Science topic

# Computer Science and Engineering - Science topic

Explore the latest questions and answers in Computer Science and Engineering, and find Computer Science and Engineering experts.
Questions related to Computer Science and Engineering
Question
20 answers
It is said that if you want to make high quality softwares than you must know data structures
Relevant answer
Answer
Data are bases for getting information, and information is basis for making decisions - it can be considered as an axiom in data analysis. Only an appropriate structure of data can enable us to get information hidden in the data. Especially, data structures are relevant in situation of longitudinally collected data, e.g. in health care, where one patient can visit his/her doctor ones, twices or many times (or never). In such cases you have at least two entities you should consider: patient and visit (one patient, more visits, 1:n relationship). The database is then the appropriate data structure. In case you have only one entity (e.g. status of somthing (or someone) in a moment) the appropriate data structure is just a file, or table (e.g. made by using Excel). Data structuring is the crucial question in probvlem solving, in data analysis, for results which could be get by using either statistics or artificial intelligence methods.
Question
1 answer
Let we change the default block size to 32 MB and replication factor to 1. Let Hadoop cluster consists of 4 DNs. Let input data size is 192 MB. Now I want to place data on DNs as following. DN1 and DN2 contain 2 blocks (32+32 = 64 MB) each and DN3 and DN4 contain 1 block (32 MB) each. Can it be possible? How to accomplish it?
Relevant answer
Answer
As I understand, Hadoop replicate data based on rack awareness.
you can read more about this here:
Question
1 answer
I implemented the FCM and KFCM in java for clustering of dataset. However I obtained same results for Obj Function:Code in KFCM:
object += 2*(Math.pow(u[i * pattern+j], m) * K(v1, v2, dimension) * K(v1, v2, dimension));
Distance code:
b= (v1[i] - v2[i]) * (v1[i] - v2[i]);
result+=(1-Math.exp(-b/4) )
Relevant answer
Answer
Dear Lokesh,
I suggset for you a links and attached files in topics.
-Comparison of K-Means and Fuzzy c-Means Algorithm Performance ...
https://www.ncbi.nlm.nih.gov › ... › PubMed Central (PMC)
- Comparative Study of Different Clustering Algorithms | Open Access ...
- ]A Comparison Study between Various Fuzzy Clustering Algorithms
Best regards
What are the processors use CORDIC as one of the functional unit?
Question
4 answers
CORDIC is used to find the special functions like COS, SIN, TAN, LOG, E^X, hyperbolic, and etc. What are the processors use CORDIC as one of the functional unit? If processors don't use CORDC, then how the above mentioned functions are computed by them?
Relevant answer
Answer
CORDIC is usually implemented in the systems where hardware multiplier is not available, e.g. FPGA. Otherwise approximation of target function with Taylor series is faster.
Question
3 answers
How does the SAX works, I have the code for the same in java but how to decide the sliding window size for a time series. My time series is not continuous and there is lot of noise in the data.
Relevant answer
Answer
There is a Java implementation of SAX in the SPMF open-source data mining library:
http://www.philippe-fournier-viger.com/spmf/ (I am the founder, by the way).
And I wrote a simple blog post about how it works with a simple example:
The code is open-source, so you could modify it to have adaptive time windows if you like. And if you do so, please share the code back ;-)
Question
12 answers
I am wondering if it is possible to change the wavelength of a photon. Can another photon intersect the path of the first and ultimately change its wavelength? Is it possible to have something which is attracted to certain photons?
Relevant answer
Answer
Two incoming photons scatter each other and in the final state there are two photons. This is not a tree level process in standard model because photons have no charge or mass, they cannot couple to each other. However, there are loop induced processes (one-loop level) where this process is possible. Experimental searches presently give an upper bound on the scattering cross section of light by light. (for example PVLAS experiment).
Question
5 answers
any researchers
Relevant answer
Or alternatively, you can also use RapidMiner. WEKA is good, but RapidMiner offers you better GUI.
Cheers,
Diky
Question
3 answers
Is there some heuristic or some rule of thumbs to decide the ideal number of map/reduce tasks for a job to be run of Hadoop cluster?
Relevant answer
Answer
It depends on your MapReduce job and on your cluster. There are some research works that aim to determine the optimal/best MapReduce parameters (number of Map/Reduce functions, Chunk size, ...) but they are generally designed for a specific application. For example, this paper deals with MapReduce-based pattern mining approaches:
Best,
Sabeur
Question
7 answers
Particularly, if yes then all major algorithms like Apriori, FP-Growth and Eclat are np-hard or only Apriori is np-hard?
Relevant answer
Answer
Yes. Usually, we say a np-hard problem instead of a np-hard algorithm.
Question
3 answers
We are trying to move our XRD evaluation package from year 2000 from an old computer to a new one, but so far we can't get windows7 to accept the dongle. Have anybody tried to do this and succeded, I will approciate to know how it was done BR Flemming
Relevant answer
Answer
We had this a while back, and I approached Bruker for an answer. I think the answer was that the old EVA was 32 bit and would not run on Windows 7 64 bit. We stayed with an old PC running Windows XP to run our copy of EVA v16. There is new version of EVA which is confusingly called EVA 3. I have seen it and hated it so I've stuck with EVA v16. If you want  Windows 7 then I think you are going to have to get the new Diffrac-suite.
Question
4 answers
Hi,
How i can differentiate if a system is a SoS or an ecosystem?
Thanks.
Lina
Relevant answer
Answer
Dear Lina,
System-of-systems
If we try to explain simply the notion of system-of-systems, it must start from the definition of a system. In summary a system is a set of elements that interact with each other according to defined rules. A system of systems therefore is a system where the elements and sub-assemblies are themselves systems. More complex way, a system of systems is a set of interconnected and coordinated autonomous systems to satisfy specific capacity and functions that the systems could not achieve independently.
A system of systems results from the collaborative operating systems components that can operate independently to fulfill their own operational mission.
Wanted by this collaboration the emergence of new behaviors used to improve the capabilities of each system component or provide new, while maintaining the operational independence and managerial constituent systems.
Ecosystem
Historically the ecosystem concept was born from the concept of biocenosis: people living together (populations are interrelated), then from the concept of ecological succession (populations are related to the environment). Ecos: means house.
An ecosystem is a dynamic set of living organisms (plants, animals and microorganisms) that interact with each other and with the environment (soil, climate, water, light) in which they live. Ecosystems dimensions can vary considerably; they can be very small, like a pond or a dead tree, or be gigantic, like Earth. An ecosystem can also be defined depending mainly on vegetation, animal species or relief, for example.
The major ecosystems are generally described as: aquatic ecosystems - saltwater or freshwater; terrestrial ecosystems - forests, grasslands, deserts, etc. Forest ecosystems are characterized by a predominance of trees, as well as the fauna, flora and ecological cycles (energy, water, carbon and nutrients) that are closely associated with them.
In ecology, an ecosystem is the set formed by an association or community of living beings (or biocenosis) and its biological environment, geological, edaphic, hydrological, climate, etc. (biotope). The components of an ecosystem develop a network for the exchange of energy and matter for the maintenance and development of life. The term was coined by Arthur Tansley in 1935 to describe the basic unit of nature.
With my best regards
Prof. Bachir ACHOUR
Question
4 answers
Is anyone aware of an overview database of ARM Cortex M processors for most licensed manufacturers, listing Flash, ROM, clock etc? I'm trying to figure out common specification classes.
Relevant answer
Answer
Not exactly what you are looking for, but seems to me that Tom's hardware have huge detailled lists of processors from long ago to the most current chips.
Question
8 answers
For example this digraph, start from s1, we can realize depth first traverse by different sequences such as { s1 t1 s2 t3 s3 t4 s1 t5 s3, s1 t2 s2} or {s1 t5 s3 t4 s1 t2 s2 t3 s3, s1 t1 s2}. I find that  in each group the start nodes is realated with the nodes whose outdegree is larger than its indegree, the end nodes in each group is realated with the nodes whose indegree is larger than its outdegree.
my question is:Given a digraph and start from the same node for depth first traverse,considering the solutions of each group, can we conclude the following properties ?
Property i: The same number of sequences in each group.
Property ii: The same kind of start nodes and the same number of each kind of start nodes for sequences in each group.
Property iii: The same kind of end nodes and the same number of each kind of end nodes for sequences in each group.
Relevant answer
Answer
What do you mean by "related with", "kind", "group", etc? (rule of technical writing #1: terms of art must be defined before use!).
From your example, it looks like you are talking about directed graphs ("digraphs"?) and by a "traversal" you mean something like an euler cricuit (i.e., a circuit passing through all the edges of the graph)? But your paths in your sets seem to cross themselves, so I can't tell!
What do you mean by "depth first"? What is your notion of depth? Do you mean depth of a node or depth of an edge? Surely depth first means doing the deepest (nodes?) first?
Why are there commas in your sets? Are those paths, separated by commas?  Those are sets of paths? Then why do some of the paths revisit s1 and/or s3, and others not?
Perhaps you could start by describing the things you intend to phrase a question about, then state the question(s).
As to why your paths tend to start at nodes with a high out-degree, well, every passage through a node by a path takes up one in-degree and one out-degree, so if there is an excess of out-degree at a node, its edges cannot all be traversed by paths passing through the node and some of its edges must instead be traversed by paths starting at the node. Indeed, as many paths must start at that node as there is excess of out-degree over in-degree.
That perhaps tells you how many paths there must be ... as many as the sum total excess of out-degree over in-degree in those nodes that have an excess of out-degree over in-degree.
We had better hope that that is equal to the sum total excess of in-degree over out-degree in those nodes that have an excess of in-degree over out degree!
In your diagram s1 has 3-out, 1-in; s2 has 2-in, 1-out; s3 has 2-in, 1-out.
Yes, the excesses match.
So it looks like you have exactly two places one path must end, s2 and s3. You  have just one place  to start a path (twice), though, s1. Oh, you have one 3-cycle plus two singleton paths from s1 (to s2 and s3 respectively).
So that all seems perfectly consistent.
Question
11 answers
To the best of my knowledge the book "Verification and Validation in Computational Science and Engineering" (1998) by Patrick Roache is one the most comprehensive engineering books detailing systematic studies to verify and validate numerical work.
I would like to find something equivalent related to spectral methods. What are the techniques to verify that a spectral method code is solving the differential equations correctly and how to validate that the model implemented spectral method code consistently replicate reality?
Relevant answer
Answer
Does it really matter, whether it's *called* ``verification'' or ``benchmarking'', or something else? The important point is to check that the code does work on known examples and its errors remain under control in the cases of interest.
Question
2 answers
Hi guys ,
I've used a structural distance-based measure to compute similarity between each pair of nodes in an undirect graph. Hence, I calculated a distance matrix "D" such that the distance value "Dij" is simply the shortest-path between node i and node j. However, obtained distance values are absolutes (i.e. 5, 19, 3...etc) and I'd like to normalize them, such that : 0<= Dij <=1.
the normalized distance value must be converted finally to a similarity value S such that Sij=1-Dij.
can any one guide me to find the appropriate function to normalize absolute distances ?
Relevant answer
Answer
There are many transformations between resemblance measures
For example you can use
s(i,j) = 1/(1+D(i,j))
d(i,j) = 1-s(i,j) = D(i,j)/(1+D(i,j))
Question
20 answers
I was working on a simulation of Heat transfer block process that contain (liquid, steam and superheated steam). Unfortunately,  all the equations of the heat transfer model consist of a 3rd order P.D.E., I already know the finite element, finite volume and finite difference method. However, applying those methodology never worked with my situation here
So, I wonder if their is a more simpler way for numerical solving of P.D.E. I am currently using MATLAB r2010a to run the simulations
Thanks in advance for any contribution to this subject
Best Regards,
Relevant answer
Answer
These are classic Navier Stokes equations and heat transfer or energy equation. I do not believe these are of 3rd order, however they are highly nonlinear and extremely difficult to solve with your own codes. It is recommended to use already available commercial software like ANSYS, COMSOL, or ABAQUS to solve such type of problems as they have the required solvers for such problems. A large number of example problems are also available with these software. They are mostly finite element or finite volume based software.
