“We present a method for the recon­struc­tion of net­works, based on the order of nodes vis­ited by a sto­chas­tic branch­ing process. Our algo­rithm recon­structs a net­work of min­i­mal size that ensures con­sis­tency with the data. Cru­cially, we show that global con­sis­tency with the data can be achieved through purely local con­sid­er­a­tions, infer­ring the neigh­bour­hood of each node in turn. The opti­mi­sa­tion prob­lem solved for each indi­vid­ual node can be reduced to a Set Cov­er­ing Prob­lem, which is known to be NP-​​hard but can be approx­i­mated well in prac­tice. We then extend our approach to account for noisy data, based on the Min­i­mum Descrip­tion Length prin­ci­ple. We demon­strate our algo­rithms on syn­thetic data, gen­er­ated by an SIR-​​like epi­demi­o­log­i­cal model.”

“Even­tu­ally, a rever­sal towards smaller farms and less pro­duc­tion of corn, soy­beans and cot­ton may even result. Also, for a few years, U.S. input costs may esca­late while other regions (not yet see­ing super­weeds) con­tinue to use lower priced Roundup, mak­ing the cost of pro­duc­tion lower in South Amer­ica, for exam­ple. Con­se­quent ris­ing input costs here in the U.S. will include poten­tially higher new GMO seeds, chem­i­cals, labor, and fuel. The vol­ume of chem­i­cal use will poten­tially become much greater, more toxic, and more expensive.”

Ide­ally, investors should be watch­dogs of a sort as well, not slaves to board­speak. “BP may face crim­i­nal charges based on their own inter­nal doc­u­ments. This would amplify any BP civil liabilities.”

“What is the pri­mary dif­fer­ence? Ulti­mately it comes down to the ques­tion of “exit”. As a founder, I have no inter­est in exit or liq­uid­ity. I am in busi­ness to run a busi­ness, not to run away from it. Or as War­ren Buf­fet puts it: Our favorite hold­ing period is forever.”

“In our final video from the last Bay Area Quan­tifed Self Show&Tell meetup, here is Dan Brown explain­ing the micro­phone he embed­ded into an air mat­tress under his bed to track motion, snor­ing, and heartrate dur­ing sleep. He describes how he assem­bled his inven­tion, the tools he used for ana­lyz­ing his data, and why he had to stop the track­ing experiment.”

“Moti­vated by constraint-​​based CAD soft­ware, we develop the foun­da­tion for the rigid­ity the­ory of a very gen­eral model: the body-​​and-​​cad struc­ture, com­posed of rigid bod­ies in 3D con­strained by pair­wise coin­ci­dence, angu­lar and dis­tance con­straints. We iden­tify 21 rel­e­vant geo­met­ric con­straints and develop the cor­re­spond­ing infin­i­tes­i­mal rigid­ity the­ory for these struc­tures. The clas­si­cal body-​​and-​​bar rigid­ity model can be viewed as a body-​​and-​​cad struc­ture that uses only one con­straint from this new class. As a con­se­quence, we iden­tify a new, nec­es­sary, but not suf­fi­cient, count­ing con­di­tion for min­i­mal rigid­ity of body-​​and-​​cad struc­tures: nested spar­sity. This is a slight gen­er­al­iza­tion of the well-​​known spar­sity con­di­tion of Maxwell.”

“I see a future com­put­ing world which is net­worked and plat­form inde­pen­dent. And that means gate­keep­ers of band­width and con­tent will be the win­ners in that world. Over the short-​​term Microsoft and the tel­cos will play their part in pro­tect­ing their legacy fran­chises in these are­nas. But ulti­mately, peo­ple just want to get their con­tent when– and where– ever they can. And that means the orga­ni­za­tions which dom­i­nate the multi-​​device inter­faces of the future will take on a lead­ing role in tech­nol­ogy, per­haps the lead­ing role.…”

“This intro­duc­tion is based on my expe­ri­ences with using Pro­to­vis in my Visu­al­iza­tion and Visual Com­mu­ni­ca­tion class ear­lier this spring. While the con­cepts involved are really not that dif­fi­cult, they are rather for­eign to stu­dents who have not been exposed to func­tional pro­gram­ming. And since that is also the case for a lot of hob­by­ists and peo­ple want­ing to do visu­al­iza­tion who do not have a com­puter sci­ence back­ground, I imag­ine they run into the same problems.”

“But even if we sub­stan­tially improve edu­ca­tion, it won’t fully solve the prob­lem. There will still be a need for qual­ity jobs that are not all that depen­dent upon knowl­edge based skills. How­ever, it’s harder to imag­ine an emerg­ing set of indus­tries that will pro­vide the large num­ber of qual­ity jobs that we need to replace those lost from indus­tries in decline.”

“If coun­tries like Ger­many start to have issues sell­ing their trea­sury bonds, how long will it take before it impacts the global bond mar­ket? It shouldn’t take long. That’s why Europe can­not afford the same quan­ti­ta­tive eas­ing as the U.S. has done in the last year. Thus, the Greece Cri­sis is not well con­tained yet. Is the Great Depres­sion Avoidable?”

