Activity for Eberhard Schruefer
-
Eberhard Schruefer posted a comment on discussion Developers
I commited a fix for the problem with gentran you reported. Your example should work now after rebuilding. Thanks for letting us know. Eberhard
-
Gentran: Strip procedure info generated by formproc. The info part was introduced after the creation of the gentran package and is of no use here.
-
Hello, For some reason string manipulation comes up here once in a while. It is not really something that is used in Reduce. What are you trying to accomplish? Anyway, to answer your question the following should do symbolic procedure explode_string u; 'list . cdr reverse cdr reverse explode u; symbolic operator explode_string; 15: explode_string "Hello world"; {H, e, l, l, o, , w, o, r, l, d} 16: Eberhard
-
Excalc: An improvement of commit 6606. When the switch 'nero' is on no extra spaces should be generated anymore.
-
Excalc: Correct printing indexed expressions when the switch 'nero' is on.
-
Hi, Thanks for reporting the issue. The message only appears with fancy printing, for example when the graphical interface is used. If you put the following after loading excalc symbolic procedure indxpri(v,u); begin scalar x,y,z; y := flatindxl allindk v; for each j in mkaindxc(y,if coposp cdr v then get(car v,'indxsymmetries) else nil) do <<x := pair(y,j); z := aeval subla(x,v); if null(!*nero and (z = 0)) then <<maprin list('setq,subla(x,v),z); if not !*nat then prin2!* "$"; terpri!* t>> else...
-
After a quite big struggle with our old snapshot-builder I managed to create a snapshot that might work on older Mac's. Please try it out. The snapshot is now on sourceforge and is named Reduce-legacy_snapshot_6550.dmg Eberhard
-
Maybe this will help you with your original problem. Remeber the definition of a quotient and remainder. p = a*q + r Now let E be p and (J - 8) be q, then r can be interpreted as a 'substitution'. In Reduce simply calculate remainder(num !E, !J - 8); Eberhard
-
Remigiusz, Without more information it is hard for me to respond. You say 'obtained spin connection doesnt follow taylor expansion'. What does that mean? Have you considered using the AdS coframe and putting the perturbation into the metric instead of having the coframe forms perturbed and the metric fixed? It might be more appropriate depending on your goal. Concerning the calculation of the Riemann connection, it is just what you find in standard textbooks (i.e. via calculating the Ricci rotation...
-
snapshots: add --old-args to RSYNC_OPTIONS for newer rsync versions to handle things like *.{deb,rpm,tgs,bz2} the way we want.
-
Besides Arthur's suggestion, here are two other possibilities: 1: p := a*x^2 + b*y^2 + c*x*y + d*x + e*y + f; 2 2 p := a*x + b*y + c*x*y + d*x + e*y + f 2: for k := 0:2 do for l := 0:2 do write x^k,",",y^l,": ", coeffn(coeffn(p,x,k),y,l); 1,1: f 1,y: e 2 1,y : b x,1: d x,y: c 2 x,y : 0 2 x ,1: a 2 x ,y: 0 2 2 x ,y : 0 6: There is also the contributed package coeff2 you could use: 1: load coeff2; 2: p := a*x^2 + b*y^2 + c*x*y + d*x + e*y + f; 2 2 p := a*x + b*y + c*x*y + d*x + e*y + f 3: coeff2(p,...
-
When you define a coframe it is assumed that all your calculations are meant to be carried out in this coframe. That's why Excalc expresses everything with respect to it. So I'm not quite sure what you are after. Anyway, here is a very hackish way to have the substitution of your post carried out 7: lisp(coord!* := nil); 8: sub(e(0)=f* d t, omr(0,1)); d t*f*lamb*r ----------------------------------- - 3 2 sqrt(-------------)*(lamb*r - 3) 2 lamb*r - 3 Note that after issuing the command in line 7...
-
I would suggest that we make this discussion off the list for a while as we have to go into more details. Eberhard
-
Following up on my last message, if you only need the duality between a coframe and a frame you could go like this 2: coframe e 1, e 2; 3: frame x; 4: x(-1) _| e 1; 1 5: x(-1) _| e 2; 0 6: d e 1; 1 d e Eberhard
-
Hi Gideon, In both, eds and excalc, frames are tied to a coordinate systems. But nevertheless you can do this sort of your calculation (if I understand your post correctly). It will be more elaborate, though. E.g. 2: tvector x k; 3: pform f = 0,e k = 1; 4: x(-3) _| d f; x _| d f 3 5: x(-3) _| e 3; 3 x _| e 3 6: let x(-3) _| e 3 = 1; 7: x(-3) _| e 3; 1 But I would really need more background and a code snippet of what you would like to accomplish to see if it can be done. Eberhard
-
Multiple definitions cleanup: two forgotten files in previous commit.