P.S. MATLAB have a very elementary PDE toolbox for solving PDEs. These equations can not be solved with that.
Question
5 answers
Hi,
I need some codes for the operation of unit commitment by using GA in Matlab.
Relevant answer
Answer
The easier  thing according to me is that: Write the code in C  and then convert to MATLAB. GA algorithms are available in many different papers.
Question
15 answers
Recently, many researchers think that using a convolutional neural network (or,CNN) on deep learning will bring a powerful and efficient performance to their researches, especially on image and face recognition.  Do you think this approach will continue to affect the era of Big Data Analytics as a useful tool of machine learning. Could you please provide some examples or trials.
Relevant answer
Answer
Yes, nost likely it will continue to impact machine learning / artificial intelligence research for quite some time.
In a nutshell: it is a game changer!
Check out the link below to see a YouTube Video on some of the revolutionary results, e.g. in medical image analysis and speech translation, that have been recently achieved using deep learning and convolutional networs.
Question
3 answers
I am using this code snippet. But I am not sure this is the correct way. Is there any other methods or API to find the execution time of a MapReduce Job?
Date date; long start, end; // for recording starting and end time of job
date = new Date(); start = date.getTime(); // starting timer
job.waitForCompletion(true)
date = new Date(); endTime = date.getTime(); //end timer
log.info("Total Time (in milliseconds) = "+ (endTime-startTime));
log.info("Total Time (in seconds) = "+ (endTime-startTime)*0.001F);
Relevant answer
Answer
I don't think that you can expect getting the same runtime constantly. As long as the standard deviation is not too big, I think you can claim that your data is credible!
Question
8 answers
Expert GIS Programmers and Computer scientists
Relevant answer
Answer
It depends on what mapping technology you are using. But if it is AJAX-PHP based, you could put on your markers a TAG with an onClick property, to execute something like:
information = function(_coords){
return \$.ajax({ type: "GET", url: 'info.php?c='+_coords, async: false }).responseText;
};
and create a info.php to treat your query.
An example is google maps, where you should use something like:
google.maps.event.addListener(marker, 'click', (function(marker, i) {
return function() {
position: new google.maps.LatLng(locations[i][1], locations[i][2]), map: map
infowindow.setContent(locations[i][0]+ " click <a onClick=javascript:information(position)>here</a>.");
infowindow.open(map, marker);
}
})(marker, i));
I hope it helps.
Question
8 answers
Here, I am trying macro modeling of infilled frame in DIANA, to check whether my interface element (CSC model) between infill and frame is working or not. But stress distribution looks uniform for whole panel as shown in attached document. Can anyone please help me to find problem and solution for this.
Relevant answer
Answer
If you say that pic looks uniformly this may be due to visualisation scale.
If you checked the stress (and strains) and they are 0, the interface elements do not work properly. Unfortunately, in the second case there are a lot of possibilities that can cause this effect.
Question
3 answers
Since from their inception, all most all the attacks on McEliece cryptosystem and its variant Niederreiter cryptosystem have been unsuccessful. They are believed to resist even quantum attacks. These two are code-based Public-key cryptosystem whose private key consists of a generator matrix of a Goppa code. All other modified versions of these cryptosystems which were proposed by replacing Goppa codes with other codes have been proven to be insecure. In view of this, what is the structural advantage that Goppa codes enjoy over other families of codes that makes them so strong for cryptographic applications?
Relevant answer
Answer
@AR Reddy: the Goppa codes used in crypto are introduced by elementary means ie
explicit generator matrices. While they can be described by means of AG codes it is hardly relevant there. Historically they were introduced **before** general AG codes on curves. Also I dont understand your comment on the size of the key which depends only on the parameters n,k,q and not on the choice of the family of codes...
Question
19 answers
I am using Hadoop-2.6.0 and Eclipse Kepler/Juno. I have executed my MapReduce Job in both way and I found that running jar file on Terminal takes more time than direct execution for Eclipse IDE. I could not understand the reason behind it. Please help, I could not decide which way I should follow ?
Relevant answer
Answer
1. I don't believe running with IDE is faster than running jar with command line for serious hadoop applications with considerable data size (not toy examples);
2. When there may be exceptions in your runnings, your comparsions would not make sense anymore, i.e. your so-called 'fast/slow" are meaningless!
Question
2 answers
if possible please include the conceptual framework and questionnaires
Relevant answer
Answer
Hi Marife,
As far as I know the questionnaires are not public.
What you can do is go to the training to see what concepts are required. Here you can find a good Study Guide for MOS Word 2013: https://ptgmedia.pearsoncmg.com/images/9780735669260/samplepages/9780735669260.pdf
Hope this helps.
Kind Regards,
Mihai Gavrilescu
I am using an Arduino Nano to acquire data from accelerometer ADXL335.How do I get the Arduino to sample at 100Hz rate?
Question
6 answers
I have tried using delay() function, but it is not working. How do I get the Arduino to sample at 100Hz rate?
Relevant answer
Answer
Hello, Maybe this works: http://electronics.stackexchange.com/questions/55652/arduino-sampling-accelerometer I hope the info is useful... Bruno
Question
7 answers
The specs I am interested in are: number of displays / projectors, total resolution, number of computers the display is driven by. I am currently trying to make some overview.
Question
11 answers
Dear all,
I have developed a new technique which solves linear integro-differential equations of fractional type. This includes Fredholm and Volterra equations.
I am looking for an application which can be modeled into such equation so I can apply my method. It can be any kind of application.
I also solve mixed system of equations e.g.1. A system of multiple Fredholm equations of different order of fractional derivatives (0, 1/2, 1, 3/2, etc..) or e.g. 2. A system of same or different kinds of Volterra equations. So, if there is an application to this kind of equation, it would be great!
I would appreciate your help. Please refer me to an article.
P.S. Only linear equations please.
Thank you very much in advance.
Sarah
Relevant answer
Answer
Dear Sarah,
Please look at the attached files.
Question
156 answers
What is the relationship between Big Data and the Internet of Things (IoT)?
Are they related and/or are they two sides of the same coin?
What are the differences between Big Data and Internet of Things (IoT)?
Is the Big Data a subset of the Internet of Things (IoT) discussion?
Relevant answer
Answer
There is strict connection between Internet of Things (IoT) and Big Data. The real value of IoT is when organizations are able to transform vast amounts of collected data, often in real time, into real business intelligence which can be taken advantage of, e.g. in the creation of new business models. Here comes the crucial role of advanced analytical systems, i.e. Big Data systems, also known as Big Data and advanced analytics (BDAA).
Question
72 answers
What is the similarity between the Internet of Things and Interconnection Networks?
Is the Internet of Things a Kind/Subset of interconnection networks?
Can the Interconnection Networks Research Group work on the subject of the Internet of Things?
Or Suitable other research groups should work on the subject of the IoT?
Relevant answer
Answer
Dear Mohammad Esmaeili,
the Internet of Things (IoT) can be viewed as an extension of the "classical" Internet with various sensor and actuator networks, in particular with different types of Wireless Sensor Networks.
The "classical" Internet is indeed a worldwide networking (interconnection) of different networks with the protocol IP. And the extension of this interconnection of networks by connecting various sensor and actuator networks (i.e. the networks with different things) leads to the genesis of a special and additional internet extension called Internet of Things. For this reason, there is a "great similarity" between IoT and Interconnection of  Networks.
This consideration and definition of the IoT can be found in the Recommendation ITU-T Y.2060 (06/2012). There IoT is defined as follows:
„Internet of Things (IoT): A global infrastructure for the information society, enabling advanced services by interconnecting (physical and virtual) things based on, existing and evolving, interoperable information and communication technologies.”
A general multi-layer model of IoT is presented in the recommendation TU-T Y.2060.
This view of the IoT  –  in accordance with the recommendation ITU-T Y.2060  –  contains my work:
IoT –  Internet of Things (in german)
See the pages:
Page 4: General definition of IoT
Page 5:  Figure 006551
Page 9 : Figure 006553
Page 12 : Figure 006554 (Multilayer model of IoT according to ITU-T Y.2060)
A Network Interconnection Working Group with the focus on "Internet Interconnection with Sensor and Actuator Networks" would still have many tasks to solve  –  in particular in relation to the integration HTTP and CoAP (protocol mapping, proxying inter HTTP and CoAP, ...). See for example:
CoAP – Constrained Application Protocol (die Pages 9 and 11)
Good luck and best regards
Anatol Badach
Question
11 answers
I am using a traditional backpropagation learning algorithm to train a neural network with 2 inputs, 3 hidden neurons (1 hidden layer), and 2 outputs.
The problem is that i obtain some negative weights and biases in the NN after its trained, although the error is pretty low  Unfortunately, for my particular research application, the network parameters need to be positive.
I need to be able to modify the algorithm in a way to produce positive weights and bias after training. I was wondering if there are any techniques with regard to that.
Thanks in advance.
Relevant answer
Answer
In the standard backpropagation  algorithm  the weights   update equation is obtained from the minimization of some error function.
( minimize,   f(e)     ).
where f(e) is the sum of square of error
you can consider  the problem  as   a constrained minimization problem as:
minimize,   f(e)
subject to
wij > 0,    wjk  > 0
and find the modified weights   update equation  that gives only positive values of  weights  wij , wjk.
Question
6 answers
i am looking for MRI brain tumor data-set(databases) with ground truth means manually segmented by radiologist doctor to evaluate the accuracy, precision and efficiency of my proposed algorithm, may i get your help?With regards.
Relevant answer
Answer
Here you can find brain MRI dataset (some include white matter lesions). Hope it helps:
How to improve as a Mathematics Teacher of engineering students?
Question
1005 answers
Dear Mathematicians and Mathematics Teachers: I would like to know how you have improved your teaching methods, techniques and procedures, since bachelor students, even engineers, complain constantly that they do not understand mathematics well.  According to your experience, Which methodology and teaching techniques have you succesfully applied in mathematics teaching for engineering students? Thanks in advance for your contributions. Sincerely Luis
Relevant answer
Answer
Recognition, competition, and collaboration all have the potential to contribute to learning in the classroom.  I think most teachers use forms of gaming in their classroom, but they don’t necessarily refer to it as gamification.  Many web-based companies continue to interweave the idea of games and learning right from the pre school to university years. good Luck
Question
7 answers
Hi,
Currently, I have a difficulties in scheduling tasks which duration is dependent on previous tasks . For example, I have the following two sets of tasks to complete on two different machines M1 and M2:
S1: T1 -> T2 -> T3
S2: T4 -> T5
Each task has duration of 2s. However, If I perform T5 immediately after T1 on the same machine, the duration of T5 can be shorten into 1s. Furthermore, If I perform T3 after T5, the duration of T3 can be shorten into 1s as well.
Therefore, the schedule below has a time span of 4, not 6.
M1: T1 -> T5 -> T3
M2: T4 -> T2
My question is: how do I model this type of "duration dependency" in my CSP model? Or which keywords I should use to search for related information?
Relevant answer
Answer
is the 2 machines have the same power and type?
it seems that executing T2 on M2 reduce the time and the same for T5 on M1
Question
4 answers
I have problems with Arduino AD8232 sensor. I use classic Arduino 1.8.2 for Arduino and Processing for visual processing of the data. According to the graph Arduino sends data in at least 100-300Hz range (graph between two R peaks (cca1 sec) is pretty precise), but when I export raw data I get a value for every approx. 17 miliseconds (around 60 Hz). I don't see why I can't get more precise raw data since more precise raw data is clearly used for the graph estimation. Are there any hardware or software obstacles I am not aware of?
Thanks!
Relevant answer
Answer
Hi Jan,
Have a look on my paper: "BITalino: A Novel Hardware Framework for Physiological Computing". I have described a firmware workflow for the acquisition of physiological signals using the ATmega328P microcontroller which is the MCU within standard Arduino Uno and Mini, etc. That firmware workflow ensures accurate sampling rates and also by streaming data via Bluetooth it may decrease the impact of the 50/60Hz noise and harmonics from the power lead into your acquisition system.
