Halving Steiner 2-designs

Halving Steiner 2-designs
Halving Steiner 2-designs

Halving Steiner2-designs

Yuichiro Fujiwara

Graduate School of Information Science,Nagoya University,

Furo-cho,Chikusa-ku,464-8601,Japan

E-mail:fujiwara@math.cm.is.nagoya-u.ac.jp

Key words:Halving,Steiner2-design,Self-complementary graph,Decomposition

Abstract

A Steiner2-design S(2,k,v)is said to be halvable if the block set can be par-

titioned into two isomorphic sets.This is equivalent to an edge-disjoint decom-

position of a self-complementary graph G on v vertices into K k s.The obvious

necessary condition of those orders v for which there exists a halvable S(2,k,v)

is that v admits the existence of an S(2,k,v)with an even number of blocks.In

this paper,we give an asymptotic solution for various block sizes.We prove that

for any k≤5or any Mersenne prime k,there is a constant number v0such that if

v>v0and v satis?es the above necessary condition,then there exists a halvable

S(2,k,v).We also show that a halvable S(2,2n,v)exists for over a half of possi-

ble orders.Some recursive constructions generating in?nitely many new halvable

Steiner2-designs are also presented.

1Introduction

A Steiner2-design of order v and block size k,brie?y S(2,k,v),is an ordered pair (V,B),where V is a?nite set of v elements called points,and

B is a set of k-element subsets of V called blocks,such that each unordered pair of distinct elements of V is contained in exactly one block of B.It is well-known that an S(2,k,v)exists only if v?1≡0(mod k?1)and v(v?1)≡0(mod k(k?1)),and that there is a constant number v0such that if v>v0and v satis?es the necessary conditions,then there exists an S(2,k,v)(see Wilson[13,14,15]).

A factor H of a Steiner2-design(V,B)is an ordered pair(V,C)where C is a subset of B.Two factors H i and H j are said to be isomorphic and denoted H i H j if there exists a bijectionφof V onto itself such that

B ={φ(x1),φ(x2),...,φ(x k)}is a block of H j if and only if B={x1,x2,...,x k}is a block of H i.Such a bijectionφis said to be isomorphism,and further if H i=H j,then the isomorphismφis said to be an automorphism of H i.The full automorphism group is the group of all automorphisms of H i.

1

An S (2,2,v )(V ,B )is equivalent to the complete graph K v on v vertices and a factor of (V ,B )can be considered to be a graph.A graph G =(V ,C )is said to be self-complementary if it is isomorphic to its complement ˉG ,that is,(V ,C ) (V ,B \C ).The isomorphism G →ˉG

is said to be a complementing permutation or an antimorphism .It is a well-know fact that a self-complementary graph G of order v exists if and only if v ≡0,1(mod 4).

An S (2,k ,v )(V ,B )is said to be halvable if there exists a partition of B =B 1∪B 2,

B 1∩B 2=/0such that (V ,B 1) (V ,B 2),that is,the block set B can be partitioned into two isomorphic factors.A halvable S (2,2,v )is just a self-complementary graph of order v .While an S (2,k ,v )can be considered to be an edge-disjoint decomposition of the complete graph K v into K k s,a halvable S (2,k ,v )is equivalent to that of a self-complementary graph G on v vertices into K k s.In this sense,we say the isomorphism between B 1and B 2to be a complementing permutation .The obvious necessary con-dition of those orders v for which there exists a halvable S (2,k ,v )is that v admits the existence of an S (2,k ,v )with an even number of blocks,that is,

(i)v ?1≡0(mod k ?1)and v (v ?1)≡0(mod k (k ?1)),

(ii)the total number of blocks v 2 / k 2 is even.

Halving Steiner 2-designs was ?rst investigated for S (2,3,v )s,known as Steiner triple systems ,by Das and Rosa [8].Around the same time Phelps [11]studied the existence of halvable S (2,4,v )s.As well as on halving Steiner 2-designs,there have also been developed many results on the problems of separating graphs or designs into some isomorphic parts.The interested reader is referred to Fronˇc ek and Meszka [10]and references there in.

We shall study the spectrum of those orders for which halvable S (2,k ,v )s exist.In general,a complementing permutation φof a decomposable self-complementary graph G need not be an automorphism of the corresponding halvable Steiner 2-design S .However,if there exists a complementing permutation of G which is also an au-tomorphism of underlying S ,then S is said to be strongly halvable .The necessary condition for the existence of a halvable S (2,3,v )is that v ≡1,3(mod 6)and v 2 / 32 is even,that is,v ≡1,9(mod 12).By observing the structure of automorphisms of certain triple systems,Das and Rosa [8]proved:

Theorem 1.1(Das and Rosa)[8]There exists a strongly halvable S (2,3,v )if and only if v ≡1,9(mod 24).

For the remaining congruence classes v ≡13,21(mod 24),in which no strongly halvable S (2,3,v )exists,they gave a partial result:

Theorem 1.2(Das and Rosa)[8]Let p be a prime,n ≥1,and p n ≡13(mod 24).Then,there exists a halvable S (2,3,p n ).

The necessary condition for k =4is that v ≡1,16(mod 24).Phelps [11]con-structed a halvable S (2,4,v )for all such orders with several exceptions,where all the remaining orders are in the class v ≡16(mod 24)and up to 616(the case v =88was missed in the literature).

2

Theorem1.3(Phelps)[11]There exists a halvable S(2,4,v)for all v≡1,16(mod24) except possibly v=88,136,184,232,328,424,616.

As far as the author knows,no further result on halving Steiner2-designs had been known except obvious examples.The main purpose of this paper is to give an asymp-totic solution for the halving problem on Steiner2-designs of various block sizes such as3,5and primes of the form2n?1,called Mersenne primes.

Theorem1.4For any k≤5or any Mersenne prime k,there is a constant number v0such that if v>v0and v satis?es the necessary conditions for the existence of an S(2,k,v)with an even number of blocks,then there exists a halvable S(2,k,v).

For integers being powers of2,we show the following:

Theorem1.5For any positive integer n≥2,the arithmetic density of the set of all orders for which a halvable S(2,2n,v)exists as compared to the set of all orders sat-isfying the necessary conditions for the existence of a halvable S(2,2n,v)is at least 1/2.

The method we shall use is also adaptable to orders v in certain congruence classes for any positive integer k≥6.Some recursive constructions generating in?nitely many new halvable Steiner2-designs are also presented.For unde?ned notions appearing in the following section,we refer the reader to Beth,Jungnickel and Lenz[5].

2Asymptotic solution and constructions

In this section,we study the spectrum of orders for which there exists a halvable Steiner 2-design.First,we give the proofs of Theorems1.4and1.5.To prove these theorems, we de?ne some notions and prove four lemmas.

A group divisible design with index one is a triple(V,G,B),where

(i)V is a?nite set of elements called points,

(ii)G is a family of subsets of V,called groups,which partition V,

(iii)B is a collection of subsets of V,called blocks,such that every pair of points from distinct groups occurs in exactly one blocks,

(iv)|G∩B|≤1for all G∈G and B∈B.

When all groups are of the same size n,all blocks are of the same size k,and|G|=t, one refers to the design as a k-GDD of type n t.If t=k,a k-GDD of type n k is called

a transversal design of order n and is referred to as a TD(k,n).Fronˇc ek and Meszka

[10]completely settled the halving problem on TD(3,n)s.They investigated TDs in the graph decomposition approach and mentioned nothing about the halvability of groups. However,in order to use GDDs in design theoretic constructions for Steiner2-designs, we shall consider the halvability of a GDD in a stronger way.

3

A GDD(V,G,B)is said to be strictly halvable if it admits a bijectionφof V onto itself,a partition of G=G1∪G2,G1∩G2=/0and a partition of B=B1∪B2,B1∩B2=/0 such thatφ(G1)=G2andφ(B1)=B2.In other words,(V,G1) (V,G2)and(V,B1) (V,B2)with the same isomorphismφ.As is the case with Steiner2-designs,we say such a bijectionφis a complementing permutation.

An S(2,k,v)(V,B)has a subsystem of order w if there exists an S(2,k,w)(W,C) such that W?V and C?B.The following is an application of the well-known Funda-mental Construction of Wilson(see Phelps[11]and Wilson[13]).

Lemma2.1Let k≥3be an integer.Assume that a halvable S(2,k,w),a strictly halvable k-GDD of type g u/g,an S(2,k,vg+w)having a subsystem of order w and a TD(k,v)all exist.Then,there exists a halvable S(2,k,vu+w).

Proof.Let(U,G,B)be a strictly halvable k-GDD of type g u/g admitting a comple-menting permutationφand partitions G=G1∪φ(G1)and B=B1∪φ(B1).Take a?nite set V of cardinality v.For each block B∈B1,let(V×B,H B,C B)be a TD(k,v)so that the set of groups H B has the form{{(a,x),(b,x),...,(c,x)}:{a,b,...,c}=V,x∈B}. Take a?nite set W of cardinality w.For each G∈G1,place onto(V×G)∪W an S(2,k,vg+w)((V×G)∪W,D G)having a subsystem(W,E).Let(W,E )be a halvable S(2,k,w)with a complementing permutationπ.De?ne a bijectionτof(V×U)∪W onto itself asτ:(i,x)→(i,φ(x))for(i,x)∈V×U and i→π(i)for i∈W.Let F={C B∪τ(C B)∪(D G\E)∪τ(D G\E)∪E :B∈B1,G∈G1}.It is straightforward to see that((V×U)∪W,F)forms an S(2,k,vu+w)and the bijectionτhalves F into two isomorphic parts.2 Lemma2.1needs several kinds of designs to generate new halvable Steiner2-designs.First,we consider the existence of an S(2,k,v)having a subsystem. Lemma2.2Let k≥2be a positive integer and w≡1(mod k(k?1)).Then there exist a constant number w0depending on k,and a constant number v0depending on w and k such that if w>w0and v>v0satisfying the conditions v?1≡0(mod k?1)and v(v?1)≡0(mod k(k?1)),then there exists an S(2,k,v)having a subsystem of order w.

