D*******a
发帖数: 3688 | 1 【 以下文字转载自 Programming 讨论区 】
发信人: DrumMania (嫑跑，兲), 信区: Programming
标 题: 运行时间比较
发信站: BBS 未名空间站 (Sat Apr 26 14:06:50 2008), 转信
有两台机器：
A: P4-2.6GHT, 512M memory, Fedora 8, GCC 4.1.2
B: P4-2.8GHT, 1.5G memory, Mac OS X 10.5.2, GCC 4.0.1 (Xcode)
用-O3编译同一个程序（min cut graph partitioning using simulated annealing)
运行时间差很远
A只用了2:21，但是B却用了5:51
按道理B应该运算能力更强，可是实际上却差那么远，怎么解释呢？ |
|
F********E
发帖数: 1025 | 2 有N(N>>1)个球，在作simulated annealing。这N个球最初为完整一组，在simulating
过程中，可能会分成不同组。希望尽量避免过于散乱的分组(ie, (N/2)+(N/2)可以接受
，但(N-1)+1就不行)。想找一个归一化函数(f(N)=0, f(1,1,...1)=1)，不知道大家有
什么建议没有。
本人学化学的，如果有复杂数学用语的话，希望能给各reference，多谢！ |
|
j*******e
发帖数: 586 | 3 我有个optimization问题，对于objective function,我现在知道的，一般情况下：
（1）函数值可以算
（2）函数的gradient也可以算出来（quasi-analytical: expression里含有error
function)
(3)一般有两个local maxima或者更多（我不清楚），其中一个最大的是我要的
请问我该怎么选择算法呢？simulated annealing可以吗？非常感谢！ |
|
w****o
发帖数: 79 | 4 can you forward me a copy if you got the paper?
Thanks,
w****[email protected] |
|
h***e
发帖数: 232 | 5 dl one for you. email? BaoZi PLS :) |
|
|
f*********h
发帖数: 392 | 7 我已经查到文件的链接，无奈
我公司没有数据库的access， 只能看第一页，真气人
肯请好心人帮忙查到，发到我个人邮箱t******[email protected]不胜感激
http://www.scientific.net/0-87849-963-6/681/
11. An in-situ post growth annealing process for the improvement of 4H-SiC/
SiO2 MOS interface prepared by CVD using TEOS, and its characteristic study
Ramanujam K, Furuichi H, Taguchi K, Yukumoto S, Nishino S : SILICON
CARBIDE AND RELATED MATERIALS 2004 Book Series: MATERIALS SCIENCE FORUM
Volume: 483 Pages: 681-684 Published: 2005 |
|
|
f*****k
发帖数: 195 | 9 Sorry one more damn question.
should I anneal it before etching or after etching?
it |
|
f*****k
发帖数: 195 | 10 annealing does work! Thanks!! |
|
a*******i
发帖数: 11664 | 11 ft, cigs的生长温度和annealling温度是几百度，然后cool down到room temperature,
主要的mismatch是在这里吧。
thermal
either |
|
s******p
发帖数: 4962 | 12 用什么样的thermal anneal process？ |
|
w********o
发帖数: 10088 | 13 要做Hall, 用一个很简单的shadow mask（不锈钢片上面几个眼），放在GaAs HEMT样品
上（有n-type cap layer的），扔到evaporator里面deposit金属contact，拿出来
anneal以后，发现sheet resistance降了10倍。charge变大，mobility变小
表面很干净，只有四个角上有金属。你们说这是为什么，神马反应能让sheet
resistance降10倍？难道以前有10倍的dopant都没有激活？不可能啊 |
|
w********o
发帖数: 10088 | 14 今天重新试了一下，deposit完金属，sheet resistance 就从85降到30了，anneal 1.5
分钟后，降到7。 见鬼了 |
|
g*z
发帖数: 124 | 15 tried adaptive simulated annealing, looks promising |
|
g*z
发帖数: 124 | 16 tried adaptive simulated annealing, looks promising |
|
k*******n
发帖数: 17 | 17 最近投了好多家都没有反应，之前说要给phone interview的几家莫名其妙就不联系我
了，都搞不懂怎么回事，现在没法了，只好过来求内推了，求rp了
semiconductor process engineer or product engineer or materials engineer
master阶段主要做的quantum dots合成和optical device，不知道是不是这个原因闹得
有点四不像？
厚着脸皮来求内推。。下面是我的简历摘要，我邮箱y********[email protected]，qq：
339037332，谢谢各位大哥大姐啦。。
SUMMARY
• Hands on electrical engineer with five years research experiences on
semiconductor and materials process, including both quantum dot synthesis
and organic/inorganic semiconductor materials, as well as l... 阅读全帖 |
|
发帖数: 1 | 18 WinS (Shenzhen Institute of Wind bandgap Semiconductors深圳第三代半导体研究
院) is seeking the following job candidates. The R&D team consisting of
senior VPs /scientist/engineers/ with 10~20 years experiences in industry
from U.S. Candidates with semiconductor materials/device/process background