Best,
Jose
Question
45 answers
Dear Friends,
Every software book and research paper on so called software components has been promoting a lie for past 50 years such as "reusable software parts are components for software". The CBD/CBE (Component Based Design, Development or Engineering) for software is defined is using such fake components. Please see attached PDF for more details.
Isn’t it an obvious lie considering this objective reality: The real components used for designing and building large physical products such as airplanes, machines, factory equipment, cars or computers, where each of the product is built by assembling the components. Each of the components is designed to be “assembled” (so that it can be disassembled and re-assembled), and in fact the component is most likely custom designed to fit just one product model, so it is unlikely be reusable or standardized to be reusable across multiple product models or product families.
I have been struggling to explain my inventions rooted in my accidental discoveries in simple laymen terms. I have been facing a huge hostile resistance when trying to expose Facts to prove this: Almost everything in the software books for so called components and research papers about the so-called software components in the context of real CBD/CBE (Component Based Design, Development or Engineering) for software is fiction and/or illusion.
What I am saying may perceive to be arrogant, disrespectful or feel offended. But it is a fact and I am determined to make the software world aware and accept the fact. If I insist that books on botany of zoology are nothing more than fiction and pure imagination or fantasy filled with lies having no basis in reality/fact, I am sure it would offend the scientists or researchers of respective scientific disciplines. But to prove that I am right, all I must do is to document and demonstrate the reality/fact (i.e. simple or obvious observable facts about the animals or trees) to expose the lies.
If all the books and research papers say dolphins lay eggs, if I disagree, I must be allowed to expose such lie by showing proof that dolphins give birth to babies. As a researcher, it is my moral duty and ethical obligation to expose such lies by showing conclusive evidence. It is an ethical obligation of every other researcher to investigate the evidence. I am sure anyone would agree that, no real scientist would defend such lies, if the lies contradict obvious observable facts about the animals or trees. How would you feel, if they insist that the reality must be ignored by insisting that is how researchers or scientists of a scientific discipline have been acquiring the BoK (Body of Knowledge) in the books since the infancy of the discipline?
The authentic source for such knowledge is observing the dolphins (but not the books). If there is a dispute about an information or fact such as an essential nature such as property of the components, unfortunately today no software expert in the world agreed to go out and observe the authentic source – the components. If there is a dispute about a fact such as weather dolphins give birth to babies or deliver eggs, to resolve the dispute the real scientists would go out and observe the authentic source – the Dolphins.
This is what software researchers and experts have been doing: If you take him to sea to prove that the dolphins are delivering babies, after confirming by observing thousands of dolphins delivering babies, what kind of sane scientist or researcher insists that the dolphins are doing it wrong? They are saying that the dolphins are not supposed to deliver babies (they must be delivering eggs), because every book and research paper is saying so.
No real scientist would be offended, if I request that we must go out and discover the reality (by observing the dolphins) to resolve such disagreement of basic facts. In fact, I should be offended, if he snubs or insults me by saying that the reality doesn’t matter or irrelevant. It is unethical, if any one resort to insults or personal attacks to keep me quite from saying such Truth.
The reality is: components are very special kind parts that can be assembled. Even after seeing conclusive proof, many software researchers have been insisting that reusable parts (that can never be assembled) are components and insisting that using such fake components is CBE/CBD, because all the software books and research publications on CBD/CBE for software have been saying so.
Even it is impossible to believe all the text books are fiction, researchers must accept the obvious or observable reality. If the software researchers observe the components, they have no option but concede that everything in the software books and research papers about the so-called software components (in the context of CBD/CBE) is fiction (or lies). But today software researchers are refusing to go out and observe the authentic source (the components), reality and empirical evidence.
Is it necessary for real components to be either reusable or standardized as the CBD/CBE books and research papers have been brainwashing everyone (from the time each of us were impressionable students)? Unfortunately, the components can’t read the software books to know that they are doing it wrong by not following what is in the books or research papers (e.g. expect the components to be doing).
The reality is 90% of large components are custom designed to fit just one product model: What is the nature and essential properties of the components, particularly in case of designing and development of large or complex one of a kind product such as prototype of an experimental spacecraft or fully tested pre-production next generation jet fighters? What percent of components are reusable? What is the percent of components that are standardized (or satisfy the properties attributed to the so called software components)? What is the percent of components that are “assembled”?
For example, let’s consider an average product: What is the percent of core components (e.g. engine, gearbox or power-train) in Toyota Camry that are reusable across product models such as Corolla or Avalon? Each of the core components are “assembled”, because each core components is most likely custom designed for a given product model (so unlikely to be either reusable or standardized).
Are all these custom design components are doing it wrong? Don’t they know what they supposed to be doing? Design, development and/or engineering of each such product (by using such components, which can be assembled but are not reusable) is not CBD/CBE, because every book on CBD/CBE is saying so.  For example, if I say that any part that can’t be assembled (but forced to be composed) is unlikely to be a real-component, one can prove me wrong by showing average components that can’t be assembled, so have no other choice but to compose. Instead software researchers insist that the objective reality is irrelevant.
The reality is that larger parts that can be assembled are known as the components. Such components (that are assembled) are very useful is under statement. The fact is, such large components (that are assembled) are the backbone or lifeblood of the modern industrial world.
The observable facts are foundation to our knowledge. The theories are used to provide logical reasoning to explain the observable facts. Software researchers are doing it wrong: How does dolphins give birth? Weather dolphins deliver babies or lay eggs? The fact is dolphins deliver babies. It would have no effect, even if every text book or research paper on dolphins say that dolphins deliver eggs.
What is the essential property of the components? Weather is it “reusable” or “assembled”? The observable fact is that components are “assembled” (while only a small percent of large core components are reusable). Unfortunately, every software text book, research paper says components are “reusable” (or large reusable parts are defined as components).
Best Regards,
Raju Chiluvuri
Relevant answer
Answer
"It is an error (or lie) to call any part a component, if the part is not conducive to be assembled (or plugged-in)"
No it isn't an error or a lie. Anybody can use any word they like for anything. That's what the section of a paper called "notation" is for. Physicists by "component" generally mean a projection onto a subspace. If the word is used for one thing in some particular subfield of a subfield, it might be used for something else in the next little niche field along, or just reinvented for the moment in any paper's whimsy. So be prepared to address yourself only to the niche subfield that cares.
Question
5 answers
Hello,
Are there researchers who are interested or working on "cross-platform development" (mobiles, web or desktop applications) subject ?
Thanks,
Relevant answer
Answer
Hi Dario, yes It's for collaborations.
Question
4 answers
ِA system can be defined as a set of states and dynamics to govern states' evolution. Are there systems that cannot be described by this formulation?
Relevant answer
Answer
Any system can be described by either input/output relation as in system identification or states of the system as in state space relation. But, if you want more control in the states of the system, it is preferred state space relation. Since it gives deep description for the system. I hope to answer your question.
Question
3 answers
Now a days we are using internet via apps and browser, What is the next alternation of apps and browser  of viewing internet?
Relevant answer
Answer
Both the browser and the app are interfaces for interacting with the services available online, including internet viewing as one of the cases.
Right now a paradigm shift happening in the interaction interface (started in 2015). A substantial part of interactions will move from mobile apps with a GUI to conversational interfaces (also known as chatbots). This trend mainly is driven by the fact that users spend most of the time on mobile devices in messengers, in fact, more than in all other apps. So services need to move to where the users are, i.e. into messengers.
On the other hand, recent development in NLP allowed making the conversational interactions more user-friendly. Practically this means that the users do not need to formulate their intent in a command line-like strict style, but can use a regular language and the machine will still be able to understand what they mean. Such understanding makes it possible to interact with a computer in a declarative style expressing what the user want to accomplish (e.g., "cut out the third face from the left on a group photo"). Before, the user-computer (in app or browser) interaction was mainly procedural. Typically the user needed to execute specific steps in their particular order (e.g., select the face on the photo, press the cut button). There is an expectation that this will enable more common interaction for less tech-savvy users, particularly pertinent for the next 4B people still to join the Internet.
Answering your initial question, some chat apps already provide developers with a possibility of creating custom messages which can be thought of as small web pages. So instead of seeing the whole webpage or an app screen at once, a bot would present small parts of the web in a form of a message particularly relevant to the conversation state.
In summary, chatbots are coming, and they will force us to rethink and redesign many aspects of the human-computer interaction we are used to today. This includes interaction with online services currently available as web pages or mobile apps.
Hope this helps.
Question
3 answers
As we know, The sensors carry the inputs into the system from its environment, and the actuators carry the outputs out of the system to its environment. So there are two kinds of timing constraint.
1. sensor-system
2. system-actuator.
Both of these cases there is a timing constraint. My question is how to test timing constraints?
Relevant answer
Answer
what criteria of testing that you want?
generally, I did a testbed for the web of things which create a dataset from sensors' readings and actuators states, measuring the time accuracy of the dataset
Question
3 answers
The diagonal elements are non-zero. So inverse of the matrix is easily computed by taking the reciprocals of each elements. Is this the complexity O(n)?
Relevant answer
Answer
Yes, if n refers to the number of rows in the matrix (or, equivalently, the number of columns in the matrix) then inverting a diagonal matrix of this size is O(n) time complexity.
Question
6 answers
I have seen in almost all processor, L2 cache associativity is some power of 2 (L1 is generally direct mapped  or some low power of 2). Although in Alpha 21164 super-scalar processor L2 cache (96KB) was 3-way set associative. Which doctrine had convinced the architect of alpha 21164 processor to not to have the associativity some power of 2 ?
Relevant answer
Answer
I don't know any particular reason why 2,4 should be preferred here. One needs to find where in three (or four :) possible locations a certain address tag is located. To do that you need three (or four :) "=" circuits and you will return one high line of three (or four) to a gate. That gate will configure a multiplexer to switch in the right column while the whole set of possibles is activated. So we're talking about a three-way mutex versus a four-way multiplexer.
Shrug. Both need two bits of addressing out of the gate for the multiplexer. It's just that the three-way multiplex doesn't do anything with the 11option, and hopefully there won't be a fourth wire in to the gate that can cause that to be signalled (what does one call a gate with 4 exclusive inputs and two outputs? A demultiplexer?)
The multiplexers will switch 3 ways or four ways. That's it.
Oh .. I do see that one needs THREE two-way multiplexers to switch four ways, but only TWO two-way multiplexers to switch three ways.
So there is a saving there.
But honestly, the extra cost of a multiplexer and its traces is zero compared with the cost of a whole extra set of locations to put the cache data in. Isn't it? Or perhaps it isn't.
Admittedly, one is going to be switching 32 or 64 bits at once. So that's a lot of extra tracery.
It depends on the way things are organised. But it does seem to me that one would perhaps indeed save a lot of real estate with a three-way design rather than a four-way design ... at least in 2D. If it's a multilayer design, all bets are off.
You need a low-level logic and layout expert!
Question
4 answers
My topic is on offloading/handoff in wireless heterogeneous networks. I want to try Machine learning techniques. I am looking around to find which kind of data I need and how to get it. Can I use simulated data to make predictions? which tools can I use to get or simulate this kind of data
Relevant answer
Answer
An up to date research topic is data offloading in cellular networks through device-to-device communication, a key enabling technology for the 5G. In such networks data packets can be directly forwarded between devices (e.g., cellphone) enabling to offload the data demand of the base station.
Several predictions could be applied to such networks. For instance, since the devices must be near to each other to communicate directly (D2D), you could use machine learning techniques to predict when, or what are the chances of two given devices getting in communication range in the near future and therefore decide if it is worth to cache a given content instead of transmitting it through the base station directly, enabling the base station offloading. For a more detailed description please take a look at the following paper:
For this type of project you can use real user contact proximity traces. Several are available at:
Hope it helps!
Best,
Ivan.
Question
11 answers
Donald Knuth said:
Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.
When is that holds to be true? How do I know that I am not wasting my time?
Relevant answer
Answer
I think what Knuth means is that you should not start optimizing because of a "gut feeling" that some algorithms in the program may run faster. You should first finish the program and then measure the performance. If the performance is sufficient, stop; else identify the problematic parts and optimize only as much as necessary.