To prove Lemma2.2,we employ Wilson’s PBD-closure theorem.

Let K be a set of positive integers.A pairwise balanced design of order v with block sizes from K of index one,brie?y PBD(v,K),is an ordered pair(V,B)satisfying the properties:

(i)V is a?nite set of v elements called points,

(ii)B is a collection of subsets of V,called blocks,such that if B∈B,then|B|∈K, (iii)every pair of distinct points occurs in exactly one block of B.

If K consists of a single number k,then a PBD(v,K)is an S(2,k,v).

The closure of a set K of any positive integers,denoted B(K),is the set of all pos-itive integers v for which there exists a PBD(v,K),that is,B(K)={v:?PBD(v,K)}. K is said to be PBD-closed if B(K)=K.De?neα(K)=gcd{k?1:k∈K}and β(K)=gcd{k(k?1):k∈K}.

4

Theorem2.3(Wilson)[15]Let K be a PBD-closed set.Then there exists a constant

number v0such that if v>v0and v satis?es the conditions v?1≡0(modα(K))and v(v?1)≡0(modβ(K)),then there exists a PBD(v,K).

Theorem2.4(Wilson)[14]Let K be a PBD-closed set.Then K is eventually periodic

with periodβ(K).

Proof of Lemma2.2.Since any positive integer in the congruence class1modulo

k(k?1)satis?es the necessary conditions for the existence of an Steiner2-design of

block size k,there is a constant number w such that if w>w and w≡1(mod k(k?1)),

then there exists an S(2,k,w).It is well-known that there is a constant number w

such that if w>w ,then a TD(k,w)exists(see,for example,Beth[4]and Wilson

[16]).Hence,there exists a constant number w0,w0>w ,w ,such that if w>w0 and w≡1(mod k(k?1)),then an S(2,k,w),a TD(k,w)and a TD(k,w?1)all exist.

Let w≡1(mod k(k?1))be such an integer greater than w0,that is,an integer in

the congruence class being large enough to ensure the existence of an S(2,k,w),a

TD(k,w)and a TD(k,w?1).De?ne K as the set of all positive integers for which

there exists an S(2,k,v)having a subsystem of order w.Putting an S(2,k,w)on each

group of a TD(k,w),we obtain an S(2,k,kw)having a subsystem of order w.Similarly,

placing an S(2,k,w)on each group of a TD(k,w?1)with one additional point,say

∞,we obtain an S(2,k,k(w?1)+1)with a subsystem of order w.Hence,we have w,kw,k(w?1)+1∈K.

On the other hand,if a PBD(x,K)exists,then by putting on each block a Steiner2-

design having a subsystem of order w,we have x∈K.Hence B(K)=K and K is PBD-

closed.By Theorem2.3,it now suf?ces to show thatα(K)=k?1andβ(K)=k(k?1).

First,we show thatα(K)|(k?1).Sinceα(K)=gcd{a?1:a∈K},α(K)divides gcd(w?1,kw?1).Write w=k(k?1)i+1.Then,

gcd(w?1,kw?1)=gcd(k(k?1)i,(k2i+1)(k?1))

=(k?1)·gcd(ki,k2i+1)

=(k?1).

Next,we show thatβ(K)|k(k?1).Sinceβ(K)=gcd{a(a?1):a∈K},we have β(K)|gcd(w(w?1),kw(kw?1),(k(w?1)+1)k(w?1)).As is the case withα(K), gcd(w(w?1),kw(kw?1),(k(w?1)+1)k(w?1))

=gcd(wk(k?1)i,kw(k2i+1)(k?1),(k(w?1)+1)k2(k?1)i)

=k(k?1)·gcd(wi,w(k2i+1),w(k2i?1)+1)

=k(k?1).

Also,by Theorem2.4,K is eventually periodic with periodβ(K).Since any integer a∈K must satisfy the necessary conditions for the existence of an S(2,k,v),we have α(K)=k?1andβ(K)=k(k?1).The proof is complete.

2 Next,we show the existence of a halvable S(2,k,v)of certain prime power orders.

A difference family over an additive abelian group G of index one is a family

B of

5

subsets of G such that every nonzero element of G occurs exactly once among all differences b ?b ,where b and b are mutually different members of the subsets in B .A subset B ∈B is called a base block .If |G |=v and for every B ∈B ,the block size |B |=k ,then one refers to the family as a DF (v ,k ,1).The translates of all the base blocks of a DF (v ,k ,1)B over G ,namely {B +g ={b 1+g ,b 2+g ,...,b k +g }:B ={b 1,b 2,...,b k }∈B ,g ∈G },form an S (2,k ,v ).Let q =et +1be a prime power.Let C 0be the subgroup of GF ×(q )of index e and order t .Wilson [17]asymptotically settled the existence problem of a difference family over GF (q ).We use the following theorem.

Theorem 2.5(Wilson)[17]Let q =et +1be a prime power satisfying q ≡1(mod k (k ?1))and e = k 2 .Let C 0h be any arbitrary set of representatives of {?1,1}in C 0.Then,there exists a constant number q 0such that if q >q 0,there exists a DF (q ,k ,1)over GF (q )having the form {μB :μ∈C 0h }for some k-element subset B of GF (q ).

By using Theorem 2.5,we show the following lemma.

Lemma 2.6Let q be a prime power satisfying q ≡1(mod 2k (k ?1)).Then,there exists a constant number q 0such that if q >q 0,then there exists a halvable S (2,k ,q ).

Proof.Let q =2k (k ?1)t +1= k 2 4t +1be a prime power.Let ωbe a primitive element of GF (q )and C 0={ωi :i ≡0(mod k 2 )}.De?ne a set C 0h of representatives of {?1,1}in C 0as C 0h ={ωj :j =0, k 2 ,2 k 2 ,...,(2t ?1) k 2 }.Since any integer in the congruence class 1modulo 2k (k ?1)lies in the congruence class 1modulo

k (k ?1),by Theorem 2.5,there exists a constant q 0such that if q >q 0and q ≡1(mod 2k (k ?1)),then there exists a DF (q ,k ,1)of the form {μB :μ∈C 0h }for some k -element subset B of GF (q ).De?ne a bijection φof GF (q )onto itself as φ:ν→ω(k 2)ν,ν∈GF (q ).Then,φforms a complementing permutation of the S (2,k ,q )developed from the DF (q ,k ,1)over GF (q ).2

Lemma 2.1employs a strictly halvable GDD.Finite geometry gives an example of such a GDD for prime power block sizes.

Lemma 2.7The points and lines of an af?ne space over GF (q )of dimension 2form a strictly halvable q-GDD of type q q or (q ?1)q +1depending on whether q is even or odd,respectively.

Proof.The points and lines of an af?ne space over GF (q )of dimension 2form an S (2,q ,q 2)(V ,B )over GF (q )×GF (q ).Suppose ?rstly that q is a power of 2.Take a line B through two points (x ,x )and (y ,y +1)for some x ,y ∈GF (q )and let G be its parallel class {B +(z ,z ):z ∈GF (q )}.Then,G ,B \G and the mapping (a ,b )→(a +1,b +1)form groups,blocks and a complementing permutation of the required strictly halvable q -GDD of type q q ,respectively.

Next,assume that q is an odd prime power.It is well-known that (V ,B )admits Z q 2?1,the cyclic group of order q 2?1,as a subgroup of the full automorphism group.Z q 2?1?xes exactly one point,say ∞,and acts regularly on the other points.De?ne a

6

bijectionφof Z q2?1∪{∞}to itself asφ:a→a+1,a∈Z q2?1andφ(∞)=∞.Then, identifying V with Z q2?1∪{∞},the bijectionφforms an automorphism of(V,B).Now all block orbits ofφhave even length q2?1or q+1,and hence taking alternate blocks from each orbit yields a halving of(V,B).By deleting the?xed point,we obtain the required strictly halvable q-GDD of type(q?1)q+1.2 We now return to the proofs of Theorems1.4and1.5.

Proof of Theorem1.4.It is easy to see that for any prime k the conditions v?1≡0 (mod(k?1))and v(v?1)≡0(mod k(k?1))are equivalent to v≡1,k(mod k(k?1)). Hence,if k is a prime,the necessary conditions for the existence of an S(2,k,v)of even

number of blocks are that

v≡1,k(mod k(k?1))and

v

2

/

k

2

is even

?v≡1,k2(mod2k(k?1))

?v≡2ak(k?1)+1,2ak(k?1)+k2(mod k(k2?1))

for a=0,1,...,(k?1)/2.(1) First,we consider the case when the block size is a Mersenne prime and let k be such a prime.Note that n(n2?1)and2an(n?1)+1are relatively prime for every

a=0,1,...,(n?1)/2if and only if n is a Mersenne prime.Hence,by Dirichlet’s theorem on primes in arithmetic progression,we can?nd for each a in?nitely many primes congruent to2ak(k?1)+1modulo k(k2?1).Hence,by Lemma2.6,we can take for each a a prime v a≡2ak(k?1)+1(mod k(k2?1))such that a halvable S(2,k,v a)exists.We assume that every v a is suf?ciently large so that,by Lemma2.2, there exists a constant number w a such that if w>w a and w satis?es(k?1)w+v a≡1,k (mod k(k?1)),then there exists an S(2,k,(k?1)w+v a)having a subsystem of order v a.Take a constant number w >max{w a}such that there exists a TD(k,w)for every integer w>w .Lemma2.7gives a strictly halvable k-GDD of type(k?1)(k+1).Putting all together with Lemma2.1,we have for all w>w satisfying(k?1)w+v a≡1,k(mod k(k?1)),there exists a halvable S(2,k,(k2?1)w+v a).If w≡0,1(mod k),then the condition(k?1)w+v a≡1,k(mod k(k?1))holds.Hence,there is a constant number i a such that if i>i a,then there exist a halvable S(2,k,k(k2?1)i+v a)and a halvable S(2,k,k(k2?1)i+(k2?1)+v a).Write v a by k(k2?1)x a+2ak(k?1)+1.Then we have

k(k2?1)i+v a=k(k2?1)(i+x a)+2ak(k?1)+1and k(k2?1)i+(k2?1)+v a=k(k2?1)(i+x a)+2ak(k?1)+k2. Hence,if v>v0=max{i a}+max{x a}and v satis?es the necessary conditions(1), then there exists a halvable S(2,k,v).