“We con­sider the prob­lem of devel­op­ing an effi­cient multi-​​threaded imple­men­ta­tion of the matrix-​​vector mul­ti­pli­ca­tion algo­rithm for sparse matri­ces with struc­tural sym­me­try. Matri­ces are stored using the com­pressed sparse row-​​column for­mat (CSRC), designed for prof­it­ing from the sym­met­ric non-​​zero pat­tern observed in global finite ele­ment matri­ces. Unlike clas­si­cal com­pressed stor­age for­mats, per­form­ing the sparse matrix-​​vector prod­uct using the CSRC requires thread-​​safe access to the des­ti­na­tion vec­tor. To avoid race con­di­tions, we have imple­mented two par­ti­tion­ing strate­gies. In the first one, each thread allo­cates an array for stor­ing its con­tri­bu­tions, which are later com­bined in an accu­mu­la­tion step. We ana­lyze how to per­form this accu­mu­la­tion in four dif­fer­ent ways.…”

“We present a method for esti­mat­ing the com­plex­ity of an image based on the con­cept of log­i­cal depth. Unlike the appli­ca­tion of the con­cept of algo­rith­mic com­plex­ity by itself, the addi­tion of the con­cept of log­i­cal depth results in a char­ac­ter­i­za­tion of objects by orga­ni­za­tional (phys­i­cal) com­plex­ity. We use this mea­sure to clas­sify images by their infor­ma­tion con­tent. The method pro­vides a means for eval­u­at­ing and clas­si­fy­ing objects by way of their visual representations.”

“We con­sider meth­ods for find­ing high-​​precision approx­i­ma­tions to sim­ple zeros of smooth func­tions. As an appli­ca­tion, we give fast meth­ods for eval­u­at­ing the ele­men­tary func­tions log(x), exp(x), sin(x) etc. to high pre­ci­sion. For exam­ple, if x is a pos­i­tive floating-​​point num­ber with an n-​​bit frac­tion, then (under rather weak assump­tions) an n-​​bit approx­i­ma­tion to log(x) or exp(x) may be com­puted in time asymp­tot­i­cally equal to 13M(n)lg(n), where M(n) is the time required to mul­ti­ply floating-​​point num­bers with n-​​bit frac­tions. Sim­i­lar results are given for the other ele­men­tary functions.…”

“In this short note we have pro­duced dif­fer­ent kind of magic squares using dig­i­tal let­ter hav­ing only the algo­risms: 0, 1, 2, 5, and 8. The inter­est­ing fact in con­sid­er­ing these five dig­its is that the day 8^th May 2010 also have these ones (08.05.2010). More­over, the magic squares pre­sented have some inter­est­ing prop­er­ties, such as: they remains the same if we rotate them by 180 degree, or see in the mir­ror, or see on the other side of the paper, etc. Two palin­drome semi-​​magic squares of order 3×3 are also given. Still, we have con­sid­ered other dates hav­ing four digits.”

“In this paper, the per­for­mances of the quasi-​​Newton BFGS algo­rithm, the NEWUOA deriv­a­tive free opti­mizer, the Covari­ance Matrix Adap­ta­tion Evo­lu­tion Strat­egy (CMA-​​ES), the Dif­fer­en­tial Evo­lu­tion (DE) algo­rithm and Par­ti­cle Swarm Opti­miz­ers (PSO) are com­pared exper­i­men­tally on bench­mark func­tions reflect­ing impor­tant chal­lenges encoun­tered in real-​​world opti­miza­tion prob­lems. Depen­dence of the per­for­mances in the con­di­tion­ing of the prob­lem and rota­tional invari­ance of the algo­rithms are in par­tic­u­lar investigated.”

‘While “self-​​styled” seems mostly a put-​​down, it is a notably weak one. The user of this phrase notes that some­one claims some­thing, but lacks an offi­cial cre­den­tial, or strong con­sen­sus, sup­port­ing this claim. But we the reader can also note that this speaker offers no stronger crit­i­cism, and is not will­ing to directly con­tra­dict the offend­ing claim. After all, instead of call­ing some­one a “self-​​styled vision­ary,” you might say “he calls him­self a vision­ary, but he’s not; he hasn’t has a vision in years.”’

“In this paper we present a novel genetic algo­rithm (GA) solu­tion to a sim­ple yet chal­leng­ing com­mer­cial puz­zle game known as the Zen Puz­zle Gar­den (ZPG). We describe the game in detail, before pre­sent­ing a suit­able encod­ing scheme and fit­ness func­tion for can­di­date solu­tions. We then com­pare the per­for­mance of the genetic algo­rithm with that of the A* algo­rithm. Our results show that the GA is com­pet­i­tive with informed search in terms of solu­tion qual­ity, and sig­nif­i­cantly out-​​performs it in terms of com­pu­ta­tional resource require­ments. We con­clude with a brief dis­cus­sion of the impli­ca­tions of our find­ings for game solv­ing and other “real world” problems.”