Faster code is less readable, and a non-critical perceived problem probably solves itself in one or two processor generations (or your optimization becomes obsolete), so keeping it simple is a very good advice. As Peter implies, the actual most critical parts are often in the most unusual and unexpected places.
So the message is to base the optimization decision on facts (measurements) and not on guessing (programmer intuition).
Question
3 answers
in ULT, scheduling is performed at user level (without the kernel involvement). but if one user level thread leave CPU and another will be schedule, then we have to change the value of program counter, Stack pointer and CPU register. without the involvement of kernel, how we can access these registers (hardware)?
Relevant answer
Answer
What's "ULT"? User Level Threading, perhaps?
Do you mean green threads, as in "Java green threads"? (OK, admittedly that's in a virtual machine, but same principle ... more or less).
If scheduling is done at user level then you are running some kind of task controller and it is signalling the threads in and out of the CPU via the kernel.  The kernel does the swapping in and out on the signal, you (user level) do the signalling. Or you've got some equivalent kernel interface to deal with, and you are hooking into that.
Why do you think you have to change the program counter in the CPU? Isn't that all taken care of in the natural course of events? The last instruction before starting a thread will be a "jump register" to the address where the thread left off! Perhaps you could explain the setup involved in "ULT" a bit more. Or perhaps the excpetion copy of the PC will be changed and the return from exception wil bring it in automatically, if you were running in an exception handler for a signal. Anyway it should not be your concern. There are a lot of possibilities as to what you may mean (my bet is on a linux kernel module that hooks into a generic scheduling interface in the kernel and runs a realtime thread to do the algorithmic work).
Question
3 answers
1. Recommend from a pool of five (5) FOS/ERPs, namely: Adempiere, Compiere, Open Bravo, WebERP, and ERP5 which one must be considered for implementation in an organisation.?
2. critical success factors that ought to be considered when selecting such an open source ERP.?
3. how would the Master File in the recommended ERP be maintained; e.g. adding new records, deleting old records, and changing existing records, etc. ?
4. I would also need steps used configuring this ERP which would be aimed at adapting specific system requirements within an organization.?
NB:  I would need references to any suggestion.
Relevant answer
Answer
ERP5 - they are using an interesting approach to automate configuration.See
Wölfel K (2014) ERP5 Starter: Open-Source-ERP-Einführung durch standardisierte Beratung. HMD 49:58–67. doi: 10.1007/BF03340663
Wölfel K (2015) Automated ERP Category Configuration Support for Small Businesses: International Journal of Enterprise Information Systems 11:1–23. doi: 10.4018/IJEIS.2015040101
Wölfel K (2017) Approaches for Automating ERP Category Configuration for SMEs. In: Tavana M (ed) Enterprise Information Systems and the Digitalization of Business Functions. IGI Global, Hershey, PA, pp 66–98
Additionally the system is very flexible.
Question
5 answers
I would like to know about it if any body share their experience.
Relevant answer
Answer
Yes, but what do you mean by "base paper"?
If you invent a term, you have to tell the rest of the world what you mean by it before we can talk to you about it! You haven't answered :-(.
Today, I'd just say something like "google scholar is your friend"! Get thee hence and chat with him/her/it. He/she/it always has something he/she/it wants to pester me to read ("your notifications") when I drop by!
Question
4 answers
IoT , no physical sensors
Relevant answer
Answer
Can you elaborate more ?? Do you mean a simulation tool or something
Question
4 answers
What are mini project ideas on Storage Area Network (SAN) can I give to engineering students, just to make them understand the concept of SAN or some real time applications ?
Question
4 answers
I need classify a set of patterns into two classes. Is it recommendable to use LVQ for that purpose?
Relevant answer
LVQ is a solid, if a bit outdated, method for pattern classification. As Victoria has said, you can use any clustering method such as SOM or k-means for initialization  and then LVQ for training a set of vectors for classification. You might be interested also in checkint out G-LVQ, a genetic method for training LVQ dictionaries
Question
3 answers
Let us assume ALU’s overflow logic were broken for unsigned addition on your MIPS processor. We don’t want to buy a new processor. Could you solve this problem in software? Specifically, create a two real MIPS instruction sequence  which places a 1 into register \$t1 if the sum of \$t2 and \$t3 would cause a unsigned overflow and a 0 otherwise.
Relevant answer
Answer
Hilariously, RG is offering me 3yr old questions a top of my feed, and has lost the "edit" button (only "delete" remains).
Yes, you can calculate the overflow. It's computational logic. Probably the easiest way is to check if a+b < a. If it is, obviously you have wrapped. (Exercise, prove "iff").
Question
4 answers
I've used Petri-Nets for the description of an audio processor. A Petri-Nets simulator is by way of example in Simplorer (ANSYS) implemented.
Relevant answer
Answer
Dear Peter, thank you for opening my question. This question is a result of another question:
In the future we have in human-machine-interfaces different architectures for audio signal processing, video signal processing and chemical structure analysis. For the inputs of these Micro- or Nanoprocessors we need not a Register-Transfer Level (RTL). We have as input only integrated converter for one- or multidimensional data streams. Also we need not a clock-synchronization. These Micro- or Nanoprocessors work with innate asynchrony clock, normally event driven. The output will be success over a data bus-system. At least the RTL is for development ancillary.
Primary for development is the structure of PETRI nets. In my “Procedure of contemporary ascertain for characteristic value, harmonics and non-harmonics of quickly alterable signals” I have used seven levels. Each level contains a PETRI net with exactly one marking. See the link
For better understand my development language: Every PETRI net consists of states, events and actions (compare with SML code of Simplorer). On every level of PETRI nets is only one state marking.
Regard Christian
Question
10 answers
I know the L1 is called first before L2 and fourth but why? Is there anybody with a theoretical and practical reason?
Relevant answer
Answer
There is no theoretical reason. The most compelling practical reason for this is cost. Maybe a minor practical reason is that is hard to implement a fast search for a larger cache.
The thing is that fast memory is very expensive. Otherwise we would use computers with the number of registers equivalent to several gigabytes of data. Because the processor logic would be quite complex for billions of registers (and thus expensive) we are using a L1 cache. This cache is still quite fast, but having a larger size of this kind of fast memory would again cost a lot. Therefore, the trade-off is again a small size. In the beginning of computing it would only take about 2 CPU cycles to access RAM (which by today's standards is also really small). With every increase in CPU speed another level of cache is needed to compensate for speed differences between CPU registers and RAM. Large and slow memory costs as much as fast and small memory. This is always a trade-off between size, speed, and money. You cannot optimize for all three at the same time. The main reason to have a cache is to hide latency to the RAM (or to the L3 cache, or to the L2 cache). Sure, this only works for programs that adhere to memory locality. Otherwise, you would have roughly 200 CPU cycles without any work; just waiting for data from the RAM.
Question
4 answers
I have an output signal such as the one in the attached file. How can I infer segments in the signal that capture when events could have possibly happened inside the system?
Relevant answer
Answer
Hi Ramy,
in terms of segmentation, Keogh's paper is a classic:
For the detection issue, try searching for: anomaly detection, state change detection, and event detection.
Regards,
Philip
How do you find the driving capability (whether it is high or low) of a digital CMOS circuit based on the arrangement of NMOS/PMOS transistors?
Question
40 answers
Driving capability for a particular digital CMOS circuit may be high or low. Ex. Driving capability of transmission gate based network is low. The arrangement of transistors impact on the driving capability of a circuit. How to decide the driving capability of a CMOS circuit is low or high based on the arrangement of transistors?
Relevant answer
Answer
Hi, It's not just the topology (i.e. the arrangement) that matters, but the sizing of the transistors is equally important. Moreover, the driving capability depends on the load that the gate is supposed to drive. Usually, the load is referred to the input capacitance of a reference inverter, which defines the FO1 load (fanout-of-one). Then, the FO4 is used to express a *typical* load (i.e. 4x the FO1 load). These loads are also used to define typical *delays* in a technology. E.g. an FO4 delay is the delay of that reference inverter loaded with 4 identical inverters. An inverter that has 2x driving capability with respect to the reference inverters, will likely have 2x wider transistors. In summary, you should consider a reference load (e.g. FO4), and a reference driving strength (e.g. that of a reference inverter), and compare the driving capability of your gate to that reference gate under the same load. Hope this helps!
Question
4 answers
I have a good computer for simulation (3.7 GHz and 32 GB ram memory). I would like to utilize the full power of my computer for my simulation. When I am running the ANSYS simulation, ANSYS is not utilizing its full power (only 10% of the total power). I have specified the sparse solver with incore option.
eqslv,sparse
bcsoption,,incore
But still the APDL is using only 10% of the total power.
Is there any way to maximize total memory usage of ANSYS APDL?
Relevant answer
Answer
Dear Ugur,
Thanks for your answer. Still some more queries, how can we increase the initial ANSYS memory allocation so that the amount of memory available to the sparse solver exceeds the in-core memory requirement ?
Regards
Question
2 answers
Is there any evidence that exists to prove seed axiom for CBSE is valid? If any one has tried to test the validity of seed axiom, kindly provide evidence?
The origins of seed axiom for existing CBSE (Component-Based Software Engineering) is usually attributed to 1968 NASA conference on software engineering. Dr. Douglas McIlroy’s speech in ‘1968 NATO software engineering conference’ on “Mass produced software components” is widely credited for launching research on ‘software components’ and CBSE.
One of the stated goals was to build software products by assembling components (i.e. COTS- Commercial Off The Shelf) from 3rd party vendors as hardware engineers build computers by assembling COTS. This goal is unrealistic, because hardware engineers deal with product models belonging to one product family while software engineers dealing with products belong to hundreds of product families (e.g. compilers, games, OS, MS-Word, PPT or ERP etc).
Please kindly understand physical components are very special kind of parts (having very unique properties), so no other kind of physical parts can be components. Today term ‘component’ (in software engineering) is used as synonym to ‘useful parts’ or ‘parts conforming to a useful model’. For example, many kinds of so called software components were defined, where each kind of component is either a software part having a given set of useful properties (e.g. reuse or standardized) or conforming to a kind of so called component model.
Isn’t it an error to define physical parts (e.g. cement, TMT-steel, plastic, metal alloys or semiconductor wafers), just because each of the parts needs complex process to make and standardized or highly reusable across many product families. The existing definitions for so called software components may be acceptable in the formative years in 1960s of software engineering, when primitive programming languages such as FORTRAN or COBAL are cutting edge technologies. Are such definitions for so called software components still valid today, since there are huge advancements in programming languages and our knowledge or expertise in software engineering and computer science?
Aren’t these baseless assumptions (originated in the early formative dark ages of software engineering) preventing us from searching for real software components that are logically equivalent to the physical functional components for achieving real CBSD (CBD for software) for large software products, where real-CBSD must be logically equivalent to the CBD (Component-Based Design) of large one-of-a-kind physical products (e.g. prototype of a new kind of spacecraft or an experimental jet-fighter)?
Of course, there is no need to search for such real software components, if there is an irrefutable proof to show that it is impossible to discover such real software components for achieving real CBSD. In the absence of such proof, relying on flawed seed axiom for advancing CBSE is insidious because such errors in seed axiom certainly sidetracks progress & eventually ends up in crisis.
The longer such error in seed axiom goes undetected the harder it would be to discover Truth for exposing the error. Relying on seed axiom (having undetected error) for advancing any branch of science or engineering (by working very hard for many decades or centuries) leads to altered reality and results in evolution of deeply entrenched paradoxical paradigm.
P.S: The geocentric paradigm evolved from the assumption that “the Earth is static at the center”, so this assumption was seed axiom for geocentric paradigm. The heliocentric paradigm evolved from the assumption that “the Sun is at the center”, so this assumption is seed axiom for heliocentric paradigm. Existing paradigm for CBSE evolved from untested definitions for so called software components, so the definitions are seed axiom for the existing CBSE paradigm.