Next,we consider the case when k≤5.Since3is a Mersenne prime and Theorem 1.3gives a halvable S(2,4,v)for all admissible v>616,we only need to consider the case k=5.Since5is a prime number and there exists a TD(5,w)for all integer w>10 (see Abel,Brouwer,Colbourn and Dinitz[1]),by following the above argument,only

7

we have to do is to?nd at least one example of halvable S(2,5,v)in each of congruence classes1,41,81(mod120).By Lemma2.6we can?nd a halvable S(2,k,v)for v≡1,41 (mod120).The DF(81,5,1)over GF(81)given by Beth,Jungnickel and Lenz[5]has the form{ωi A:i=0,1,4,5},whereωis a primitive element and A a certain5-tuple. The corresponding S(2,5,81)can be halved by multiplyingω.The proof is complete.

2 Proof of Theorem1.5.Let k be a power of2.As with the proof of Theorem1.4, the necessary condition of those v for which there exists a halvable S(2,k,v)can be described as v≡2ak(k?1)+1,2ak(k?1)+k2(mod k2(k?1))for a=0,1,...,(k?

1)/2.By using the same technique as in the proof of Theorem1.4,we show that there exists a halvable S(2,k,v)for all v≡2ak(k?1)+1(mod k2(k?1))with a?nite number of possible exceptions.By Dirichlet’s theorem on primes in arithmetic progression and Lemma2.6,we can?nd for each a a prime v a≡2ak(k?1)+1(mod k2(k?1))such that there exists a halvable S(2,k,v a).Since v a≡1(mod k(k?1)),by Lemma2.2,there exists a constant number w a for each a such that if w>w a and w satis?es kw+v a≡1,k (mod k(k?1)),then there exist an S(2,k,kw+v a)having a subsystem of order v a and a TD(k,w).Applying Lemma2.1to these designs with the strictly halvable k-GDD of type k k given by Lemma2.7,we obtain a halvable S(2,k,k2w+v a)for all w>w a satisfying kw+v a≡1(mod k(k?1)).When w is divisible by k?1,the condition kw+v a≡1(mod k(k?1))holds.Hence,there exists a constant number i a such that if i>i a,then there exists a halvable S(2,k,k2(k?1)i+v a).Write v a by k2(k?1)x a+ 2ak(k?1)+1,and we have k2(k?1)i+v a=k2(k?1)(i+x a)+2ak(k?1)+1.Hence, if v>v0=max{i a}+max{x a}and v≡1(mod2k(k?1)),then there exists a halvable S(2,k,v).2 Lemma2.7generates no strictly halvable GDD of a nonprime power block size. The following deals with more general block sizes.

Lemma2.8If there exists a halvable S(2,k,2v+1)or a halvable S(2,k,2v),then there exists a strictly halvable k-GDD of type(k?1)(2v)/(k?1)or type12v,respectively. Proof.Since a complementing permutation of a self-complementary graph on an odd number of vertices?xes exactly one vertex,a complementing permutation of a halvable S(2,k,2v+1)S also?xes one point.By deleting the?xed point from S,we obtain a k-GDD of type(k?1)(2v)/(k?1).

Every cycle of a complementing permutation of a self-complementary graph on even number vertices is divisible by four,and hence it is even.Regarding a halvable S(2,k,2v)as a k-GDD of type12v,we obtain the required GDD.2 Lemma2.2and2.6admit any block size greater than?ve.Also,Lemma2.6gen-erates a halvable Steiner2-design,and hence we obtain from Lemma2.8a strictly halvable k-GDD for any integer k≥6.Hence,applying Lemma2.1to these designs, we can obtain in?nite classes of halvable S(2,k,v)s for any block size k≥6.

In general,the constant number v0in the proof of Theorem1.4is large.However, we can reduce the number for particular k to being relatively small.To lower the con-stant,known results on subsystems and difference families are useful.In fact,Phelps

8

[11]proved Theorem1.3by using the results on subsystems due to Rees and Stinson

[12].The following is known as the Doyen–Wilson theorem.

Theorem2.9(Doyen and Wilson)[9]There exists an S(2,3,v)which has a subsystem of order w whenever v≡1,3(mod6),w≡1,3(mod6)and v≥2w+1.

Combining Theorems1.1,1.2,2.9,Lemmas2.1and2.8,we show the following. Theorem2.10There exists a halvable Steiner triple system of order v for all v≡1,9 (mod12)except possibly v=21.

Proof.We only need to consider the congruence classes13,21modulo24.Applying Lemma2.8to the unique S(2,3,9),which is halvable,we obtain a strictly halvable3-GDD of type24.By Theorem1.2,there exists a halvable S(2,3,13).Also,by Theorem

2.9,there exists an S(2,3,2w+13)having a subsystem of order13for all w≡0,1(mod

3)and w≥7.Since a TD(3,w)also exists for every positive integer w≥7,by applying Lemma2.1,we obtain a halvable S(2,3,v)for all v≡13,21(mod24)and v>61. Theorem1.2provides halvable Steiner triple systems of order37and61.To complete the proof we construct a halvable S(2,3,45)over GF(9)×Z5.Letωbe a primitive element of GF(9)such thatω3=ω2+1.De?ne a set B of triples on GF(9)×Z5 as B={{(0,4),(1,0),(ω3,1)},{(0,1),(1,0),(ω3,4)},{(0,4),(0,1),(ω3,0)}}.It is

straightforward to see that B∪(ω2,2)·B forms a block set of a3-GDD of type95 with groups G={{a×{i}:a∈GF(9)}:i∈Z5}.Filling each group with the lines of AG(2,3),we obtain a halvable S(2,3,45)with a complementing permutationφ:x→(ω2,2)·x.2 Similarly,the analog of the Doyen–Wilson theorem given by Able,Ge,Greig and Ling[2](see also Bennett,Chang,Ge and Greig[3])and the complete resolution of the existence of DF(q,5,1)over GF(q)due to Chen and Zhu[7]can considerably reduce the v0in Theorem1.4for k=5.Furthermore,this technique with the recent results on Steiner2-designs over cyclic groups due to Chen and Wei[6]completely settle the halving problem for S(2,4,v)s.

An S(2,k,v)is said to be cyclic if its full automorphism group contains the cyclic group Z v as a subgroup acting regularly on the point set.Each block orbit of a cyclic S(2,k,v)has length v or v/k.For the case when the block size k=4and the order v≡16(mod24),these two values v and v/k are both even,and hence every cyclic S(2,4,v),v≡16(mod24),is halvable(see Das and Rosa[8]).

Theorem2.11There exists a halvable S(2,4,v)if and only if v≡1,16(mod24). Proof.We only need to remove the seven exceptions indicated in Theorem1.3.Every cyclic S(2,4,40)has the unique orbit of length10forming a parallel class,and hence it can be considered as a strictly halvable4-GDD of type410.There exist an S(2,4,76) having a subsystem of order16,a halvable S(2,4,16)and TD(4,15).Applying Lemma 2.1to these designs,we obtain a halvable S(2,4,616).All the smaller exceptions are covered with cyclic Steiner2-designs constructed by Chen and Wei[6].2 In the remainder of this paper,we mention some recursive constructions for halv-able S(2,k,v)s.

9

Theorem2.12If there exist a halvable S(2,k,v),a halvable S(2,k,2w+1)and a TD(k,v),then there exists a halvable S(2,k,v(2w+1)).

Proof.Let(V,B)and(W,C)be two halvable Steiner2-designs,where|V|=v,|W|= 2w+1and bijectionsφ,πpartition B,C into B=B1∪φ(B1),C=C1∪π(C1),re-spectively.For each C∈C1,let(V×C,G C,D C)be a TD(k,v)so that G C has the form{(a,x),(b,x),...,(c,x):{a,b,...,c}=V,x∈C}.For each x∈W,place a fac-tor(V,B1)onto the point set V×{x}and name its block set B1x.De?ne a bijec-tionτof V×W onto itself asτ:(i,x)→(φ(i),π(x))for(i,x)∈V×W.Then, D={D C∪B1x:C∈C1,x∈W}forms a half of an S(2,k,v(2w+1)).The image of D byτcompletes the remaining half.2

Combining Lemma2.1with Lemma2.8,we have the following corollaries similar to the construction above.

Corollary2.13If there exist a halvable S(2,k,2v),an S(2,k,w+u)having a sub-system of order u,a halvable S(2,k,u)and a TD(k,w),then there exists a halvable S(2,k,2vw+u).

Corollary2.14If there exist a halvable S(2,k,2v),an S(2,k,w)and also a TD(k,w), then there exists a halvable S(2,k,2vw).

Corollary2.15If there exist a halvable S(2,k,2v+1),an S(2,k,w(k?1)+u)having a subsystem of order u,a halvable S(2,k,u)and a TD(k,w),then there exists a halvable S(2,k,2vw+u).

