-ordered p-group G isomorphic to A x H of genus ≤2, class ≤2, and exponent p, return the canonical tame genus signature as a list with four entries. For groups with genus ≤1, the last two entries will be empty. We describe the entries.c1xd + c2 xd - 1y + c3 xd - 2y2 + ... + cd + 1yd.
- (1)
- The first is a triple of three integers (q, d, e) such that associated fully nondegenerate commutator tensor of H has the form GF(q)d x GF(q)d ↣ GF(q)e. Here, GF(q) is the largest field F which the commutator tensor is F-bilinear.
- (2)
- The second entry is a pair of integers (r, c), where A isomorphic to (Z/(p))r and c=0.
- (3)
- The third entry is a sequence of odd integers corresponding to the dimensions of the flat indecomposable spaces.
- (4)
- The fourth and last entry is a list sequences in GF(q) of possibly different lengths. A sequence [c1, ..., cd + 1] in the second entry corresponds to the coefficients of the homogeneous polynomial in x and y in degree d:
Because this is a canonical label, two groups are isomorphic if, and only if, their tame genus signatures are equal. See [BMW17] for details on how the last two entries describe a genus 2 group.
Given an alternating, directly indecomposable tensor t:V x V ↣ W of genus ≤2, where V and W are GF(q)-vector spaces for odd q, return the canonical tame genus signature as a list with four entries. For tensors with genus ≤1, the last two entries will be empty. We describe the entries.c1xd + c2 xd - 1y + c3 xd - 2y2 + ... + cd + 1yd.
- (1)
- The first is a triple of three integers (qf, d, e) such that associated fully nondegenerate tensor of t has the form GF(qf)d x GF(qf)d ↣ GF(qf)e. Here, GF(qf) is the largest field F which the associated fully nondegenerate tensor is F-bilinear.
- (2)
- The second entry is a pair of integers (r, c), where the radical of t is r-dimensional and the codimension of the image of t in W is c, both are determined over GF(q).
- (3)
- The third entry is a sequence of odd integers corresponding to the dimensions of the flat indecomposable spaces.
- (4)
- The fourth and last entry is a list sequences in GF(q) of possibly different lengths. A sequence [c1, ..., cd + 1] in the second entry corresponds to the coefficients of the homogeneous polynomial in x and y in degree d:
Because this is a canonical label, two tensors are pseudo-isometric if, and only if, their tame genus signatures are equal. See [BMW17] for details on how the last two entries describe a genus 2 tensor.
> blocks := [1, 1, 1, 2, 2, 3, 3, 5, 6]; > G := RandomGenus2Group(9, blocks); > #G eq 9^(&+blocks + 2); true
Because G was created with the field GF(9) and since there are three 1-dimensional blocks, we expect |A| = 93 = 36.
> TGSignature(G); [* <9, 21, 2>, <6, 0>, [ 3, 3, 5 ], [* [ 0, 1 ], [ 1, 1 ], [ 1, 2, 2, 2 ] *] *]
The first entry of this list tells us that the group H has order 921 + 2 and that the commutator tensor of H is GF(9)-bilinear. The second entry reports that A isomorphic to (Z/(3))6. The third and fourth entries show that H is a central product of six indecomposable groups: three flats of orders 93 + 2, 93 + 2, and 95 + 2, and three sloped with Pfaffians in GF(9)[x, y] equivalent to
y, x + y, x3 + 2x2y + 2xy2 + 2y3.
> TG_SG := SmallGroupProcess(3^7, IsTameGenusGroup); > > repeat > G := Current(TG_SG); > _, ID := CurrentLabel(TG_SG); > print "Small group ID:", ID; > TGSignature(G); > Advance(~TG_SG); > until IsEmpty(TG_SG);
The loop yields the following output.
Small group ID: 9106 [* <3, 3, 2>, <2, 0>, [ 3 ], [* *] *] Small group ID: 9107 [* <3, 4, 2>, <1, 0>, [], [* [ 1, 0 ], [ 0, 1 ] *] *] Small group ID: 9108 [* <3, 4, 2>, <1, 0>, [], [* [ 0, 0, 1 ] *] *] Small group ID: 9109 [* <9, 2, 1>, <1, 0>, [* *], [* *] *] Small group ID: 9110 [* <3, 5, 2>, <0, 0>, [ 3 ], [* [ 0, 1 ] *] *] Small group ID: 9111 [* <3, 5, 2>, <0, 0>, [ 5 ], [* *] *] Small group ID: 9302 [* <3, 2, 1>, <4, 0>, [* *], [* *] *] Small group ID: 9305 [* <3, 4, 1>, <2, 0>, [* *], [* *] *] Small group ID: 9308 [* <3, 6, 1>, <0, 0>, [* *], [* *] *] Small group ID: 9310 [* <3, 7, 0>, <7, 0>, [* *], [* *] *]