Getting wrong answer to root question “which planet is at center” side tracked scientific progress and resulted in a scientific crisis. Likewise, if there exits an accurate answer to root question “what is a component (i.e. accurate definition or unambiguous objective description for the components)”, getting wrong answer to the root question certainly side tracks progress of CBSE paradigm and results in software engineering crisis.
Is there any evidence to prove that, it is impossible to find right answer to this root question for the CBSE paradigm? Many software researchers insist that no right answer could exist for the root question, but no one ever showed any proof to indicate that any one ever even have investigated to find right answer.
Relevant answer
Answer
@Raju: Excellent question!
Component-based development works well in the domain of hardware because the domain of hardware is organised into highly specialised sub-domains; all the different disciplines of engineering. In such a setting, it is highly unlikely that components from a Car will be used in a Ship, nor vice versa --- as you have rightly pointed out.
Component-based development will work well in Software Engineering, too, as soon as we develop highly specialised sub-domains here as well. Currently, however, "Software Engineering" is still far too general to date. Obviously we cannot re-use Compiler components in software systems which are not Compilers, (for example). However in those areas, where specialisation is progressing, the re-use of components is becoming a reality in software, too: Think about, for example, the computer-games industry, wherein the so-called "game engines" are already being used and re-used in a variety of different games to be played.
[14-Nov-14]
Question
17 answers
Java parallel programming code.
Relevant answer
Answer
Yes and No. In a multicore CPU, some subsystems are common to all the cores. The "computing" work is physically separated in the cores, some communications (bus) are common. I confess I have no exact idea of this design on the global performance. However, I suspect that on modern 64bit CPU's that embed enough first levels caches, the bottleneck of the computer is probably more located at the HD, RAM or network levels... It was always the case in my own experiences.
Just one tip: if possible, keep one core free for OS and JVM operations...
Question
8 answers
I've successfully imported a webpage into NVivo but I can't successfully select (in order to add a node) text that is hyperlinked, or pictures. Is this a limitation of the software, or of the user?
Does anyone have any suggestions for an alternative to NVivo that can code pictures (a la screenshots)?
Before NCapture I was using Webpage Screenshot to capture websites in full, and NVivo didn't like those pdfs much either as it couldn't recognise text or select areas (think: click and drag to make a box) to code. Does anything else have this type of capability?
Relevant answer
Answer
If you install a .pdf printer you can output the web page as a pdf with full text recognition, you will then be able to tag it in NVivo. It works the same as printing the page to paper, but it 'prints' the document to a .pdf instead.
Here are some Mac options:
Here are some Windows options:
Question
8 answers
I wrote an image processing algorithm on MATLAB and I want to know the CPU time consumption of the same algorithm if it was written using OpenCV.
Relevant answer
Answer
The only way to answer your question is to write the same algorithm using OpenCV and time it on the same machine.
Question
9 answers
I am looking to find standard reinforcement learning implementations in C, C++ or Python, to be able to adapt to my problem which is compiler optimizations.
Looking at the link attached bellow it seems all the C++ implementations are bad linked. Anyone has any suggestions?
Relevant answer
Answer
Dear Amir H. Ashouri,
You may get details of a PhD work related to this topic. Hope that may help you. With best wishes, Good Luck.
Question
15 answers
I have a workstation equipped with a Dual Intel E5-2650 v2 processor and 64 GB RAM. I am running heavy computational simulations which needs about a month for a single run. I heard that RAM Disk is a good option for improving the performance. Can anybody tell me the proper way of setting the RAM Disk?
Relevant answer
Answer
I doubt that a ramdisk will help you much.
Linux systems use Virtual Memory which consists of pages.
Most if not all Linux systems now will cache files if the programs in memory are not using those pages.   For data that only read, not modified, if the files are small enough to fit in memory (what you would need for a ramdisk) then you would read once and never again,
For writes, the data must eventually be flushed to disk.  Pages with modified data are called dirty pages, so any cached write would be in a dirty page.  You can affect how many dirty pages there can be and how often they are flushed to disk via syctl settings like  vm.dirty_background_ratio  and vm.dirty_writeback_centisecs these allow more pages which have been modified to stay in memory so writes to a file will not be flushed to disk as often.
sysctl -a will show you all syctl settings.  You can modify on the fly with
sysctl -w
commands, or make a permanent change by modifying the file /etc/sysctl.conf
Question
8 answers
Reduce the hand tremors in assistant medical robot
Relevant answer
Answer
Hi, Albert,
Well, in some application maybe. But I doubt about the concept of automation in surgery... hehe, even now in laparoscopic surgery, surgeons are still arguing if it is necessary to use robots. So far, seems only in some procedures, in urology, it is well received that using surgical robot has superior performance than direct operation with conventional tools.
After all, it is the critical decision making, at least in the near future, artificial intelligence is not capable of replacing human. So automation... is good for those highly repeatable procedures, like those hectic work in industry. But in medical field, it is not impossible, but would take quite a long way to go. That is my personal opinion.
Question
17 answers
I would like to know how I can perform the velocity of the CCSD(T) calculation. Should I put more CPU or increment the RAM memory like in MP2 calculation? What is a suitable way? Thank you very much.
Relevant answer
Answer
Can you please  tell what program or package you are using to calculate the CCSD calculation? Normally it depends on the cluster size. If you are calculating for small clusters (bare clusters upto 2 atoms) then 3-4 Gb is more than enough. Incase you have ligands attached to you clusters then it requires more memory.  If you are working with complexes with transition metal clusters then I think your memory requirement should be higher.
Question
27 answers
To design a controller for multi-degrees of freedom actuators which type of controller is better (Sliding Mode Controller or Backstepping Controller)?
Relevant answer
Answer
There are 4 issues I have seen in this discussion:
1. Backstepping can compensate theoretically exactly only uncertainties decreasing together with the state variables. No comparison between backstepping and SMC in this sense.
2. Matched and unmatched uncertainties
For perturbed chain of integrators there are no unmatched perturbations. If you will differentiate the output till the system order all the perturbations will be matched! That is why the only reasonable to control the systems with unmatched uncertainties is a combination between backstepping and SMC, where backstepping compensates state dependent part and SMC(see Estrada et al IJRNC,27(4),2017,DOI : 10.1002/rnc.3590,Automatica 2010(11), TAC 2010(11), J. Davila TAC 2013, Ferreira et al Journal of Franklin Institute, 2014, 351(4),doi:/10.1016/j.jfranklin.2013.12.011, Automatica, September 2015, Vol. 59, 10.1016/j.automatica.2015.06.020)
3. SMC can not be smooth. Sliding mode (for definition!) is a motion on the sliding surface!
4. Chattering
4.a. There is no way to keep the sliding mode and eliminate the chattering.
4.b. It is wrong opinion that it continuous HOSM controllers (like super-twisting, Kamal et al DOI:10.1016/j.automatica.2016.02.001, Torres et al doi :10.1016/j.automatica.2017.02.035, Laghrouche et al TAC,2017) everytime produce less chattering that discontinuous ones (see https://www.researchgate.net/publication/317229996_Is_It_Reasonable_to_Substitute_Discontinuous_SMC_by_Continuous_HOSMC).
It is true for the systems with fast actuators only.
Question
7 answers
Given an extrinsic matrix, it is granted that it transforms a point in world coordinate to a point in camera coordinate.
Let E be the extrinsic matrix:
R11 R12 R13 t1
R21 R22 R23 t2
R31 R32 R33 t3
0       0      0    1
Technically, extrinsic matrix gives you the orientation of world coordinate frame with respect to camera coordinate frame. It also gives you the translation of world coordinate frame with respect to camera coordinate frame. Thus, t1 should be distance from camera coordinate origin to world coordinate origin in x direction and t2 in y direction and t3 in z direction. This indirectly means that camera coordinate origin is located -t1 in x direction, -t2 in y direction and -t3 in z direction from the world coordinate origin. That is, (-t1, -t2, -t3) is the camera coordinate origin in world coordinate frame. Camera coordinate origin should correspond to camera position and thus (-t1, -t2, -t3) should correspond to camera position in world coordinate frame. However, it isn't so.
Let Xw be the camera position in world coordinate frame and Xc be the camera position in camera coordinate frame. We need to find Xw.
Then:
E * Xw = Xc
Xc is (0,0,0) and hence:
Xw = -1 * (R)' * (t), where R (3*3) and t (3*1) are the rotation and translation components of E respectively.
Why is Xw not equal to (-t1, -t2, -t3)?
Relevant answer
Answer
Hi Dennis,
You have answered your own question by deriving Xw...
If you are looking for the intuition of why this is the case, think about it this way: the extrinsic matrix transforms a point from the world coordinates into the camera's coordinates. it applies rotation first, and then translation:
Pc = R * Pw + t
The order matters. To find camera's location in the world coordinates, we need to reverse the transformation: apply the negative of the translation, followed by the inverse of rotation.  That is why Xw = R' * [-t1, -t2, -t3], and not just [-t1, -t2, -t3].
Question
26 answers
I'd like to know about the best method to defined a fuzzy rule base for fuzzy logic controllers, most of references said experience knowledge is the best way, but what are the other parameters?
Relevant answer
Answer
Hi
Question
2 answers
We're planning to conduct a basic sensory and motor audit of patients within the QE Hospital ICU here in Birmingham and in support of some of our virtual restorative environment projects I'd like to include a couple of popular tests (EFT) addressing cognitive style and mental rotation (Shepard & Metzler).  Is anyone aware of (a) valid "cut-down" versions of these tests and (b) the existence and availability of computerised versions?
Relevant answer
Answer
Yes. We have a Witkin´s GEFT version on-line for research purposes. Write to  christian.hederich@gmail.com
Question
10 answers
Hello Scholars,
I am trying to make a common domain for input and output of a function. I divide this domain in input scheme and output scheme.
Input scheme contains a set of possible inputs of a function and output schemes contain a set of possible outputs of a function.
For example: function parameter, global variables, inFile pointer are members of input scheme and return variable, reference variable, outFile pointer are member of output scheme.
My question is what else can we add in this domain? These members can be dependent or independent.
Can i add i/o device in this domain?
Relevant answer
Answer
What's "input/output element of a function"? In fact, what is "a function"?
The problem is the word "element". Elements are members of sets. What set are you talking about?  Do you perhaps just mean "input or output of a function"? (no "element" word).
Well, if your inputs and outputs are interchangable, what's the point in distinguishing them at all? Then both inputs and outputs are in some domain D. And you have a side-effecting function f:DxS->DxS, where S is the global state. If you have addresses A as part of D, that's just somethig you would write down, thus:
A <= D
So please write down your semantic domains like that and we can see what you mean. Whatever you think of, call it something, say B, C, D, E, etc, and just write down
B <= D
C <= D
and so on.
If you want to firther specify some interior detail in some domain, then go ahead. I already suggeted that addresses might consist of read and write functions, so
A = (S->V) x (S->V->S)
might be a suitable definition. I.e. each address is a something that gives you a value v:V (bearing in mind the current global state s:S) or allows you to change something.
Question
35 answers
In general the dataset used for classification using Lilinear classifier is scaled in the range 0 to 1. now if the scaling range  of 0.1 to 1 is feeded to the same classifier, would it effect the result?
Relevant answer
Answer
In theory it would not matter which normalization scale you choose for the data set. It only affects the visual separation. I would probably speculate that large margin classifiers like SVM could handle a collection of data with a mixture of both 0-1 and 0.1-1 normalization.
Question
7 answers
I transferred 2D slices in 3D single Matrix.
Relevant answer
Answer
It seems to be a volume rendering problem. There are some free implementations on the web for volume rendering.
Question
2 answers
I'm so glad that ask my third question on my favorite site.
Infact i ran into multiple choice question in recent exam on Compiler Course.
Suppose T1, T2 is created with SLR(1) and LALR(1) for Grammar G. if G be a SLR(1) Grammar which of the following is TRUE?
a) just T1 has meaning for G.
b) T1 and T2 has not any difference.
c) total Number of non-error element in T1 is lower than T2
d) total Number of error element in T1 is lower than T2
My solution:
we know table size and state of LALAR(1) and SLR(1) is the same. but someone say number of reduced state in LALAR(1) is lower than SLR(1) (free space in LALR(1) is more than SLR1(1) because using lookahead instead of follow) and so (d) is correct. but in answer sheet we see (b) is correct. anyone can describe it for us? which of these is true?