Acknowledgements

The author would like to thank Professor Masakazu Jimbo,Professor Junya Satoh and anonymous referees for helpful comments and valuable suggestions.In particular,the referees’suggestions improved the proof of Lemma2.1.One of the referees pointed out the existence of a halvable S(2,4,616)and the recent results by Abel,Ge,Greig and Ling[2].The author also thanks the referee for giving him an essential idea for the construction of the halvable S(2,3,45)in the proof of Theorem2.10.Research of the author is supported by JSPS Research Fellowships for Young Scientists.

References

[1]R.J.R.Abel,A.E.Brouwer,C.J.Colbourn,J.H.Dinitz,Mutually orthogonal

Latin squares,in: C.J.Colbourn,J.H.Dinitz(Eds.),The CRC Handbook of Combinatorial designs,CRC Press,Boca Raton,FL,1996,pp.111–142. [2]R.J.R.Abel,G.Ge,M.Greig,A.C.H.Ling,Further results on(v,{5,w?},1)-

PBDs,preprint.

[3]F.E.Bennett,Y.Chang,G.Ge,M.Greig,Existence of(v,{5,w?},1)-PBDs,Dis-

crete Math.279(2004),61–105.

10

[4]T.Beth,Eine Bemerkung zur Absch¨a tzung der Anzahl orthogonaler lateinischer

Quadrate mittels Siebverfahren,Abh.Math.Sem.Univ.Hamburg53(1983),284–288.

[5]T.Beth,D.Jungnickel,H.Lenz,Design theory,Cambridge University Press,

Cambridge,England,1999.

[6]K.Chen,R.Wei,A few more cyclic Steiner2-designs,https://www.360docs.net/doc/034310937.html,bin.13

(2006),Research Paper10,13pp.(electronic).

[7]K.Chen,L.Zhu,Existence of(q,k,1)difference families with q a prime power

and k=4,5,https://www.360docs.net/doc/034310937.html,bin.Des.7(1999),21–30.

[8]P.K.Das,A.Rosa,Halving Steiner triple systems,Discrete Math.109(1992),

59–67.

[9]J.Doyen,R.M.Wilson,Embeddings of Steiner triple systems,Discrete Math.5

(1973),229–239.

[10]D.Fronˇc ek,M.Meszka,Halving transversal designs,https://www.360docs.net/doc/034310937.html,bin.Des.8(2000),

83–99.

[11]K.T.Phelps,Halving block designs with block size four,https://www.360docs.net/doc/034310937.html,bin.3

(1991),231–234.

[12]R.S.Rees,D.R.Stinson,On the existence of incomplete block designs of block

size four having one hole,Utilitas Math.35(1989),119–152.

[13]R.M.Wilson,An existence theorem for pairwise balanced designs,I:Composi-

tion theorems and morphisms,https://www.360docs.net/doc/034310937.html,bin.Theory Ser.A13(1972),220–245. [14]R.M.Wilson,An existence theorem for pairwise balanced designs,II:The struc-

ture of PBD-closed set and the existence conjecture,https://www.360docs.net/doc/034310937.html,bin.Theory Ser.A13 (1972),246–273.

[15]R.M.Wilson,An existence theorem for pairwise balanced designs,III:Proof of

the existence conjecture,https://www.360docs.net/doc/034310937.html,bin.Theory Ser.A18(1975),71–79.

[16]R.M.Wilson,Concerning the number of mutually orthogonal Latin squares,Dis-

crete Math.9(1974),181–198.

[17]R.M.Wilson,Cyclotomy and difference families in elementary abelian groups,

J.Number Theory4(1972),17–47.

11

二叉树习题及答案

1.设一棵完全二叉树共有699 个结点,则在该二叉树中的叶子结点数? 1根据二叉树的第i层至多有2A(i - 1)个结点;深度为k的二叉树至多有2A k - 1 个结点(根结点的深度为1)”这个性质: 因为2A9-1 < 699 < 2A10-1 , 所以这个完全二叉树的深度是10,前9 层是一个满二叉树, 这样的话,前九层的结点就有2A9-1=511 个;而第九层的结点数是2A(9-1)=256 所以第十层的叶子结点数是699-511=188 个;现在来算第九层的叶子结点个数。由于第十层的叶子结点是从第九层延伸的,所以应该去掉第九层中还有子树的结点。因为第十层有188 个,所以应该去掉第九层中的188/2=94 个;所以,第九层的叶子结点个数是256-94=162,加上第十层有188 个,最后结果是350 个 2完全二叉树:若二叉树中最多只有最下面两层的结点的度可以小于2,并且最下面一层的结点 (叶结点) 都依次排列在该层最左边的位置上,这样的二叉树为完全二叉树。 比如图:完全二叉树除叶结点层外的所有结点数(叶结点层以上所有结点数)为奇数,此题中,699 是奇数,叶结点层以上的所有结点数为保证是奇数,则叶结点数必是偶数,这样我们可以立即选出答案为B!如果完全二叉树的叶结点都排满了,则是满二叉树,易得满二叉树的叶结点数是其以上所有层结点数+1 比如图: 此题的其实是一棵满二叉树,我们根据以上性质,699+1=700,700/2=350,即叶结点数为350,叶结点层以上所有结点数为350-1=349。 3完全二叉树中,只存在度为2 的结点和度为0 的结点,而二叉树的性质中有一条是: nO=n2+1 ; nO指度为0的结点,即叶子结点,n2指度为2的结点,所以2n2+1=699 n2=349 ; n0=350 2.在一棵二叉树上第 5 层的结点数最多是多少一棵二叉树,如果每个结点都是是满的,那么会满足2A(k-1)1 。所以第5 层至多有2A(5-1)=16 个结点! 3.在深度为5 的满二叉树中,叶子结点的个数为答案是16 ~ 叶子结点就是没有后件的结点~ 说白了~ 就是二叉树的最后一层~ 深度为K 的二叉树~ 最多有2Ak-1 个结点~ 最多有2A(k-1) 个结点~ 所以此题~ 最多有2A5-1=31 个结点~ 最多有2A(5-1)=16 个叶子结点~ 4.某二叉树中度为2 的结点有18 个,则该二叉树中有几个叶子结点?结点的度是指树中每个结点具有的子树个数或者说是后继结点数。 题中的度为2 是说具有的2 个子树的结点;二叉树有个性质:二叉树上叶子结点数等于度为2 的结点数加1。 5.在深度为7 的满二叉树中,度为2 的结点个数为多少,就是第一层只有一个节点,他有两个子节点,第二层有两个节点,他们也都有两个子节点以此类推,所以到第6 层,就有2的5次方个节点,他们都有两个子节点最后第7 层都没有子节点了。因为是深度为7 的。 所以就是1+2+4+8+16+32 了 2深度为1的时候有0个 深度为2的时候有1个 深度为3的时候有3个 深度为4的时候有7个 深度为n的时候有(2的n-1次方减1 )个 6?—棵二叉树中共有70个叶子结点与80个度为1的结点,则该二叉树中的总结点数为?

(完整版)逻辑思维方法训练及技法训练大全,推荐文档

怎样提高逻辑思维能力 一、逻辑思维的概念 “逻辑”,或称为“理则”。源自古典希腊语λ?γο?(logos),最初的意思有词语、思想、概念、论点、推理之意。1902年严复译《穆勒名学》,将其 意译为“名学”,音译为“逻辑”;日语则译为“论理学”。在现代汉语词典里, 逻辑的涵义是思维的规律或客观的规律性,逻辑学被定义为研究思维形式和规 律的科学。 逻辑思维(Logical thinking),人们在认识过程中借助于概念、判断、推理等思维形式能动地反映客观现实的理性认识过程,又称理论思维。它是作为 对认识着的思维及其结构以及起作用的规律的分析而产生和发展起来的。只有 经过逻辑思维,人们才能达到对具体对象本质规定的把握,进而认识客观世界。 它是人的认识的高级阶段,即理性认识阶段。 二、逻辑思维的方法分类 (一)、系统思维法: 1.系统结构: 系统的上下级是归属关系,同级之间是并列关系。 例如:某所高中系统,分为高一、高二、高三这三个子系统,其中高一这个子系统又分为一班、二班。可见,系统的上下级之间(年级和班级)是归属 关系,同级之间(年级之间或班级之间)是并列关系。 例如:人体由八大系统组成,既运动系统、神经系统、内分泌系统、循环系统、呼吸系统、消化系统、泌尿系统、生殖系统。 其中的消化系统又由消化管和消化腺这两个子系统组成。 其中的消化管又由口腔、咽、食管、胃、小肠、大肠这些更小的系统组成。 其中的小肠又由十二指肠、空肠、回肠这些更更小的系统组成。 2.系统中,同级的事物之间的关系: 系统中同级的事物之间,如果存在相互的关系,通常按组织结构分为合作和对立两种,按变化过程,分为因果和阶段两种。 (1)合作关系。 例如:餐馆是一个系统,里面的厨师、服务员、老板他们相互合作。 例如:消化系统中的胃和小肠是合作关系,都是在消化食物。 (2)对立关系。 例如:全国足球联赛是一个系统,里面两只比赛的足球队是对立关系。 例如:激素系统中的胰岛素和胰高血糖素是对立关系,胰岛素降低血糖,胰高血糖素升高血糖。 合作关系与对立关系的事例:在一个群落系统中,羚羊之间是合作关系,一批羚羊休息时,另一批羚羊要放哨,而羚羊和狮子是对立关系。 (3)因果关系。 我去人也就有人!为UR扼腕入站内信不存在向你偶同意调剖沙