“Thanks to recent tech­no­log­i­cal advances, it is now pos­si­ble to track with an unprece­dented pre­ci­sion and for long peri­ods of time the move­ment pat­terns of many liv­ing organ­isms in their habi­tat. The increas­ing amount of data avail­able on sin­gle tra­jec­to­ries offers the pos­si­bil­ity of under­stand­ing how ani­mals move and of test­ing basic move­ment mod­els. Ran­dom walks have long rep­re­sented the main descrip­tion for micro-​​organisms and have also been use­ful to under­stand the for­ag­ing behav­iour of large ani­mals. Nev­er­the­less, most ver­te­brates, in par­tic­u­lar humans and other pri­mates, rely on sophis­ti­cated cog­ni­tive tools such as spa­tial maps, episodic mem­ory and travel cost dis­count­ing. These prop­er­ties call for other mod­el­ing approaches of mobil­ity pat­terns. We pro­pose a for­ag­ing frame­work where a learn­ing mobile agent uses a com­bi­na­tion of memory-​​based and ran­dom steps. We inves­ti­gate how advan­ta­geous it is to use mem­ory for exploit­ing resources in het­ero­ge­neous and chang­ing environments.…”

“Google and the USPTO have entered into an agree­ment to make the fol­low­ing USPTO prod­ucts avail­able to the pub­lic at no charge:

Patents (grants, appli­ca­tions, assign­ments, clas­si­fi­ca­tion infor­ma­tion, and main­te­nance fee events)
Trade­marks (grants, appli­ca­tions, assign­ments, and TTAB proceedings)

All data orig­i­nated from the USPTO. Google is host­ing this data unchanged, except for repack­ag­ing into zip files.”

“The goal of topol­ogy opti­miza­tion is to find opti­mal mate­r­ial dis­tri­b­u­tion in the given design domain sub­ject to some con­straints gov­erned by cer­tain phys­i­cal prop­er­ties and/​or some other prac­ti­cal con­straints dur­ing the design. In the past two decades, advances in the the­ory of homog­e­niza­tion, opti­miza­tion, numer­i­cal analy­sis as well as newly devel­oped engi­neer­ing approaches make topol­ogy opti– miza­tion tech­niques to become a stan­dard tool of engi­neer­ing design, in par­tic­u­lar in the field of struc­tural mechanics.”

“Shorter — there really is an urgent and per­ilous threat to Israel. It’s called “the Israeli government”.”

“Algo­rith­mic his­to­ri­og­ra­phy was pro­posed by Eugene Garfield in col­lab­o­ra­tion with Irv­ing Sher in the 1960s, but fur­ther devel­oped only recently into HistCite^{TM} with Alexan­der Pudovkin. As in his­tory writ­ing, HistCite^{TM} recon­structs by draw­ing intel­lec­tual lin­eages. In addi­tion to cited ref­er­ences, how­ever, doc­u­ments can be attrib­uted a mul­ti­tude of other vari­ables such as title words, key­words, jour­nal names, author names, and even full texts. New devel­op­ments in mul­ti­di­men­sional scal­ing (MDS) enable us not only to visu­al­ize these pat­terns at each moment of time, but also to ani­mate them over time. Using title words, co-​​authors, and jour­nal names in Garfield’s oeu­vre, the method is demon­strated and fur­ther devel­oped in this paper (and in the ani­ma­tion at this http URL). The vari­ety and sub­stan­tive con­tent of the ani­ma­tion enables us to write, visu­al­ize, and ani­mate the author’s intel­lec­tual history.”

“We for­mal­ize the pro­ce­dure of adap­ta­tion of a TSP neigh­bor­hood for GTSP and pro­pose effi­cient algo­rithms to explore the obtained neigh­bor­hoods. We also gen­er­al­ize all other exist­ing and some new GTSP neigh­bor­hoods. Apart from these the­o­ret­i­cal results, we also pro­vide the results of a thor­ough exper­i­men­tal analy­sis to com­pare the pro­posed algo­rithms imple­men­ta­tions and find out which neigh­bor­hoods are the most effi­cient in practice.”

“We use three-​​dimensional phase-​​field sim­u­la­tions to inves­ti­gate the dynam­ics of the two-​​phase com­pos­ite pat­terns formed upon dur­ing solid­i­fi­ca­tion of eutec­tic alloys. Besides the spa­tially peri­odic lamel­lar and rod pat­terns that have been widely stud­ied, we find that there is a large num­ber of addi­tional steady-​​state pat­terns which exhibit sta­ble defects. The defect den­sity can be so high that the pat­tern is com­pletely dis­or­dered, and that the dis­tinc­tion between lamel­lar and rod pat­terns is blurred. As a con­se­quence, the tran­si­tion from lamel­lae to rods is not sharp, but extends over a finite range of com­po­si­tions and exhibits strong hys­tere­sis. Our find­ings are in good agree­ment with experiments.”

“Since mul­ti­vari­ate Boltz­mann sam­plers can be obtained in any sit­u­a­tion where the dis­tri­b­u­tion is well-​​concentrated, one may envi­sion exten­sions to other classes, includ­ing con­strained trees, permu– tations with a fixed num­ber of cycles, func­tional graphs with a con­trolled num­ber of components.…”