Relevant answer
Answer
b
Is there something better than entropy for measuring the variability of the output from a calculation?
Question
8 answers
Addition mod 2 is always zero on the diagonal x=y yet an entropy calculation says that the output has entropy 1, given independent inputs x,y of total entropy 2. Common sense says that if you are trying to code up your calculations so that a baddie cannot recognize them, then you want the output to have as much variability as possible so the output can't be relied on by the baddie to take a particular form - even statistically. So a lot of entropy is good. That is what addition has. Yet a clever baddie will just put in inputs to addition mod 2 that are equal - he/she does not know what they are - and get an output that is guaranteed to be zero mod 2, which is knowledge the baddie ought not to have had.  Why doesn't the entropy measure give any clue that there is an easily selectable full subspace of the inputs along which the output has variability zero, even though the output taken over the whole set of input possibilities formally has 'a lot' (1 bit) of variability, as the entropy measure says? Yes, entropy says we lost one whole bit of variability from the inputs, but there is no hint that we can easily force the output to have no variability at all just by putting x=y. Any insights into this would be appreciated and I want to know  if there is something better to look at here than entropy.
Relevant answer
Answer
I am afraid I am a little bit far from the subject, but found your question interesting, though. What about conditional entropy, or mutual information? I think that, besides entropy, which is a measure of how big the information space is, you want a measure of how the occurrence of an input, or an output, correlates with other occurrences. If different occurrences are statiscally independent, mutual information should be zero. If it is non-zero, then by knowing one, you reduce the uncertainty of knowing the other. What do you think? Cheers, Fernando https://en.wikipedia.org/wiki/Mutual_information
Question
5 answers
Please answer in relation to image analysis on kidney biopsy.
Relevant answer
Answer
??? Convert to text-file? How, but more important, why?
How many classes are there in kidney images? A lot, I guess; depends on the diseases that could have been diagnosed.
You want to diagnose yourself, using image-processing? Good luck!
Or do you want to train an ANN, using diagnoses made by doctors? If this latter is the case, the classification is already done by the doctors, so ask THEM what method they used. Don't be surprised when that method is not readily captured in 0's and 1's.
Question
4 answers
We have HP workstations one with 8 core Intel processors and 16 GB RAM and another with 12 core processor and 16 GB RAM. Are we able to cluster these two workstations and implement OpenMP threaded C code?
Relevant answer
Answer
In general this is possible. But, the usual way to go is to use MPI instead of OpenMP to parallelize on two different physical computers. There are things like PVM which allow you to write OpenMP programs to work on multiple computers, but these kind of tools have very poor performance. The performance gain you will get is so small (because of the overhead) that it is not worth the effort.
My advice is to use MPI instead. It will scale quite well, but then you are required to rewrite your software to use MPI. Look for an MPI library (there are several) that already supports MPI 3.0, e.g. OpenMPI. MPI 3.0 allows for true shared memory so that you do not have to replicate most of the data for processes that run on the same machine. For two computers it is still feasible to call mpiexec by hand. But, if you want to have more comfort you should look for a job scheduling system. Quite easy is HTCondor (although, quite frankly, I recently had a lot of trouble with that in a company's network environment). HTCondor will even allow you to only use other workstations when nobody is currently working on them. A more sophisticated system would use the PBS scheduler instead.
Question
4 answers
I need install a Snort platform, but the new version does not support connecting MySQL. I have used cygwin to compile Barnyard2, but it failed. Are there some parameters that I have to modify? Can anyone help me?
Relevant answer
Answer
have you read the error? when you try to compile & build barnyard2, could you add a little bit of detail{the error for instance};
Do you think parameters should be designed like impact factors to decide on the standards of upcoming research issues as well?
Question
1 answer
Areas related to computer science and engineering
Relevant answer
Answer
Hi, Chetna: Interesting question. As we know, Information systems journal rankings and ratings particularly help scholars focus their publishing efforts and are widely used surrogates for judging the quality of research. Over the years, numerous approaches have been used to rank IS journals, approaches such as citation metrics, school lists, acceptance rates, and expert assessments. However, for some the results of these approaches often conflict due to a host of validity concerns! In a recent research study (Lowry et al., 2013) argued that their current scientometric study made significant strides toward correcting for these limitations in the ranking of mainstream IS journals. They compare expert rankings to bibliometric measures such as the ISI Impact Factor™, the h-index, and social network analysis metrics. Among other findings, they concluded that bibliometric measures provide very similar results to expert-based methods in determining a tiered structure of IS journals, thereby suggesting that bibliometrics can be a complete, less expensive, and more efficient substitute for expert assessment. The authors also find strong support for seven of the eight journals in the Association for Information Systems Senior Scholars' "basket" of journals. A cluster analysis of our results indicates a two-tiered separation in the quality of the highest quality IS journals-with MIS Quarterly, Information Systems Research, and Journal of Management Information Systems belonging, in that order, to the highest A+ tier. Journal quality metrics fit nicely into the sociology of science literature and can be useful in models that attempt to explain how knowledge disseminates through scientific communities! Lowry, P. B., Moody, G. D., Gaskin, J., Galletta, D. F., Humpherys, S. L., Barlow, J. B., & Wilson, D. W. (2013). EVALUATING JOURNAL QUALITY AND THE ASSOCIATION FOR INFORMATION SYSTEMS SENIOR SCHOLARS' JOURNAL BASKET VIA BIBLIOMETRIC MEASURES: DO EXPERT JOURNAL ASSESSMENTS ADD VALUE?. MIS Quarterly, 37(4), 993-A21. Hope this helps! Nadeem
Question
36 answers
I'm working on a molecular simulation package in Linux mint, I want to make an executable file to start simulation by ''make'' command in the terminal, but I receive this error
defaultCompiler -c eispack.F
make: defaultCompiler: Command not found
make: *** [eispack.o] Error 127
Relevant answer
Answer
gfortran will be fine (backward compatible), but you might want to read the installation, README, or other documents.
The error you received, essentially tells you that the make file is trying to execute the command 'defaultCompiler'. You will find that there is no such binary on your system.
Usually there is a step in preparing a distribution prior to compiling that should replace 'defaultCompiler' with the appropriate command, e.g., gfortran or g77. It would appear that the step was not performed or not successful. Precisely what the step is, depends on the distribution. It is therefore impossible to say what should be done to correct it.
Which method would you deploy to set the optimal decision threshold to plot a meaningful sensitivity vs. (1 - specificity) graph, 'aka' "ROC curve"?
Question
40 answers
I am currently assessing the robustness of various Artificial Intelligence-based classifiers on the same series of highly dimensional and nonlinear datasets. I would like to ask you if you could please suggest me how you would set the optimal decision threshold or cut-off value for a 'tailor-made' ROC curve. Many thanks.
Relevant answer
Answer
Dear Costanzo,  Sorry for my late reply. Thanks for your answer.  I agree with Prof P R Yarnold. Firstly, could you please specify with respect to which other method and in which manner the Youden's J statistic you suggested is 'more formal'? Did you mean it seems to be more 'rigorous' than the one deployed in Prof P R Yarnold's paper? If this is what you think, how is it 'more formal' and why? Secondly, I am very new to the 'Youden's J statistic'. Talking 'ex abrupto' about its optimisation sounds pretty vague to me. How would you optimise it, please? Were you thinking of any algorithm in particular? Best Regards,  Luca
Question
6 answers
Swarm Intelligence (SI) techniques are more and more used by researchers in a wide range of areas: physics, economy, materials, biomedical sciences, computer sciences, engineering...
Recently, new published papers propose enhanced variant of SI techniques to improve the optimal solutions and achieve better results than other conventional optimization techniques, namely ACO, PSO, differential Evolution (DE)...
Does anyone have worked on these enhanced variant of SI techniques?
Regards.
Amin
Relevant answer
Answer
I have some experience in developing constrained versions of ACO and PSO  and hybridizing them for better performance. You can find some of these works in my researchgate contributions.
Question
2 answers
In a Weighted Directed Graph G (with positive weights), with n vertex and m edges, we want to calculate the shortest path from vertex 1 to other vertexes. We use 1-dimensional array D[1..n], and D[1]=0 and others filled with +infinity for updating the array we can just use Realx (u, v).
If D[v] > D[u] + W(u, v) then D(v):=D(u)+W(u,v)
in which W(u, v) is the weight of u-->v edge, how many time must we call the relax function to ensure that for each vertex u, D[u] equals to length of shortest path from vertex 1 to u?
Solution: I think this is Bellman-Ford and m*n times we must call.
Could anyone help me? In Dijkstra and Bellman-Ford and others we have Relax functions. How do we detect which one of them?
From CLRS Book:
The algorithms in this chapter differ in how many times they relax each edge and the order in which they relax edges. Dijkstra’s algorithm and the shortest-paths algorithm for directed acyclic graphs relax each edge exactly once. The Bellman-Ford algorithm relaxes each edge |V|-1 times.
Relevant answer
Answer
Hello,
I will try to give you the answer for the Dijkstra algorithm. From what I remember, you only need to relax each edge at most once, as stated in your question. Here is why. Starting with vertex 1, let's assume it is connected to vertices v_1,...v_i. The relaxing function will write for all k in {v_1,...,v_i}, D[k]=W(1,k). Then, you consider that vertex 1 is out of the graph, and apply the same function on vertex v_j such that D[v_j] is minimal in D (except for D[1]). Without loss of generality, v_j, has edges to vertices {v_l,...v_z}. For k' in {v_l,...v_z}, the function writes: D[k']=min(D[k'],D[v_j]+W(v_j,k)). Once this is done for all k', v_j is taken out of the graph, and the function is applied to the vertex v_p that has minimal D[v_p] (except for D[1] and D[v_j]).
By iterating this process, you will end up:
- applying the function only once to each vertex. Since each vertex is taken out after the function is applied on it, each edge is relaxed exactly once in a non-directed graph, and at most once in a directed graph (depending on whether or not you have vertices linked by a double edge, with one in each direction).
- with for all k in [|1,n|], D[k] is the cost of the shortest path from vector 1 to vector k.
This algorithm can be used in a more general way in the presence of cyclic graphs, when looking for the shortest cycle. In this case the complexity is a bit higher.
The wikipedia article on Dijkstra's algorithm is a good resource to understand it : http://en.wikipedia.org/wiki/Dijkstra%27s_algorithm
I hope this answers your question.
Question
3 answers
ACC needs 'Headway distance' and 'Desired speed' as input from user.
If vehicle is just stared and without driving manually for certain speed or distance, can ACC be activated ?
Relevant answer
Answer
That depends on the type of ACC. The "old" one can only be activated above 30km/h. The new ones (ACC stop and Go) work until stand still. On of the reasons is that the old brake systems were too uncomfortable and noisy at low speed (the two piston pump is inducing too much vibration leading to an oszillating deceleration and noises).
Question
6 answers
I am using hadoop-2.6.0 and eclipse kepler on Ubuntu-14.04. I am confused about library file of hadoop, as there are so many jar files in contrast to hadoop-1x. So how many jar files should be added in any MapReduce project. For example what should add in word count program ?