and FAB experience are highly recommended. Package is negotiable, benefits
comparable to SUSTech University（南方科技大学）
Candidates with Ph.D from top 150 universities are qualified to get “孔雀计
划A/B/C” with 3.0 to 1.6... 阅读全帖 |
|
发帖数: 1 | 19 WinS (Shenzhen Institute of Wind bandgap Semiconductors深圳第三代半导体研究
院) is seeking the following job candidates. The R&D team consisting of
senior VPs /scientist/engineers/ with 10~20 years experiences in industry
from U.S. Candidates with semiconductor materials/device/process background
and FAB experience are highly recommended. Package is negotiable, benefits
comparable to SUSTech University（南方科技大学）
Candidates with Ph.D from top 150 universities are qualified to get “孔雀计
划A/B/C” with 3.0 to 1.6... 阅读全帖 |
|
|
|
r****r
发帖数: 14 | 22 In most of time, the end group should be excluded from the crystalline region
and accumulated on the crystal surface because it will affect the stability of
crystals. But in some extreme case like low temp, which leads to fast
kinetics, the end group or some other defects can be trapped into the
crystalline phase. In that case, kinetics plays more important role than
thermodynamics. Those trapped defects may still have chance to be excluded
through annealing at higher temperature.
There are bunc |
|
z**h
发帖数: 224 | 23 if you are running neutron or x-ray reflectivity, your sample has three
different layers at first. after you annealed it, these layers mergered
together. will the critical angle change or not? Thanks. |
|
z**h
发帖数: 224 | 24 the critical angle is determined by the average refractive index of the whole
sample. Does the average index change after annealing? I think only the
distribution changed. |
|
c*******n
发帖数: 1648 | 25 since I stopped surfing for 2 days, so many holes have appeared just like
damned roads outside:)
1st question: Do you study plastics? or sth else?
If it's for plastics, then warping usually is a common problem for various
molding processes. Molecules don't like the shape of molds, they want to relax
after mold release. Then macroscopically you see warping. Maybe you know why
it's alway difficult to open the drawers of an old furnituer, actually the
same
thing, warping of celluose!
Annealing may |
|
c*****e
发帖数: 238 | 26 The length scale of phase separation is mostly governed by the annealing time
and preparing condition. So in principle there is no upper bound as to the
size of the phases.
It is definitely possible that the two materials are not mixed good enough,
but if the resulting morphology shows some sort of order, that is very
interesting.
Finally for thin films, interface interaction is very important and will
affect the density distribution in the longitudinal direction, i.e.,
perpendicular to the surf |
|
c*****e
发帖数: 238 | 27 Still some questions remain.
First, can you tell the composition within the phase boundaries?
Second, does the pattern persist if you increase/decrease the annealing time?
Third, what kind of change is induced by changing the volume fractions?
Fourth, what kind of monomer interactions (\chi) can be assumed for A and B?
是
的
同 |
|
s****e
发帖数: 2934 | 28 Your question is about curing or annealing? |
|
x****a
发帖数: 1041 | 29 AFM想看相图吧。可以试试用硅烷处理下硅片。spin coat 以后也可以试试anneal .