-
Multiple definitions cleanup: Another tiny step mainly related to excalc siblings.
-
Multiple definitions cleanup: excalc and related packages.
-
There is nothing special to do. Just start Reduce and then type the following into it load excalc; coframe e th = d th, e ph = sin th*d ph; killing_vector u; That's it. Eberhard
-
The coframe statement gathers the coordinates of the manifold from the differentials of the given coframe forms. Therefore e(4) = (A + d w) cannot work. You could try to use an explicit expression for A i.e. somsething like e(4) = axd x + ayd y + azd z + (aw + 1)d w Eberhard
-
Hi Oscar, There is an example in the Excalc test file for calculating the PDE's for the Killing vectors of the 2-sphere: 2: coframe e th = d th,e ph = sin th*d ph; 3: killing_vector u; ph th {@ *u + @ *u , ph th th {@ (u ), th ph 2 th @ (u )*sin(th) + @ (u ), th ph th ph cos(th)*u + @ (u )*sin(th)}} ph The first entry of the returned list gives the explicit form of the Killing vector with undetermined coefficients and the rest are the PDE's for its coefficients. Eberhard
-
Web interface experiment: Use equation environment for multiline output.
-
MathLive web-interface: Fix for some issues with cut and paste of the multiplication operator. Make df, solve, ws and lists available in the interface.
-
More work on a MathLive based web interface. It is very crude and mainly meant as a playground to get ideas from. It can be tried on www.ca-musings.de . The rendering is done by mathlive and not by mathjax. On the left of the input field is a button for a virtual keyboard. It is populated only by a few items with no proper keycaps. It serves only as an example and not for real use. Positioning of the cursor is at the moment sometimes tricky. Comments and volunteers are, as always, very welcome.
-
Proof of concept for a MathView based web-interface. It shows that the input equation editor is functional. Cut and paste of 2-dimensional expressions work. Currently, only polynomials and sqrt's are supported as no other serializers are written yet. Unfortunately, to try this out a reduce web image has to be build with the following modification to csl.cpp
-
A playground for a Web interface using MathLive. The script launch_MathLiveTests.sh starts the interface. It is very primitive. In the aera above the submit button a Reduce command can be given, e.g. b-a; . After pressing the submit button Reduce's answer, which should be -a + b, will be displayed in the area below. For more complicated expressions an outputhandler and a latex parser on the Reduce site is needed, unless we decide that the conversion better lives in the interface code.
-
Excalc: this extends vardf such that the Lagrangian can also contain Hodege-terms of exterior degree 0.
-
Yes, 'conc' and 'join' are synonyms in the for each statement.
-
Everything that is not 'nil' or equivalently '{}' is considered as 't' (true). Your example with the function 'divable' cannot work. The function extractlist as it is coded in the contributed package assist expects a function of 1 argument.
-
There are of course many other possibilities, e.g. procedure extract_divable(l,q); for each p in l conc if remainder(p,q) = 0 then {p} else {}; This would give you 2: extract_divable({x,x,y,2},x); {x,x} Eberhard
-
As 'extractlist' is programmed in the package assist you need to return nil as the boolean value for false, i.e. you should use procedure ismultx(p); if remainder(p,x) = 0 then 1 else nil; Eberhard
-
Wikipedia says ([https://en.wikipedia.org/wiki/Exponential_integral]) for positive values of x, we have − E 1 ( x ) = Ei ( − x ) Eberhard
-
Sync revision number for snapshot.
-
Just a few remarks from my site. Using web technologies for a GUI for REDUCE is a very attractive project. We need to distinguish between using a browser just for display and talking to REDUCE via some sort of sockets versus embedding REDUCE into a browser. Roughly speaking the first route is taken by Francis project Run-Reduce-FX. Whereas the second is more or less along using jlisp/jslisp and, what you now propose, webassembly. There are pros and cons for either way like portability w.r.t. the...