Relevant answer
Answer
You can start by using maven project and then search for dependencies in the project you should choose hadoop-hdf.xxx.jar and hadoop-mapreduce.xxxx.jar and hadoop common-xxx.jar. The maven will hold the rest of the other jar files dependencies
in case you don't want to use maven project, i prefer to create a user library with the following jars to avoid adding them in future projects, you need to change the version of hadoop from 2.2 to 2.6
/hadoop-2.2.0/share/hadoop/common/hadoop-common-2.2.0.jar
/hadoop-2.2.0/share/hadoop/common/hadoop-common-2.2.0-tests.jar
/hadoop-2.2.0/share/hadoop/common/hadoop-nfs-2.2.0.jar
/hadoop-2.2.0/share/hadoop/hdfs/hadoop-hdfs-2.2.0.jar
/hadoop-2.2.0/share/hadoop/hdfs/hadoop-hdfs-2.2.0-tests.jar
/hadoop-2.2.0/share/hadoop/hdfs/hadoop-hdfs-nfs-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-app-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-common-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-core-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-hs-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-hs-plugins-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-jobclient-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-jobclient-2.2.0-tests.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-client-shuffle-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/aopalliance-1.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/asm-3.2.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/avro-1.7.4.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/commons-compress-1.4.1.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/commons-io-2.1.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/guice-3.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/guice-servlet-3.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/hadoop-annotations-2.2.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/hamcrest-core-1.1.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/jackson-core-asl-1.8.8.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/jackson-mapper-asl-1.8.8.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/javax.inject-1.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/jersey-core-1.9.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/jersey-guice-1.9.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/jersey-server-1.9.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/junit-4.10.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/log4j-1.2.17.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/netty-3.6.2.Final.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/paranamer-2.3.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/protobuf-java-2.5.0.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/snappy-java-1.0.4.1.jar
/hadoop-2.2.0/share/hadoop/mapreduce/lib/xz-1.0.jar
Question
2 answers
I have executed my MapReduce program on these two version of Hadoop and observe that Hadoop-2.6.0 takes about double time in comparison to Hadoop-1.2.1. And also confused about the jar files of Hadoop-2 which is complicated while in Hadoop-1 it was so easy to add jar files. I am using eclipse kepler and Ubuntu 14.04 64 bit. Please help me.
Relevant answer
Answer
Try to look at the log files to characterize your job - for instance, in which part of the job execution does the YARN/MR2 run take longer - map phase? reduce phase? You can also check this article: http://blog.cloudera.com/blog/2014/04/apache-hadoop-yarn-avoiding-6-time-consuming-gotchas
Question
4 answers
Some people say VMI is coined based on the concept of Virtual Desktop Infrastructure (VDI), what is your definition of VMI?
Do you think VMI is practical?
What are the use cases for VMI?
Do you know any VMI reference architecture you can share?
What are the pre-requisites before VMI can be implemented?
Relevant answer
Answer
Are you referring to mobile devices in your query?
If so, than mobile device virtualization is next frontier after server virtualization. The difference is in the underlying instruction set architectures, and the support they provide for efficient virtualization. For example x86 and ARM. Yes virtualization of mobile devices is practical because they are rapidly increasing in power and memory. The use case is to reduce hardware and host multiple OS on single hardware.
Question
9 answers
I need some basic scientific paper list of speech to text technology.
Relevant answer
Answer
This is a rather broad field. It would be helpfull if you could be more specific.
In any case I guess it is worth looking at the Publications of the CMU SCS Speech Group
Question
16 answers
Please kindly let me summarize the reality of CBD of the large physical products: In case of the CBD (Component Based Design) of physical products the following things are taken for granted: Nearly 97% of the features and functionality are implemented as components, where each component is 100% free from spaghetti code hence nearly 97% of the product is free from spaghetti code. Each component can be disassembled (or unplugged) to refine or redesign for making it as best as it can be (e.g. to optimize and to perform at its best). This component can be tested individually to make sure it performs its functionality as expected before assembling into the product.
After assembling into its container-component (or product), each component is tested to make sure its coupling interfaces fit perfectly and it performs as expected within the product by properly collaborating with other components. If it doesn't fit perfectly or to further improve its performance and functionality, it can be unplugged (or disassembled) for redesigning for refining its interfaces or functionality and testing it individually (outside of the product). Each component can be disassembled for redesigning to further improve its functionality, to add features or to refacor for eliminating any internal inefficient design (e.g. internal spaghetti code) and tested (outside of the product) individually before reassembling.
Today even after leveraging latest technologies, tools and all the suitable reusable libraries, software engineers are forced to implement 100s of thousands to millions of lines of custom application code for creating large software products. In case of software products today nearly 90% of the custom application code ends up as spaghetti code. Since essential properties of the physical functional components are not yet known, no conscious effort is made to identify each component. Hence almost no real software component (equivalent to the physical functional components by having the essential properties uniquely and universally shared by every known physical functional components).
An ideal CBD for software product must satisfy the following conditions: (i) Over 97% of the functionality and features of the software product must be implemented as a replaceable components, (ii) where it should require implementing no more that 3 to 5 lines to assemble each of the replaceable components, and (iii) removing 3 to 5 lines implemented to assemble each of the replaceable component must effectively remove the component without leaving any traces of spaghetti code. That is, if the application is created by assembling each of the components into a generic application template, the leftover code must be the generic application template, if all the components are disassembled by removing each of the components one at a time.
For example, the airplanes, automobiles or computers are created by assembling components as hierarchy of replaceable components. Each of the products would be left with empty frame (i.e. skeleton or holder), if all the components are disassembled by removing each of the components one at a time. Each component is designed and can be disassembled to redesign and test individually outside of the product. Each component can be disassembled to replace or redesign individually outside of the product (free from spaghetti code implemented for any other component). The component is reassembled into the product after redesigning and testing the component individually outside of the product (without being forced to see even a single line of spaghetti code implemented for any other component, while redesigning and testing).
Today no conscious effort is made to identify software parts equivalent to the physical functional components to implement them as loosely coupled, so that each can be assembled by implementing 3 to 5 lines and disassembled by removing the 3 to 5 lines (e.g. to redesign and test individually free from exposing to spaghetti code). There is no valid reason, why it is not possible to modularize 97% of the features and functionality (i.e. implemented in the custom code to satisfy product’s unique needs) into such replaceable components, where each component requires no more than 3 to 5 lines to disassemble/reassemble (so that each of the replaceable components can be redesigned and tested free from exposing to the spaghetti code implemented for any other part of the application).
It is impossible to find any explicit explanation, why software engineering paradigm for CBSD (CBD of software products) deviated from the reality of CBD (Component Based Design) of physical products? It is impossible to find any rational explanation or justification anywhere. The first principles for basic sciences are taught in high schools. Likewise, the first principles for new branches of scientific disciplines (e.g. CBSD, virology, mycology, parasitology, and bacteriology) are taught in first couple of chapters of respective text books.
My best guess of the undocumented hidden reasoning: It is not possible to discover accurate description for the physical functional components, for example, where the accurate description is defined by essential characteristics uniquely and universally shared by each and every known physical functional component. It is not possible to invent real software components (having the essential characteristics) equivalent to the physical functional components, even if it is possible to discover such essential characteristics for the physical functional components.
Likewise, it is not possible to discover accurate description for the CBD of the physical products, for example, where the accurate description is defined by essential aspects uniquely and universally shared by each and every known CBD of physical product. It is not possible to invent real CBD for software products (having the essential aspects) equivalent to the CBD of physical products, even if it is possible to discover such essential aspects for the CBD of physical products.
Can any one summarize any other tacit reasoning, why software engineering paradigm for CBSD (CBD of software products) deviated from the reality of CBD (Component Based Design) of physical products?
Although those reasons might be acceptable 50 years ago (when FORTRAN and assembly languages were leading edge technologies), are such tacit assumptions still valid? Documenting the reasons that resulted in deviating software engineering paradigm for CBSD from reality would help address them to put the software engineering paradigm for CBSD on right tracks.
Relevant answer
Answer
All the "research" works I found from you were the 7 articles you published on ResearchGate and here are the facts:
1 - You cite no source (excepted yourself) to support your claims.
Just one example: "The researchers of computer science and software engineering defined CBD (“Component Based Design”) for software is using software parts either having a given set of useful properties (e.g. reuse or standardized) or conform to a given so called component models.". If it is true, then you should cite which papers you considered to ensure that. Was it written explicitly in another paper? Did you interpret that from reading several papers? How much of them? How good are they? etc. No reader can properly check your claims until you properly support them with citations. This is all about research rigor, or intellectual honesty, or repeatability of your discovery (people can redo the same path to ensure that they reach the same conclusion). If you don't want to be rigorous in your writting, why one should be rigorous in reading you? Or said another way, why should we take you seriously if you do not take your readers seriously by giving them the way to ensure that what you say is not pure invention? The only citations I saw were when you cite your blog, but not only self-citation are arguably a support (they ar emere extension of the paper) but a blog has no research power. It is nice to share ideas, but not to assess their validity.
2 - Your papers lack in modesty and rigor.
History of science as shown us that we are correct unless someone proves the opposite. We never know when our assumptions will be broken, and this is only by the inability to break it that it obtains its Truth value. Inability which can reduce along time, due to the evolution of the knowledge, with the possibility to reach a time when someone else finally find the counter-example which break the assumption. Thus we have to write our claims with the necessary modesty to show that we know that it "could be" wrong. But the way you write your papers just completely lack this modesty:
"The erroneous assumption that the Earth was at the center was absolutely wrong (i.e. a lie)."
"The Sun is at the center and the planets going around the sun is an absolute Truth."
"This discovery still holds True, and it is universally accepted as an absolute truth."
"Today no one can deny this absolute Truth."
Who are you to claim what should be universally accepted as an absolute truth? Did you go around the world, asking to any single individual whether or not (s)he believes that claim? The Earth is at the center is wrong? Well, it is at the center of the orbit of the moon, and it is at the center of its own revolution, so it is at the center to some extents. The sun being at the center is an absolute truth? Well, the solar system is clearly not in the center of the galaxy, so now way for the sun to be in the center of the galaxy too, and probably not in the center of the universe, so it is not always in the center. Your "absolute" claims are breakable with a simple perspective switch. If I can show that easily that you are wrong here, why should I consider that you are correct elsewhere?
3 - Your papers lack in respect towards other researchers.
"why brilliant minds of computer science and software engineering unable to see falsity of this seed axiom."
"Most researchers have hard time accepting simple obvious facts and refuse to investigate the Truth"
"Most researchers refuse to accept simple facts about real CBD"
Then why are you even discussing with us, if we are so stupid at the first place? Are you some kind of prophet who comes to save us by delivering the Divine Laws? Why should we respect you if you are not respecting us? Let me rephrase those sentences of yours:
"why Raju unable to see that his behaviour towards the community of researchers is inadapted."
"Raju has hard time accepting that nobody is perfect, including him, and refuse to investigate his own behavioral flaws"
"Raju refuses to accept simple facts about real research"
Do you think you could have a constructive discussion with someone claiming that?
4 - None of your papers have been published in any peer-reviewed context (conference, journal, whatever).
Peer review is the way to assess that your work does not have obvious flaws. If your paper is not published in a peer-reviewed context, then it potentially means that you never tried to validate what you think by having constructive feedback in order to improve your work. You probably just went the way you personally think it should go and finally reached a state where you think it is consistent enough so that it cannot be wrong. Sorry for your ego, but it is not enough to prove absolute Truth. Unless you define absolute Truth by "accepted by Raju Chiluvuri", definition that will obviously be rejected by many researchers today. But if you fix the previous points, probably you have your chance to be accepted for publications.
So here is my conclusion: you are not a researcher. Or said another way, you are a self proclaimed researcher. You are working alone in your cave, building your own conceptual model of the world, and relying only on your own, personal interpretations to claim what is a proper Truth or not, and then you come here to say "I know better than you". This is what spoiled children do. If your ideas are so strong that they deserve the strong claim you write, then you should be able to provide all the necessary evidences, with proper citations, to show that what you say is actually true and can be assessed by doing the same process that you did: checking all these citations, assessing their value and find the same conclusion. If you can't, then what you write is not research, it is an auto-biography. But in this case, you should go to Facebook, not ResearchGate.
Question
5 answers
I am working on the code to trace mouse location on screen using the Kalman filter. But I wanted to know which parameters (with values) I should consider? What values should be taken in the transition matrix? And most important what range of values to consider for the X position of the mouse, the Y position of the mouse, the velocity change along the X axis, and the velocity change along the Y axis (these are rows of transition matrix).
help.
Relevant answer
Answer
This is an interesting problem. I assume you would need a position-velocity-acceleration model which gives you a target tracking model.