定量分析方法总结

一、灰色关联分析 灰色关联分析是系统态势的一种量化比较分析,其实质就是比较若干数列所构成的曲线到理想数列所构成的曲线几何形状的接近程度,几何形状越接近,其关联度就越大。可见,灰色关联分析是一种趋势分析,它对样本的大小没有太高的要求,一般情况下比较适合小样本,贫信息的数据,并且样本数据不需要典型的分布规律,因而,具有广泛的适用性。 灰色关联分析模型的建立: (1)确定比较数列与参考数列; 设Xi={xi(1),xi (2),…xi(n)}为创业板上市公司的财务指标形成的比较数据列,其中,i=1,2…17.同时,把每项指标中的最优值作为最优指标集X0,可得到参考数列:X 0={x 0(1),x 0(2),…x 0(n)} (2)无量纲化处理;无量纲化的处理方法通常有初值化、均值化、规范化三种方法,而本文采用的是不同指标的标准化处理方法,如前文所示。 (3)各个指标权重的确定w (k ); (4)计算关联系数δi(k); (5)计算关联度r i 设参考数列为:X 0={x 0(1),x 0(2),…x 0(n)},关联分析中被比较数列记为X i ={x i (1),x i (2),…x i (n)},i=1,2,…28;n=1,2,3…12. 对于一个参考数列X 0,比较数列Xi ,可用下述关系表示各比较曲线与参考曲线在各点的差: |(k)x -(k) x |ρm ax m ax |(k)x -(k)x | | (k)x -(k)x |m ax m ax ρ |(k)x -(k)x |m inm in (k)δi o i o i o i o i ++=

式中,δi(k)是第k 个时刻比较曲线x i 与参考曲线x o 的相对差值,这种形式的相对差值称为x i 对x 0在k 时刻的关联系数。ρ为分辨系数,ρ∈(0,1),引入它是为了减少极值对计算的影响。在实际计算使用时,一般取ρ=0.5. 若记:Δmin=minmin|x o (k)-x i (k)|, Δmax= maxmax|x o (k)-x i (k)|,则Δmin 与Δmax 分别为各时刻x o 与x i 的最小绝对差值与最大绝对差值,从而有 ρΔm ax |x -x |ρΔm ax Δm in δi(k)0(k))k i(++= 根据关联系数计算关联度,得到灰色关联模型为: r i = ∑=n 1i )(*)(k w k i δ 二、层次分析法构建经营绩效评价模型 层次分析法(Analytic Hierarchy Process 简称AHP)是美国运筹学家匹茨堡大学教授Saaty 于二十世纪70年代初期提出的。层次分析法(AHP ),它是系统工程中对非定量事件作定量分析的一种简便方法,也是人们对主观判断进行客观描述的一种有效方法。它将复杂问题分解成若干个层次,逐步进行分析。这种做法,首先要求把问题层次化,根据问题的性质和要得到的目标,将问题分解为不同的组合因素,并将问题按不同的层次聚集组合,形成一个多层次的分析结构模型。通过两两比较的方法,确定层次中诸因素的相对重要性,然后组合人们的判断以决定诸因素相对于总目标的相对重要性数值或相对优劣次序的排序。 层次分析法的核心思想可以归纳为“先分解后综合”,应用层次分析法进行上市公司经营绩效评价进,应包括如下基本步骤[27]: (1)建立层次结构 应用层次分析法进行综合经营绩效评价时,首先建立评价问题的层次结构(Hierarchy)。层次结构是应用层次分析法把复杂问题分解简化的关键,必须建立在对决策问题深刻分析和对决策目标以及决策主体意图的充分理解之上。层次结构的建立过程是首先确定决策目标,其次罗列出与该目标相关的各种因素,然后

二叉树的4个普遍性质和2个特殊性质的完善推导过程

二叉树的5个性质 1、在二叉树的第k 层上,最多有2k-1(k ≥1)个结点 证明:在二叉树的第i 层上最多有2 i-1 个节点 1层 1个 20 2层 2个 21 3层 4个 22 ..... i 层 2 i-1个 2、二叉树中如果深度为k,那么最多有2k -1个节点 证明:在具有相同深度的二叉树中,仅当每一层都含有最大结点数时,其树中结点数最多。因此利用性质1可得,深度为k 的二叉树的结点数至多为: 20+21+…+2k-1=2k -1 故命题正确。 3、在任意一棵二叉树中,若终端结点的个数为n 0,度为2的结点数为n 2,则n o =n 2+1。 . 证明:n 0=n 2+1 n 0表示度数为0的节点 n 2表示度数为2的节点 推导过程 根据两个公式 1). n=n 0+n 1+n 2 n 表示二叉树中的节点总个数,n 1表示度数为1的节点个数 2). n-1=2n 2+n 1 通过观察二叉树我们可知,除了根节点之外,其余的任何节点都有一个入口分支(或其他节点都有一个入口分支),那么节点的总分支数等于节点个数减一,度数为2的节点有2个出口分支,度数为一的有1个出口分支,度数为0的节点没有出口分支 所以总的分支个数为 2n 2+n 1,因此有n=2n 2+n 1+1, 3).比较n=n 0+n 1+n 2和n=2n 2+n 1+1两式,可得n 0=n 2+1。 5.在完全二叉树中,具有n 个节点的完全二叉树的深度为[log2n]+1,其中[log2n]+1是向下取整。 证明: 根据性质 2: 假设深度为k 的满二叉树的节点个数一定为2k -1,那么n=2k -1推得满二叉树的深度数为k=log 2(n+1);——深度为m 的二叉树最多有2k -1个节点,即是满二叉树的情形。 设完全二叉树是具有n 个节点的二叉树,若按层序编号那么其编号与同样深度的满二叉() 11122111n m m n a q S q --===---() 11111220n k k n a a q k ---==?=≥

定性与定量分析

定性--用文字语言进行相关描述 定量--用数学语言进行描述 定性分析与定量分析应该是统一的,相互补充的;; 定性分析是定量分析的基本前提,没有定性的定量是一种盲目的、毫无价值的定量;; 定量分析使之定性更加科学、准确,它可以促使定性分析得出广泛而深入的结论 定量分析是依据统计数据,建立数学模型,并用数学模型计算出分析对象的各项指标及其数值的一种方法。定性分析则是主要凭分析者的直觉、经验,凭分析对象过去和现在的延续状况及最新的信息资料,对分析对象的性质、特点、发展变化规律作出判断的一种方法。相比而言,前一种方法更加科学,但需要较高深的数学知识,而后一种方法虽然较为粗糙,但在数据资料不够充分或分析者数学基础较为薄弱时比较适用,更适合于一般的投资者与经济工作者。因此,本章以后几节所做的分析基本上以定性分析为主。但是必须指出,两种分析方法对数学知识的要求虽然有高有低,但并不能就此把定性分析与定量分析截然划分开来。事实上,现代定性分析方法同样要采用数学工具进行计算,而定量分析则必须建立在定性预测基础上,二者相辅相成,定性是定量的依据,定量是定性的具体化,二者结合起来灵活运用才能取得最佳效果。 不同的分析方法各有其不同的特点与性能,但是都具有一个共同之处,即它们一般都是通过比较对照来分析问题和说明问题的。正是通过对各种指标的比较或不同时期同一指标的对照才反映出数量的多少、质量的优劣、效率的高低、消耗的大小、发展速度的快慢等等,才能为作鉴别、下判断提供确凿有据的信息。 应用: 在证据法学研究中,定性分析方法和定量分析方法各有长处,可以相辅相成。但是由于我国证据法学的研究人员比较熟悉定性分析方法,所以有必要特别强调定量分析方法的功能和重要性。例如,我们不仅要分析某个证据规则是好还是不好,而且要分析其利弊比例……等等 专利分析法分为定量分析和定性分析两种。定量分析即对专利文献的外部特征(专利文献的各种著录项目)按照一定的指标(如专利数量)进行统计,并对有关的数据进行解释和分析。定性分析是以专利的内容为对象,按技术特征归并专利文献,使之有序化的分析过程。通常情况下需要将二者结合才能达到较好的效果。 定性分析与定量分析应该是统一的,相互补充的;定性分析是定量分析的基本前提,没有定性的定量是一种盲目的、毫无价值的定量;定量分析使定性分析更加科学、准确,它可以促使定性分析得出广泛而深入的结论。 定量分析是依据统计数据,建立数学模型,并用数学模型计算出分析对象的各项指标及其数值的一种方法。 定性分析则是主要凭分析者的直觉、经验,凭分析对象过去和现在的延续状况及最新的信息资料,对分析对象的性质、特点、发展变化规律作出判断的一种方法。相比而言,前一种方法更加科学,但需要较高深的数学知识,而后一种方法虽然较为粗糙,但在数据资料不够充分或分析者数学基础较为薄弱时比较适用,更适合于一般的投资者与经济工作者。但是必须指出,两种分析方法对数学知识的要求虽然有高有低,但并不能就此把定性分析与定量分析截然划分开来。事实上,现代定性分析方法同样要采用数学工具进行计算,而定量分析则必须建立在定性预测基础上,二者相辅相成,定性是定量的依据,定量是定性的具体化,二者结合起来灵活运用才能取得最佳效果。 不同的分析方法各有其不同的特点与性能,但是都具有一个共同之处,即它们一般都是通过比较对照来分析问题和说明问题的。正是通过对各种指标的比较或不同时期同一指标的对照才反映出数量的多少、质量的优劣、效率的高低、消耗的大小、发展速度的快慢等等,才能作为鉴别、下判断提供确凿有据的信息。数学的时候,才能称得上是一门科学。数学的时候,才能称得上是一门科学。 我所接触的稿件基本上都是运用科技统计数字作定量分析的。按常规推理,这种定量分析有扎实的统计数

习题8(二叉树的定义和性质)

