1 / 25

South China University of Technology

South China University of Technology. Stochastic Process. Xiaobao Yang Department of Physics. www.compphys.cn. Generation of random number. circumference ratio. Generation Non-uniform random number. The possibilities of events A and B is 2:1.

Télécharger la présentation

South China University of Technology

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. South China University of Technology Stochastic Process Xiaobao Yang Department of Physics www.compphys.cn

  2. Generation of random number

  3. circumference ratio

  4. Generation Non-uniform random number The possibilities of events A and B is 2:1. The possibilities of events A1, A2 … An is R1, R2 … Rn . The possibilities of events A1, A2 … An is f(1), f(2) … f(n) .

  5. Weibull distribution

  6. Non-uniform cases uniform distribution x=rand(1000,1); y=-log(x); hist(y)

  7. Non-uniform cases (0, Py(max)) If Py(y1) < P test, we remove y1 from the sequence; otherwise it is kept. for ii=1:length(y) if exp(-y(ii))>rand x=[x y(ii)]; end end y=rand(1000,1)*6; x=[];

  8. Spatial correlation

  9. 帕斯卡 惠更斯

  10. Simulation of gambling The big bet induces chaos.

  11. SixGL.m Simulation of HK Six Marks There are Events 1~12. You win 10 when you hit it. for ii=2:N banker=floor(rand*12)+1; tmp=floor(rand(10,1)*12)+1; buyers(:,ii)=buyers(:,ii-1)-1+10*(tmp==banker); end

  12. Gambling.m Events of Big/Small/Tie. banker=floor(rand(1,3)*5)+1; tmp=floor(rand(10,1)*2)+2; if re==1 buyers(:,ii)=buyers(:,ii-1)-1; end if re==2 buyers(:,ii)=buyers(:,ii-1)+(tmp==2)*2-1; end if re==3 buyers(:,ii)=buyers(:,ii-1)+(tmp==3)*2-1; end if length(unique(banker))==1 re=1; end if length(unique(banker))~=1&sum(banker)<=10 re=2; end if length(unique(banker))~=1&sum(banker)>10 re=3; end

  13. Summary • Select a proper distribution • Construct the Stochastic Process • Store the data for histogram.

  14. The existence of god of gamblers? 1 1 1 0 0 1 0 0 1 0 …. When you bet N times, the possibility of all win is 2^(-N). Don't get crush on him and he is just a legend.

  15. Random walks

  16. Random walks • Concept of random walk. x0=0 • Informative quantity:

  17. 1-D random walks • clear • N=300; • M=100; • re=zeros(M,N); • for ii=1:N-1 • for jj=1:M • re(jj,ii+1)=re(jj,ii)+sign(rand-0.5); • end • end • plot(sum(re.^2,1)/M) rw1d.m

  18. Random walks • Questions: To make it more realistic: • What is the relation of <r2> vs. t for a random walk in 2-D, 3-D systems? • How about allow the steps to be a random length in [-1,1]?

  19. 2-D random walks • clear • N=300; • M=100; • re=zeros(M*2,N); • for ii=1:N-1 • for jj=1:M • if rand <0.5 • re(jj*2-1,ii+1)=re(jj*2-1,ii)+sign(rand-0.5); • re(jj*2,ii+1)=re(jj*2,ii); • else • re(jj*2,ii+1)=re(jj*2,ii)+sign(rand-0.5); • re(jj*2-1,ii+1)=re(jj*2-1,ii); • end • end • end • plot(sum(re.^2,1)/M) rw2d.m

  20. Random walk in hexagonal lattice O (0.00,0.00) A (0.00,1.00) B (-0.87 1.50) C ( 0.87 1.50) T=

  21. Self-Avoiding Walks A typical polymer in solution may be described by SAW. Keep all possible SAW configurations equivalent. To generate 4th steps if a new possible way is chosen when “self-avoiding” • There are 3 possible ways. (only has 2/3 the probability as in b) • There are only 2 possible ways . How to correct the problem? • Discard the whole RW when restriction condition occurred. CPU time ! • “enumeration” instead of “simulation”.

  22. saw.m jj=1:sx(1) newtmp=[x(jj,1:2:end)' x(jj,2:2:end)']; % find all the used sites in the same link! tmp=x(jj,end-1:end)+[0 1]; if length(intersect(tmp,newtmp,'rows'))==0; allx=[allx x(jj,:) tmp]; end

  23. Self-Avoiding Walks Log-log scale Average of 106 walks Average of 1000 walks ν=1/2: simple random walk ν=3/4: 2D SAW ν~3/5: 3D SAW

  24. Walker with memory 可以且仅可以记忆它上一步走过的方向并避免重复, 比如当从(0,0)走到(0,1)时,下一步只有3个可能的位移[(0,2), (1,1), (-1,1)], 避免了从(0,1)走回(0,0)位置。 while (re(jj*2-1,ii+1)==re(jj*2-1,ii-1)&re(jj*2,ii+1)==re(jj*2,ii-1)) if rand <0.5 re(jj*2-1,ii+1)=re(jj*2-1,ii)+sign(rand-0.5); re(jj*2,ii+1)=re(jj*2,ii); else re(jj*2,ii+1)=re(jj*2,ii)+sign(rand-0.5); re(jj*2-1,ii+1)=re(jj*2-1,ii); end walkwith.m

  25. Homework Random walk in triangular or hexagonal lattice. Sending to 17273799@qq.com when ready For lecture notes, refer to http://www.compphys.cn/~xbyang/

More Related