[dotx \\ ddot x \\ dddotx] = [1 DT DT^2\2 \\ 0 1 DT \\ 0 0 1] [[x \\ dotx \\ ddotx] + noise
y = [1 0 0] [x \\ dotx \\ ddotx] + measurement noise
I hope you understand the notation (latex).
Question
4 answers
Do all method achieve Pareto set? I have read about the different methods, but I don't understand which is best one.
I ask about the reference  which can assist me to decide which method is best.
constrained, wight, compromise, goal, lexicographic
Relevant answer
Answer
Hi Abdallah,
If you are not sure that an heuristic method achieves the Pareto optimal set, you can apply it on a toy example, where the solutions can be derived analytically. The reference below provides such examples. If you want to know to have an idea the best configuration of your algorithm to approximate the Pareto front of your problem in the case where it cannot be derived analytically (most of the time when you deal of dataset of normal size), then you can try to build a problem with analytic solutions that looks similar.
I hope this late answer could still help you.
Question
3 answers
I was wondering if modern, high-end processors' register files are double-pumped, and how many ports they can have.
Relevant answer
Answer
As stated , "pipelining" I guess should not suffer if registers are multi-port ! (stalling results). Again it depends on clock phases (0,90,180,270,360) if the unit(s) support this or these are designed for a reasonable performance (best / average/worst case) . To get independent atomic operations so that multiple instructions are processed in a clock period there is a need for bargaining ! (2-phase clock circuitry) remember that there should not be any violations of set-up time/hold time etc .
Scared to use these words - "hyper threading"  use the control and general purpose registers for greater parallelism .
Question
5 answers
A tank has i fishes, of size n1, n2, ni.  A person puts an infected fish of size 'x' into the tank. This infected fish can eat a smaller size fish (say, size 's') and grow to x=x+s size. This way it can keep eating any smaller fish and grow in size, but it cannot eat a bigger one.
At any point, if the tank has fishes where all the remaining of ni  fishes are larger than x, then the scientist can make a move to either add a small fish to make x grow larger than the smallest of ni or remove a larger fish so that the smallest of remaining fishes are smaller than x.
What is the MINIMUM number of moves the person needs to make, so that at the end of all iterations only one fish - the infected one - remains?
Relevant answer
Answer
There must be a given relation between x and s in your question.
Question
7 answers
I have accessed the following page to download hadoop-2.6.0 but confused which file I should download. Whats are the use of each.
Relevant answer
Answer
If you have a plan to build the binary by yourself, just download hadoop-2.6.0-src.tar.gz, extract and build. If you use windows and need executable files, download hadoop-2.6.0.tar.gz and extract it. It should contain binary files which can run on windows. *.mds files are checksums.
Question
4 answers
In driver class I have added Mapper, Combiner and Reducer classes and executing on Hadoop 1.2.1. I have added NumReduceTasks as 2. The output of my mapreduce code is generated in a single file part-r-00000. I need to generate output in more than one file. Does it need to be add our own Partitioner class?
Relevant answer
Answer
HashPartitioner is the default Partitioner in hadoop. you may use other partitioner provided by hadoop or create custom partitioner class too as per your requirements.
following link may be useful to you. kindly go through all the content because it will answer almost all your questions related to hadoop map-reduce.
Question
2 answers
Relevant answer
Answer
The terms “dependability”‎ and “security”‎ have been used interchangeably to describe the properties of secure and trusted software.
However, the extant literature shows that dependability attributes are considered as the cure for security threats, abnormal behavior and untrustworthy issues in a software system.
A system is considered dependable when it can be depended on to produce the consequences for which it was designed, with no adverse effect in its intended environment. Dependability comprises several attributes that imply availability, confidentiality, integrity, reliability, safety, and maintainability.
Methods and tools to attain the dependability attributes have been discussed in details in my previous publications.
I hope this can help you. Good luck.
How do I correctly perform a Zero adjustment/Zero offset operation? Comments on signal filtering is also added.
Question
60 answers
Hello,  I'd like to minimise all errors of a force plate signal. This device has got a tare button which I use to push before any new measurement.  However, if I measure during an unloaded condition, of course there is some little shift (drift) and o offset too. The attached figure shows a recording of 5s under no load condition after resetting the force plate (tare button).  Should I take the mean of this "baseline" ( zero) and then add this value to the measure vector under experimental conditions? Is this correct? I'm a bit confused between Zero adjustment and Zero offset.
Relevant answer
Answer
Adjust the baseline of data collection by subtracting the entire curve by averaging the values you collected without charge. Do not worry, this is a procedure widely used for adjustment. There seems to be a bit of noise in the signal, then, remember that at the time will process your data should filter them properly, using a good filter structure and appropriate cut-off frequency. In this link, you find a routine implemented in matlab environment to perform residuals analysis and identify the optimal cutting frequency for your signal.  http://g-biomech.blogspot.com.br/2013/12/analise-residual.html http://g-biomech.blogspot.com.br/2013/12/analise-residual.html
Question
8 answers
I need a tool / software / code in java that can represent the abstract tree in a graphical way i.e. showing labeled nodes connected via labeled edges.
Relevant answer
Answer
One of the most flexibles API's to obtain this is JUNG (http://jung.sourceforge.net). An example of tree at http://jung.sourceforge.net/applet/treecollapse.html
You can also try Graph/JGraphT (http://jgrapht.org)
Question
1 answer
We have a 'datapath' from one CPU, see attachment figure. If the next instruction address is in 'PC Register', how many 'clock cycles 'need to follow 'word add' instruction fetched and executed?
Memory is 10-bits and each instruction at least is 2-words. All register is 10-bits and has 'INC (increment), CLR (clear), LD(load)' instructions. 'Addr' means direct addressing. Onstruction will work on 10-bits word:
Wadd (src1), (src2), (dst)
My instructor solved that it's on a 15 clock cycle. Any hint or idea on how this is calculated?
My try is attached via the attachment figure, but I don't know whether I'm right.
Relevant answer
Answer
Dear Huiman, you really need to ask your instructor. You cannot expect strangers to your course to load their mind with an unfamiliar and incompletely described design. Moreover, the answer completely depends on the microcoding, as well as the abstract cicrcuit scheme. If that scrap of paper is indeed the microcode required, and it is 15 steps long, then 15 clock cycles is what it takes.
If you can think of a microcode that does the job faster, then good for you! Show it to your instructor and let him say if there is something wrong with it.
What I notice is that the microcode on the paper seems to include memory reads ("M[AR]"), so you have been holding out on us .. that is not a "word add", but a "memory word  add". Yes .. the final step in the microcode opens the path from register DR to memory M[AR].
The schema does not distinguish between multiplexers and adders/ALUs, which I find confusing. But I can live with it :-). The best hints are in the microcode! There I see the instruction fetch and decode in the timing. I would not have expected that to share the same circuitry. Certainly it is notionally a different "datapath", so I would not expect the datapath you have shown to be relevant to it. A datapath is a configuration of the multiplexers which allows data to plow in a certain way through the circuit. You should have 15 datapaths, one for each microcode instruction (a microcode instruction is a configuration of the circuit, hence a datapath).
Given that the instruction fetch and decode occupies the same functional elements as the memory data load and store and arithmetic, I would not be surprised to hear that 15 clock cycles are required overall (where is the decode unit in your diagram?). If the instruction fetch and decode were parallel to the rest of the circuit, as is normal, then register add would be expected to take just one cycle. An extra load and store would take one maybe two cycles each - it looks like your clock is slow, so memory access can be one clock. But you also have the instruction fetch, and a separate decode.
I also note that the PC counter is incremented 4 times in that micro-instruction sequence. Why? Surely only one instruction is read? Is the data embedded following the instruction? Ah .. the ISA is shown, and yes, addresses of sources and destination follow the instruction. So that's why there are four advances of PC and four reads of the instruction stream (into the DR register). Then each address needs to be moved from DR to the AR register to be looked up in memory and the result placed in the DR register, and maybe moved somewhere safe. Then the arithmetic is done. Then the address is looked up and the result moved there from the DR register. Yes, that's a lot.
But the bottom line is really to interrogate your instructor.
If that scrap of paper is really your explanation, as you claim, then why are you asking us? The writing looks confident, and I would have said that was an Instructor writing it - since it contains enough information for me to follow, whereas your question above does not.
TYPE out the microcode on the piece of paper and add your own commentary to it. That should persuade you one way or the other. It starts:
1. move PC to AR in order to read the instruction next
2. read instruction at AR into DR, as promised
3. decode DR and advance PC for read of following addresses, move PC to AR
4. read address at AR into DR, as promised, and advance PC for next read
5. move  DR to AR preparatory to reading data from memory at AR
6. read data from memory at AR to DR
7  ...
Keep going! What can you do better?
Question
12 answers
How to cope with available computing power as the machine size increases?
Relevant answer
Answer
Amdahl's law was seen as a pessimistic outlook by the parallel processing community. There is also a law called Gustafson's law which you may want to read about.
Question
5 answers
I need a software tool for training a neural network with the Artificial Bee colony optimization technique or the ACO or the firefly. Can an one help me ?
Relevant answer
Answer
Dear Parminder,
send me an email in order to send you some matlab code. my gmail:
best regards,
Hamid
Question
4 answers
I have an array say A=[1    4   -3   2   4   6   7   8   17    -23    -34   4    5    6
6   7   -5   4    7  41  2   5   23    -3        4    7    7    41 ]
Now I need to get the output as:
[1    4   -3   2      6    7   8   17    -23    -34       5                                                      6    7   -5   4      41  2   5   23    -3        4        7
1    3    1    1     2    1    1    1      1        1        1]
Where the third row gives the count of occurrence of the column (given by first two rows in the output matrix) uniquely.
Relevant answer
Answer
I did not get exactly the question. So what do you mean by "the count of occurrence of the column (given by first two rows in the output matrix)"
So for example what does each element of third row shows? the third column of third row is 1; does it mean the number of occurrence of -3 or -5 is 1? probably i missed understood some part.
Question
38 answers
In the literature and in the RG site pages, the “kinetic model” term sometimes occurs. Using it, authors of laboratory studies would like to let riders know beforehand that they don’t state that this model describes the real mechanism but only propose the model as if capable of predicting the reaction rate under some conditions.
However, on the one hand, it is a question whether the irresponsible proposal of a mechanism is useful for the catalytic science and, on the other hand, maybe, it would be more useful to have an experimental unit real tube filled with a catalyst or an experimental shelf with a catalyst at the chemical plant and to determine the map of the catalyst productivity under different ambient conditions with no models.
Relevant answer
Answer
Hi
Valid questions and comments Prof Victor Ostrovskii. Excellent.
Depending on whether you are a chemist or chemical engineer, the meaning, scope and application of kinetic models can be different and diverse.
There are a number of important points to make initially: reaction kinetics, unlike thermodynamics, have to start in the lab with experimentation under "certain" conditions, then data collected fitted to mathematical models based on "assumed or postulated" mechanisms. Clearly, a number of mechanisms can be imagined and put to test. One can end up with a few possibilities that can be resolved using statistical approaches for the "best fit". These models are then published for other people to use in simulations (predictions). This is the case for "intrinsic " kinetics. But chemical engineers are also interested in mass transfer limitations and therefore add extra terms (like external, internal mass transfer resistance, etc...) will be added to the kinetic models. This is well documented.
For chemical reactor design, these reaction kinetics models are essential and are specific to reaction systems. Such is the real challenge in reactor design: we are at the mercy of the quality of kinetic studies in the lab. To mention an important point: quality reaction kinetics studies are very expensive in terms of equipment and are therefore highly appreciated. For every single kinetic study in the literature, you may find 10's or 100's of design or simulation studies that are cheaper to carry out and can be published quickly.
The purely experimental data have limited use for engineers and can be useful for catalyst development chemists. I can see value in both studies and each will have different "clients".
An emerging field such as molecular dynamics may one day replace lab experimentation if it proves to be reliable and validated independently by experts. That is really the major breakthrough we have been waiting for. To cut costs in expensive lab tests and be coupled with high power design/simulation tools.
We are not there yet. Some efforts being deployed in that direction though.