习题8(二叉树的定义和性质) 一、选择题 1、除个别结点外,其余结点只能有1个前驱结点,可有任意多个后继结点,这样的结构为( B )。 A)线性结构 B)树形结构 C)图形结构 D)拓扑结构 2、在下述结论中,正确的是( D )。 ①只有一个结点的二叉树的度为0 ②二叉树的度为2 ③二叉树的左右子树可任意交换 ④深度为K的完全二叉树的结点个数小于或等于深度相同的满二叉树 A)①②③ B)②③④ C)②④ D)①④ 3、下列有关树的概念错误的是( B )。 A)一颗树中只有一个无前驱的结点 B)一颗树的度为树中各个结点的度数之和 C)一颗树中,每个结点的度数之和等于结点的总数减1 D)一颗树中每个结点的度数之和与边的条数相等4、对任一颗树,设它有n个结点,这n个结点的度数之和为d,下列关系式正确的是( D )。 A)d=n B)d=n-2 C)d=n+1 D)d=n-1 5、下列说法中正确的是( D )。 A)二叉树中任何一个结点的度都为2 B)二叉树的度为2 C)任何一棵二叉树中至少有一个结点的度为2 D)一棵二叉树的度可以小于2 6、以二叉链表作为二叉树的存储结构,在具有n个结点的二叉链表中(n>0),空链域的个数为( C )。 A)2n-1 B)n-1 C)n+1 D)2n+1 7、树最适合用来表示( C )。 A)有序数据元素 B)无序数据元素 C)元素之间具有分支层次关系的数据 D)元素之间无联系的数据8、由4个结点可以构造出多少种不同的二叉树( C )。 A)4 B)5 C)14 D)15 9、一个二叉树具有( A )种基本形态。 A)5 B)4 C)3 D)2 10、二叉树的第I层上最多含有结点数为( C )。 A)2I B)2I-1-1 C)2I-1 D)2I -1 11、深度为5的二叉树至多有( C )个结点. A)16 B)32 C)31 D)10 12、一个满二叉树,共有n个结点,其中m个为树叶,则( B )。 A)n= m+1 B)m=( n +1)/2 C)n =2 m D)n =2 m 13、深度为h的满m叉树的第k层有( A )个结点。(1=

常用的逻辑思维方法技巧

常用的逻辑思维方法技巧 例如:一切化学元素在一定条件下发生化学反应。 惰性气体是化学元素,所以,惰性气体在一定条件下确实能够发生化学反应。 这里运用的就是演绎推理方法。 演绎推理的主要形式是三段论法。 三段论法就是从两个判断中进而得出第三个判断的一种推理方法。 上面的例子就是包含着三个判断。 第一个判断是一切化学元素都在一定条件下发生化学反应"-提供了一般的原理原则,叫做三段论式的大前提。 第二个判断是"惰性气体是化学元素"--指出了一种特殊情况,叫做小前提。 联合这两种判断,说明一般原则和特殊情况间的联系,因而得出第三个判断:"惰性气体在一定条件下确定能够发生化学反应"--结论。 只要作为前提的判断是正确的,中间的推理形式是合乎逻辑规则的,那么,必然能够推出“隐藏在前提中的知识,这种知识,尽管没有超出前提的范围,但毕竟从后台走到了前台,对我们来说,往往也是新的,而且由于我们常常是为了某种实际需要才做这种推理,其结论很可能具有应用价值。 这样演绎推理的结论就可能既具有新颖性,又具有实用性。 2.归纳推理法1)完全归纳推理从一般性较小的知识推出一般性较

大的知识的推理,就是归纳推理。 在许多情况下,运用归纳推理可以得到新的知识。 按照一定的目标,运用归纳推理的思维方法,取得新颖性结果的过程,就是归纳推理法。 1000只大象都是灰色,第1001只大象为白色的可能性总是存在。 所有的金属都有导电性,但电阻大小一样,使用场合与效果也不一样,用铁丝充当保险丝就难以胜任,强而为之,后患无穷。 2)简单枚举归纳推理简单枚举归纳推理是列举某类事物中一部分对象的情况,根据没有遇到矛盾的情况,便做出关于这一类事物的一般性结论的推理。 例如:花开的时间、天鹅的颜色简单枚举归纳推理的意义虽然它的结论是或然的,但不一定是错误的,有的是正确的,也就可以提供新的知识。 在它的结论的基础上,可以继续研究,如果证明是正确的,就得到了新的知识。 即使证明了是错误的,也从另一方面给了我们新的知识。 两种不完全归纳推理的区别:它们的根据不同,前者只要没有发现矛盾的情况就可以做出结论,后者要根据发现的因果之间的必然联系才能下结论前者的结论是或然性的,后者的结论要可靠的多。 提高前者结论的办法是多找事实,提高后者的结论是对事实情况作出科学的分析,找出因果关系。

定量分析方法 重点整理

1、公共管理:是一门研究公共组织尤其是政府组织的管理活动及其规律的学科。公共管理研究的内容:①公共组织的结构、功能、环境和运行机制;②行政管理体制改革、中央与地方的关系;③市场经济条件下政府的职能与作用、政府与市场、政府与企业、政府与社会的关系;④公共人力资源的开发与利用;⑤公共管理中的规划、计划与决策、监督与控制,公共项目评估,行政立法、司法和执法;⑥公共信息管理和咨询服务;⑦财政管理、教育管理、科技管理和文化管理。 2、定量分析方法的主要内容 系统模型与系统分析、线性回归预测分析、社会调查程序与方法、统计分析方法、线性回归预测分析、马尔可夫预测方法、投入产出分析方法、最优化方法(线性规划、运输问题、动态规划、资源分配问题)、评价分析方法、层次分析法、对策论、风险型决策与多目标决策、管理系统模拟、排队论、系统动力学方法、网络计划方法 3、为什么在系统分析中广泛使用系统模型而不是真实系统进行分析?人类认识和改造客观世界的研究方法,一般有实验法和模型法。实验法是通过对客观事物本身直接进行科学实验来进行研究的,因此局限性比较大。公共管理问题大多是难以通过实验法直接进行研究,广泛使用系统模型还基于以下五个方面的考虑:①系统开发的需要只能通过建造模型来对系统或体制的性能进行预测;②经济上的考虑对复杂的社会经济系统直接进行实验,成本十分昂贵;③安全性、稳定性上的考虑对有些问题通过直接实验进行分析,往往缺乏安全性和稳定性,甚至根本不允许;④时间上的考虑使用系统模型很快就可得到分析结果;⑤系统模型容易操作,分析结果易于理解 4、系统分析的要点和步骤 要点(1)任务的对象是什么?即要干什么(what);(2)这个任务何以需要?即为什么这样干(why);(3)它在什么时候和什么样的情况下使用?即何时干(when);(4)使用的场所在哪里?即在何处干(where);(5)是以谁为对象的系统?即谁来干(who);(6)怎样才能解决问题?即如何干(how)。步骤(1)明确问题与确定目标。当一个有待研究分析的问题确定以后,首先要对问题进行系统的合乎逻辑的阐述,其目的在于确定目标,说明问题的重点与范围,以便进行分析研究。(2)搜集资料,探索可行方案。在问题明确以后,就要拟定解决问题的大纲和决定分析方法,然后依据已搜集的有关资料找出其中的相互关系,寻求解决问题的各种可行方案。(3)建立模型。为便于对各种可行方案进行分析,应建立各种模型,借助模型预测每一方案可能产生的结果,并根据其结果定性或定量分析各方案的优劣与价值。(4)综合评价。利用模型和其他资料所获得的结果,对各种方案进行定性与定量相结合的综合分析,显示出每一种方案的利弊得失和效益成本,同时考虑到各种有关因素,如政治、经济、军事、科技、环境等,以获得对所有可行方案的综合评价和结论。(5)检验与核实。 5、简述霍尔三维结构与切克兰德“调查学习”模式之间的区别。 1)霍尔三维结构将系统的整个管理过程分为前后紧密相连的六个阶段和七个步骤,并同时考虑到为完成这些阶段和步骤的工作所需的各种专业管理知识。三维结构由时间维、逻辑维、知识维组成。霍尔三维结构适用于良结构系统,即偏重工程、机理明显的物理型的硬系统。2)切克兰德“调查学习”模式的核心不是寻求“最优化”,而是“调查、比较”或者说是“学习”,从模型和现状比较中,学习改善现存系统的途径,其目的是求得可行的满意解。适用于不良结构系统,偏重社会、机理尚不清楚的生物型的软系统。3)处理对象不同:前者为技术系统、人造系统,后者为有人参与的系统;4)处理的问题不同:前者为明确、良结构,后者为不明确,不良结构;5)处理的方法不同:前者为定量模型,定量方法,后者采用概念模型,定性方法;6)价值观不同:前者为一元的,要求优化,有明确的好结果(系统)出现,后者为多元的,满意解,系统有好的变化或者从中学到了某些东西。 6、定性分析的方法:目标--手段分析法、因果分析法、KJ 分析法 7、社会调查的含义:是人们有意识、有目的地通过对社会现象的考察、了解和分析,来认识社会生活的本质机器发展规律的实践活动和认识活动。 基本原则①客观性原则,核心是实事求是,这是社会调查的立足点和出发点;②实证性原则,要求社会调查的结论以及与此相关的各种观点,都必须有真实、可靠的疏忽和资料做支持;③系统性原则,要求对社会现象要进行系统、综合的分析和研究。 8、预测分析的一般步骤①明确预测目标;②收集、整理资料和数据;③建立预测模型;④模型参数估计;⑤模型