Dip coat 也可以试试 |
|
m*******l
发帖数: 220 | 30 是这样吗:先随便算几个点来estimate，然后stochastic approximation？
这和simulated annealing有什么不同呢?可以找到现成的软件吗？ |
|
b****t
发帖数: 114 | 31
if your problem does not have a closed-form objective
function, simulation-based methods may be the good start.
e.g. random search (genetic, annealing, stoch ruler...).
these can gaurantee the global convergence, but quite slow.
Using sample-path, you may consider using derterministic
optimization tools to solve a seqence of approximating
problems to get better solutions. But it is hard to get global
solutions for general problems using nonlinear programming techn...
beet |
|
F********E
发帖数: 1025 | 32 有N(N>>1)个球，在作simulated annealing。这N个球最初为完整一组，在simulating
过程中，可能会分成不同组。希望尽量避免过于散乱的分组(ie, (N/2)+(N/2)可以接受
，但(N-1)+1就不行)。想找一个归一化函数(f(N)=0, f(1,1,...1)=1)，不知道大家有
什么建议没有。
本人学化学的，如果有复杂数学用语的话，希望能给各reference，多谢！ |
|
S*****n
发帖数: 6055 | 33 比如说我们的CNT是catalytic CVD长出来的,那么可以认为一端是金属粒子,那么另一端
呢?是open end还是closed with half fullerene cap?
如果把金属粒子用酸洗掉，那么这一端又是不是open的呢？有文献说这一端full of
defects，defects造成这一端实际上是close的，只有高温annealing才能打开，真的么？
一直没想明白ing。。。
哪位大虾解释一下？多谢！ |
|
d***p
发帖数: 306 | 34 这个anneal总是要有个过程吧，不可能cdse一下从核里蹦出来，中间停止反应就行了。
如果是一下就彻底分
开那才是奇怪呢。
CdSe |
|
l******u
发帖数: 2314 | 35 试试这个：把CNT悬浮到某溶剂里(表面活性剂/水，有机溶剂都应该可以），把悬浮液
滴到你的substrate上，用氮气顺着槽的方向吹干。如果用了表面活性剂，要anneal掉
吧。
记得这是数年之前Nano Lett上看到的一个作alignment的方法。至于管子能不能进槽，
俺就不知道了。赫赫。 |
|
l*******s
发帖数: 65 | 36 I suppose you are talking about growing TiO2 nanotube on ITO substrate. C.
Grimes at PSU did some work on this.The advantage is incident light can
directly reach the active layer in a front-side illumination mode. However,
it is hard to deposit thick Ti on substrate with high quality, which is very
important for the formation of TiO2 NT; also, after anodization and
annealing, the contact between TiO2 NT and substrate might be a problem.
It is also possible to use hydrothermal approach to grow T |
|
x********8
发帖数: 167 | 37 1，硝酸回流处理后碳管损失量有多大
2，表面氧化以后，碳管似乎可以叫做氧化物了，synthon说氧含量有10%。
3，什么条件下可以把硝酸处理后的表面基团全部annealing掉，气氛选择，温度选择
上次称了0.2克单壁管煮了一天，一超声分散发现溶液太透明了。 |
|
n*****7
发帖数: 19 | 38 请问你得copolymer是自己合成的还是买的？
如果是合成，用的什么方法？谢谢 |
|
n**********s
发帖数: 145 | 39 水热合成的氧化物（价态最高，不能再被氧化;微米／纳米材料）能被有机氨溶解，但
是若合成后在空气或者氩气中在４００C anneal一个小时后，发现在相同的条件下基本
不溶解。大家给点线索吧？谢谢！祝周末愉快！ |
|
g******e
发帖数: 760 | 40 google Q.-Y. Tong and U. Gösele. |
|
p*****0
发帖数: 1100 | 41 我的第一篇文章投出去快四个月，现在审稿意见回来，第一个审稿人问了很多问题，第
二个审稿人问题很少，基本是修改格式。编辑的意见是拒稿，但是建议修改后重新投，
他愿意重新处理。
现在头疼的是第一个审稿人列出了八九篇文献要求阅读，但是其中三篇不能直接下载。
可怜我们实验室很穷，没有下文章的免费账号。自己掏钱，又不知道花100多刀下那三