各类型二叉树例题说明

5.1树的概念 树的递归定义如下:(1)至少有一个结点(称为根)(2)其它是互不相交的子树 1.树的度——也即是宽度,简单地说,就是结点的分支数。以组成该树各结点中最大的度作为该树的度,如上图的树,其度为3;树中度为零的结点称为叶结点或终端结点。树中度不为零的结点称为分枝结点或非终端结点。除根结点外的分枝结点统称为内部结点。 2.树的深度——组成该树各结点的最大层次,如上图,其深度为4; 3.森林——指若干棵互不相交的树的集合,如上图,去掉根结点A,其原来的二棵子树T1、T2、T3的集合{T1,T2,T3}就为森林; 4.有序树——指树中同层结点从左到右有次序排列,它们之间的次序不能互换,这样的树称为有序树,否则称为无序树。 5.树的表示 树的表示方法有许多,常用的方法是用括号:先将根结点放入一对圆括号中,然后把它的子树由左至右的顺序放入括号中,而对子树也采用同样的方法处理;同层子树与它的根结点用圆括号括起来,同层子树之间用逗号隔开,最后用闭括号括起来。如上图可写成如下形式: (A(B(E(K,L),F),C(G),D(H(M),I,J))) 5. 2 二叉树 1.二叉树的基本形态: 二叉树也是递归定义的,其结点有左右子树之分,逻辑上二叉树有五种基本形态: (1)空二叉树——(a); (2)只有一个根结点的二叉树——(b); (3)右子树为空的二叉树——(c); (4)左子树为空的二叉树——(d); (5)完全二叉树——(e) 注意:尽管二叉树与树有许多相似之处,但二叉树不是树的特殊情形。 2.两个重要的概念: (1)完全二叉树——只有最下面的两层结点度小于2,并且最下面一层的结点都集中在该层最左边的若干位置的二叉树; (2)满二叉树——除了叶结点外每一个结点都有左右子女且叶结点都处在最底层的二叉树,。 如下图: 完全二叉树 1页

个人逻辑思维能力需求分析资料

个人逻辑思维能力需 求分析

《现代人力资源培训与开发》课程作业题目:个人逻辑思维能力需求分析 姓名:王天渝 学号:20143350135 专业:人力资源管理 班级:1班

思考的深度决定一个人思想的深度,逻辑思维能力的强弱则决定着一个人思考的深度。做一个有深度的思想者才能发现本质而非表面,才能冷静客观地分析各种现象而不是被现实蒙蔽了双眼。思考是思想的出口,逻辑则是打开出口大门的钥匙。用逻辑来学会思考,是每一个自由的个体要拥有的能力。 在这篇对自我逻辑思维能力的需求分析报告中,将采用不同的维度对我进行从大学这两年来度过的“逻辑无意识”(without consciousness of using logic)的学习生活进行反思及对逻辑思维能力的需求进行分析,试图在形成“逻辑意识”的基础上,规划今后逻辑思维训练的方案。 一,逻辑思维能力的需求分析模型 (1)关于逻辑思维能力的体现: 根据逻辑思维能力的定义:逻辑思维能力是指正确、合理思考的能力。即对事物进行观察、比较、分析、综合、抽象、概括、判断、推理的能力,采用科学的逻辑方法,准确而有条理地表达自己思维过程的能力。逻辑思维更多的是一种技术,需要进行长时间的有效的训练,才能掌握这门技术,它帮助我们在日常生活中更好地去表达、思考和学习(如图1) 逻辑思维能力1,表达思路:思维分块化的基础上建立逻辑链。表达方法:回溯建立的逻辑链。 2,思考能力:具体问题具体分析;抽象思维,理解概念。 3,学习能力:提高信息的转化效率;学会对庞杂的知识建立体系。 图1:逻辑思维能力的构成

所谓的在日常生活中运用这些能力,对于一个大学生来说,就是降低运用直觉思维的能力的次数。直觉思维是逻辑思维的对立面,批判性思维的对立面就是定势思维,信息转化的高效率的对立面就是在阅读时的“逻辑无意识”,“逻辑无意识”是因为人在生理上和精神上都不擅长于逻辑思考,试图运用直觉经验来解释需要进行逻辑思考的问题。 下面对我大学两年来的一些涉及到需要进行表达,思考和学习的活动中,我的不突出的表现来分析我的现实的逻辑思维能力存在的不足:大学生能力的培养主要基于专业学习和实践活动,专业学习包含了如何去运用思考和学习能力,而实践活动需要良好的表达能力,当然大学生最重要的是培养综合素质,但如前文所说,我认为的逻辑思维是一种技术,一种建构和解构的技术,建立和分解逻辑结构都是更好地去运用它,逻辑思维在大学生素质构成中应该占有较大的比重。(2)我的表现 1,专业学习: a.专业课程:人力资源专业课程的学习包括六大模块,其中目前已学习了薪酬管理、人员招聘与配置以及人力资源培训与开发。在三门课程中,目前已知薪酬管理的学习收获和效果。在这门课程中,我认为需要掌握的是一门技术类知识的内在构成逻辑,即如何去使用这门课程中操作性的方法,而我的缺陷则是将这些结构化的知识分散记忆,没能学会对这些知识建立起属于自身的体系并运用。

二叉树习题及答案

1.设一棵完全二叉树共有699个结点,则在该二叉树中的叶子结点数? 1根据“二叉树的第i层至多有2^(i ? 1)个结点;深度为k的二叉树至多有2^k ? 1个结点(根结点的深度为1)”这个性质: 因为2^9-1 < 699 < 2^10-1 ,所以这个完全二叉树的深度就是10,前9层就是一个满二叉树, 这样的话,前九层的结点就有2^9-1=511个;而第九层的结点数就是2^(9-1)=256 所以第十层的叶子结点数就是699-511=188个; 现在来算第九层的叶子结点个数。 由于第十层的叶子结点就是从第九层延伸的,所以应该去掉第九层中还有子树的结点。因为第十层有188个,所以应该去掉第九层中的188/2=94个; 所以,第九层的叶子结点个数就是256-94=162,加上第十层有188个,最后结果就是350个 2完全二叉树:若二叉树中最多只有最下面两层的结点的度可以小于2,并且最下面一层的结点(叶结点)都依次排列在该层最左边的位置上,这样的二叉树为完全二叉树。 比如图: 完全二叉树除叶结点层外的所有结点数(叶结点层以上所有结点数)为奇数,此题中,699就是奇数,叶结点层以上的所有结点数为保证就是奇数,则叶结点数必就是偶数,这样我们可以立即选出答案为B! 如果完全二叉树的叶结点都排满了,则就是满二叉树,易得满二叉树的叶结点数就是其以上所有层结点数+1比如图: 此题的其实就是一棵满二叉树,我们根据以上性质,699+1=700,700/2=350,即叶结点数为350,叶结点层以上所有结点数为350-1=349。 3完全二叉树中,只存在度为2的结点与度为0的结点,而二叉树的性质中有一条就是:n0=n2+1;n0指度为0的结点,即叶子结点,n2指度为2的结点,所以2n2+1=699 n2=349;n0=350 2.在一棵二叉树上第5层的结点数最多就是多少 一棵二叉树,如果每个结点都就是就是满的,那么会满足2^(k-1)1。 所以第5层至多有2^(5-1)=16个结点! 3、在深度为5的满二叉树中,叶子结点的个数为 答案就是16 ~ 叶子结点就就是没有后件的结点~ 说白了~ 就就是二叉树的最后一层~ 深度为K的二叉树~ 最多有2^k-1个结点~ 最多有2^(k-1)个结点~ 所以此题~ 最多有2^5-1=31个结点~ 最多有2^(5-1)=16个叶子结点~ 4、某二叉树中度为2的结点有18个,则该二叉树中有几个叶子结点? 结点的度就是指树中每个结点具有的子树个数或者说就是后继结点数。 题中的度为2就是说具有的2个子树的结点; 二叉树有个性质:二叉树上叶子结点数等于度为2的结点数加1。 5、在深度为7的满二叉树中,度为2的结点个数为多少, 就就是第一层只有一个节点,她有两个子节点,第二层有两个节点,她们也都有两个子节点以此类推,所以到第6层,就有2的5次方个节点,她们都有两个子节点 最后第7层都没有子节点了。因为就是深度为7的。 所以就就是1+2+4+8+16+32了

二叉树习题及答案(考试学习)

1.设一棵完全二叉树共有699个结点,则在该二叉树中的叶子结点数? 1根据“二叉树的第i层至多有2^(i ? 1)个结点;深度为k的二叉树至多有2^k ? 1个结点(根结点的深度为1)”这个性质: 因为2^9-1 < 699 < 2^10-1 ,所以这个完全二叉树的深度是10,前9层是一个满二叉树, 这样的话,前九层的结点就有2^9-1=511个;而第九层的结点数是2^(9-1)=256 所以第十层的叶子结点数是699-511=188个; 现在来算第九层的叶子结点个数。 由于第十层的叶子结点是从第九层延伸的,所以应该去掉第九层中还有子树的结点。因为第十层有188个,所以应该去掉第九层中的188/2=94个; 所以,第九层的叶子结点个数是256-94=162,加上第十层有188个,最后结果是350个 2完全二叉树:若二叉树中最多只有最下面两层的结点的度可以小于2,并且最下面一层的结点(叶结点)都依次排列在该层最左边的位置上,这样的二叉树为完全二叉树。 比如图: 完全二叉树除叶结点层外的所有结点数(叶结点层以上所有结点数)为奇数,此题中,699是奇数,叶结点层以上的所有结点数为保证是奇数,则叶结点数必是偶数,这样我们可以立即选出答案为B! 如果完全二叉树的叶结点都排满了,则是满二叉树,易得满二叉树的叶结点数是其以上所有层结点数+1比如图: 此题的其实是一棵满二叉树,我们根据以上性质,699+1=700,700/2=350,即叶结点数为350,叶结点层以上所有结点数为350-1=349。 3完全二叉树中,只存在度为2的结点和度为0的结点,而二叉树的性质中有一条是:n0=n2+1;n0指度为0的结点,即叶子结点,n2指度为2的结点,所以2n2+1=699 n2=349;n0=350 2.在一棵二叉树上第5层的结点数最多是多少 一棵二叉树,如果每个结点都是是满的,那么会满足2^(k-1)1。 所以第5层至多有2^(5-1)=16个结点! 3.在深度为5的满二叉树中,叶子结点的个数为 答案是16 ~ 叶子结点就是没有后件的结点~ 说白了~ 就是二叉树的最后一层~ 深度为K的二叉树~ 最多有2^k-1个结点~ 最多有2^(k-1)个结点~ 所以此题~ 最多有2^5-1=31个结点~ 最多有2^(5-1)=16个叶子结点~ 4.某二叉树中度为2的结点有18个,则该二叉树中有几个叶子结点? 结点的度是指树中每个结点具有的子树个数或者说是后继结点数。 题中的度为2是说具有的2个子树的结点; 二叉树有个性质:二叉树上叶子结点数等于度为2的结点数加1。 5.在深度为7的满二叉树中,度为2的结点个数为多少, 就是第一层只有一个节点,他有两个子节点,第二层有两个节点,他们也都有两个子节点以此类推,所以到第6层,就有2的5次方个节点,他们都有两个子节点

常用的逻辑思维方法有哪些

常用的逻辑思维方法有哪些 假设法 假设法就是对于给定的问题,先做一个或多个假设,然后根据已知条件来分析,如果与题目所给的条件矛盾,就说明假设错误,然 后再用其它的假设。 排除法 排除法:已知在有限个答案中,只有一个是正确的,对于一个答案,不知道它是否正确,但是知道这个答案之外的其它答案都是错 误的,所以推断这个答案是正确的。 著名侦探福尔摩斯说过:“当排除了所有其它的可能性,还剩一个时,不管有多么的不可能,那都是真相。” 反证法 反证法是“间接证明法”一类,是从反面的角度的证明方法,即:肯定题设而否定结论,从而得出矛盾。具体地讲,反证法就是从反 论题入手,把命题结论的否定当作条件,使之得到与条件相矛盾, 肯定了命题的结论,从而使命题获得了证明。 常见步骤: 第一步:假设命题结论不成立,即假设结论的反面成立。 第二步:从这个命题出发,经过推理证明得出矛盾。 第三步:由矛盾判断假设不成立,从而肯定命题的结论正确。 等级和阶段 等级:事物的发展过程分为多个等级,具备一定的条件,才能进入相应的等级。

阶段:事物的发展过程分为多个阶段,具备一定的条件,才能进入相应的阶段。 等级和阶段的作用: (1)区分作用。一些事物可以按照所处的等级或阶段来进行区分。 (2)描述事物变化、发展的过程。例如:我们常说一个事物发展 到什么阶段了,或者一个事物发展到什么等级了。 筛选思维 筛选:通过淘汰的方式对事物进行的挑选。 对于多层筛选,需要为每层都设置通过的条件,符合条件的事物可以通过,不符合条件的事物被淘汰掉,那些符合条件的事物再进 入到下一级别筛选,从而实现一层一层的筛选。 限定思维 限定是为了缩小范围。语言中的定语就是为了限定主语和宾语,从而缩小主语和宾语的范围。 (1)用形容词限定主语: 例如:“猫”→“黑色的猫”。“黑色的”这样的限定,就缩小了指定的猫的范围。 (2)用名词所有格限定主语: 例如:“猫”→“小明的猫”。“小明的”这样的限定,就缩小了指定的猫的范围。 (3)用数词限定主语: 例如:“两只猫。”“两只”是数量上的限定。 计算法 (1)计算法解决关于定性的问题: 例如:计算分数判断考试及格还是不及格。

二叉树练习题答案

一、下面是有关二叉树的叙述,请判断正误 (∨)1. 若二叉树用二叉链表作存贮结构,则在n个结点的二叉树链表中 只有n—1个非空指针域。 ( X )2.二叉树中每个结点的两棵子树的高度差等于1。 (∨)3.二叉树中每个结点的两棵子树是有序的。 ( X )4.二叉树中每个结点有两棵非空子树或有两棵空子树。 ( X)5.二叉树中所有结点个数是2k-1-1,其中k是树的深度。 ( X )6.二叉树中所有结点,如果不存在非空左子树,则不存在非空右子树。 ( X )7.对于一棵非空二叉树,它的根结点作为第一层,则它的第i层上最 多能有2i-1个结点。 (∨)8.用二叉链表法存储包含n个结点的二叉树,结点的2n个指针区域中有n+1个为空指针。 ( X )9. 具有12个结点的完全二叉树有5个度为2的结点。 二、填空 1.由3个结点所构成的二叉树有 5 种形态。 2. 一棵深度为6的满二叉树有 26-33 个分支结点和 32 个叶子。 3.一棵具有257个结点的完全二叉树,它的深度为 9 。

4. 设一棵完全二叉树具有1000个结点,则此完全二叉树有500 个叶子结点,有 499 个度为2的结点,有 1 个结点只有非空左子树,有 0 个结点只有非空右子树。(分析:完全二叉树中三种节点个数n0,n1,n2,其中 n1为0或1;n0=n2+1;总节点个数N=n0+n1+n2=n0+n1+n0-1=2*n0-1+n1.由此推 出当完全二叉树中节点个数为偶数时n1为1,否则为0,则可计算本题) 5. 二叉树的基本组成部分是:根(N)、左子树(L)和右子树(R)。因而二叉 树的遍历次序有六种。最常用的是三种:前序法(即按N L R次序),后序法(即 按 LRN 次序)和中序法(也称对称序法,即按L N R次序)。这三种方法 相互之间有关联。若已知一棵二叉树的前序序列是BEFCGDH,中序序列是FEBGCHD,则它的后序序列必是 FEGHDCB 。 6. 用5个权值{3, 2, 4, 5, 1}构造的哈夫曼(Huffman)树的带权路径长度 是 (1+2)*3+3*2+(4+5)*2= 33 。 7.一个深度为h的二叉树最多有 2h-1 结点,最少有 h 结点。

定量分析的一般步骤知识分享

第二章分析试样的采取和预处理方法 教学要求:1、掌握定量分析的一般步骤;2、掌握试样采取得一般原则、无机和有机试样的分解方法;3、了解试样的制备和保存方法 试样的分析过程,一般包括下列步骤:试样的采取和制备、试样的预处理、干扰组分的掩蔽和分离、定量测定、分析结果的计算和评价。 §2-1 分析试样的采取和制备 试样采取得重要性和意义: 分析试样的采集: 指从大批物料中采取少量样本作为原始试样,所采试样应具有高度的代表性,采取的试样的组成能代表全部物料的平均组成,否则分析结果再准确也是毫无意义的。 一、采取试样的一般原则 1、现场勘察并收集资料; 2、代表性; 3、采用量符合要求; 4、合理保存 二、固体试样的采取 (一)矿石试样 1、采样点的布设(汽车、火车、轮船、矿堆、传送带等) 2、湿存水的去除(100-105o C烘干) 3、制备 制备试样分为破碎,过筛,混匀和缩分四个步骤。 粗碎(过4-6号筛)、缩分、中碎(过20号筛)、缩分、碾磨、缩分、分析试样。 大块矿样先用压碎机破碎成小的颗粒,再进行缩分。常用的缩分方法为“四分法”,将试样粉碎之后混合均匀,堆成锥形,然后略为压平,通过中心分为四等分把任何相对的两份弃去,其余相对的两份收集在一起混匀,这样试样便缩减了一半,称为缩分一次。每次缩分后的最低重量也应符合采样公式的要求。如果缩分后试样的重量大于按计算公式算得的重量较多,则可连续进行缩分直至所剩试样稍大于或等于最低重量为止。然后再进行粉碎、缩分,最后制成100-300克左右的分析试样,装入瓶中,贴上标签供分析之用。

通常试样的取样量可按下面的经验公式(亦称采样公式)计算: m = Kd a 式中:m为采取拭样的最低重量(公斤);d为试样中最大颗粒的直径(毫米); K和a为经验常数,可由实验求得,通常K值在0.02-1之间,a值在1.8—2.5 之间。地质部门规定a值为2,则上式为:m=Kd2 筛号(网目) 20 40 60 80 100 120 200 筛孔大小/mm 0.83 0.42 0.25 0.177 0.149 0.125 0.074 一般要求通过100-200目筛。 (二)土壤试样(了解) 1、采样点的布设:根据地形地貌和地块大小决定。有梅花形、棋盘式、蛇 形布点法等。 2、采样时间:根据分析内容确定 3、采样深度:根据作物的根系确定:农作物0-20,水果0-60 4、采用量:每点采1-2kg,经压碎、风干、粉碎、过筛、缩分等步骤,取 粒径小于0.5 mm的样品作分析试样。 5、保存:根据测定项目,对挥发酚、氨氮、氰化物等不稳定组分需要新鲜 土壤,其他多数需要风干土壤。风干时注意阳光直射和尘土落入,保存在玻璃 或聚四氟乙烯塑料容器中。 (三)金属或金属制品(了解) 由于金属经过高温熔炼,组成比较均匀,因此,对于片状或丝状试样,剪 取一部分即可进行分析。但对于钢锭和铸铁,由于表面和内部的凝固时间不同, 铁和杂质的凝固温度也不一样,因此,表面和内部的组成是不很均匀的。取样 时应先将表面清理,然后用钢钻在不同部位、不同深度钻取碎屑混合均匀,作 为分析试样。对于那些极硬的样品如白口铁、硅钢等,无法钻取,可用铜锤砸 碎之,再放入钢钵内捣碎,然后再取其一部分作为分析试样。 (四) 粉状或松散物料试样(了解)

相关文档
最新文档