篇文献值不值，有可能最后还是被拒了。请教各位，如果有那三篇文献的，能不能发到
我邮箱？鄙人将感激不尽！！
1. Title: Zinc oxide, a multifunctional material: from material to device
applications. Source: APPLIED PHYSICS A-MATERIALS SCIENCE & PROCESSING
Volume: 96 Issue: 1 Special Issue: Sp. Iss. SI Pages: 197-205
2. Title: Nonvolatile resistive switching in single crystalline ... 阅读全帖 |
|
|
m**********e
发帖数: 12525 | 43 可以，理论上monte carlo是基于各态历经性，用时间平均来代替系宗平均。
所以只要外界能量输入速度没快到破坏各态历经性，monte carlo方法都是有效
的。
具体技巧可以参考下monte carlo方法模拟simulated annealing和quench
部能 |
|
m**********e
发帖数: 12525 | 44 你找本simulated annealing,或者化学它们做的分子模拟方面的书,
都有详细解答,一般是先把dynamic equ转为hamilton,然后就可以用
monte carlo了.这方面有大量软件,甚至自己都不用写程序
+ |
|
Ac
发帖数: 50 | 45 我说的是这种
Enhancement of ferromagnetism upon thermal annealing in pure ZnO
Appl. Phys. Lett. 91, 182501 (2007); DOI:10.1063/1.2804081
Published 29 October 2007 |
|
a********r
发帖数: 177 | 46 That's true. However, can MCD dectect a enhanced signal in FM sample ( e.g.
annealed one) compared to the NFM in materials like in ZnO? The normal
Zeeman spliting is much smaller than the sp-d splitting in my inpression.
Do you happen to know any paper apply MCD to d0 materials?
Thanks!
发信人: Ac (Ac), 信区: Physics
标 题: Re: 有人做磁性半导体，自旋器件的么？
发信站: BBS 未名空间站 (Thu Mar 27 16:33:30 2008), 转信
mcd signal reflect the split of energy band
In DMS, the sp-d interaction just enhances the split.
In pure ZnO, t |
|
d*******n
发帖数: 62 | 47 Lots of people grow things on top of ultrathin Al2O3 films prepared by
annealing NiAl single crystals in oxygen. You can still do STM on top of
these films. Check related literature regarding thickness and resistivity,
and you may get some idea. |
|
t***r
发帖数: 119 | 48 按照下文说法，招过来直接自己manage their own book? 这个leverage group是个猎
头？
http://jobview.monster.com/Quantitative-Portfolio-Manager-Job-N
Company
The Leverage Group
Location
New York City, NY
About the Job
Quantitative Portfolio Manager (with Optimization Focus)
Unique opportunity to develop systematic trading strategies in a fast-paced,
results-oriented environment: each Portfolio Manager (PM) is responsible
for managing their own book and for the continuous evolution of their
strategies.
Candidates must pos... 阅读全帖 |
|
C******n
发帖数: 9204 | 49 最惨不是坑的问题，是大家都在做悲催的simulated annealing |
|
s***e
发帖数: 911 | 50
这个是自旋玻璃里面的概念. 比如系统Hamiltonian是:
H(X;P)
其中X是自由度矢量,P是系统参数矢量(比如相互作用强度)并且被假设为具有某种
随机分布. quenched disorder是指参数P虽然是随机量, 但是在实验相关的时间尺度
内, P可以看作是不变的. 固定P计算系统配方函数,最后再对P按P的概率计算
平均就是quenched average了.
P是很快因为热运动进入了热力学平衡态的分布,这时候计算系统配分函数必须同时
计如对P的计算. 也就是说(X,P)必须合起来看成系统的自由度, trace是对X,P同时
作的,这就是annealed average了. |
|