-
Using Arthur's procedure, wouldn't just procedure piecewisevalue(e,v); sub(x=v, piecewiseeval(e, v)); procedure piecewiseeval(e, x); if e = {} or first first e = infinity then 0 else if x < first first e then second first e else piecewiseeval(rest e, x); do what you want? E.g. 48: pwc := {{0,0},{1,x**2/2},{2,( - 2*x**2 + 6*x - 3)/2},{3,(x**2 - 6*x + 9)/2},{infinity,0}}; pwc := {{0,0}, 2 {1,0.5*x }, 2 {2, - x + 3*x - 1.5}, 2 {3,0.5*x - 3*x + 4.5}, {infinity,0}} 49: 49: on rounded; 50: piecewisevalue(pwc,1.99);...
-
I just commited a fix that cures the problem to sourceforge. It will be contained in the next snapshot. Eberhard
-
Reform sq's in tmprint and fmprint so that an order command has an effect, e.g. order x,t; f(x-t) is printing now as expected.
-
Thanks. Now I see. The problem occurs when using the GUI. It is o.k. with the CLI as I had shown. We will investigate the issue. Eberhard
-
I can't reproduce what you are reporting. I get 1: operator f; 2: f(x-t); f( - t + x) 3: order x,t; 4: f(x-t); f(x - t) Waht is your version of REDUCE? Eberhard
-
I can only guess. Could it be that you define the above 'for all' statement in your module without specifying to be in algebraic mode? Beware that the module statement puts REDUCE into symbolic mode. So please try again with an 'algebraic;" statement before your 'for all' statement. It should work. Eberhard
-
Fix for a regression with printing matrices while the redfront output handler is active.
-
rlisp88: Hyphens need to be escaped in the test file.
-
Towards the use of the TeXmacs interface under Windows.
-
Since with the signature clause a name for a metric is not provided it can of course not be printed. If you would like to do so you need to specify it i.e. instead of signature use coframe e t = d t, e phi = r*d phi with metric g = -e t*e t+e phi*e phi; g(-a,-b); You can of course use any other name like eta which is also very common for the metric. Eberhard
-
The signature clause in the coframe statement means that the metric is diagonal with the 1's or -1's as the entries in the diagonal. You can define a coframe with any sensible metric. E.g. coframe e t = d t, e ph = sin th *d ph with metric g = - e t*e t + r**2 * e ph*e ph; and print the metric coefficients by 3: g(-a,-b); g := -1 t t g := 0 t ph 2 g := r ph ph You will get an error message when the metric is for example singular i.e. has a different dimension than the coframe basis forms. Does this...
-
If you define a coframe together with a signature the metric is implicitly defined. E.g. given the coframe coframe e t = d t, e phi = r*d phi with signature -1,1; the metric is the symmetric rank 2 tensor - e t ⊗ e t + e phi ⊗ e phi I do not quite understand what you mean by calculating the metric. Eberhard
-
Here is the link https://sourceforge.net/p/reduce-algebra/code/HEAD/tree/trunk/generic/texmacs Eberhard
-
Please take a look into the generic/texmacs directory on sourceforge. It is a while I created this and don't know if it still works. Eberhard
-
Thanks again for pointing this out. If you paste the following into your session after 'load_package excalc;' symbolic procedure assgn_indxpri u; <<if caddr u memq '(first only) then if !*fancy then fancy!-terpri!* t else terpri!* t; if null cadr u and null atom car u and null flagp(caar u,'indexvar) then indxpri1 car u else indxpri(if null cadr u then car u else caadr u, car u)>>; does it fix the problem? Eberhard
-
Excalc: correct a problem when printing indexed quantities while using the gui (here a fancy-terpri* instead of terpri* has to be used).
-
Thanks for checking. It will be included in the next snapshot. Eberhard
-
rlfi: add support for printing binomial. rprint: adapt code for printing assignments.
-
Actually, rlfi doesn't know about binomial. If you read the appended code into your terminal and then use rlfi does this work for you? Eberhard put('tex,'evfn,'texevfn); symbolic procedure texevfn(u,v); reval1(cadr u,v); symbolic procedure binomprla(u,p); begin prin2la "\binom"; prin2la "{"; maprintla(car u,0); prin2la "}{"; maprintla(cadr u,0); prin2la "}" end; put('binomial,'laprifn,'binomprla);
-
Excalc: Issue a terpri* before printing indexed quantities for better readability.
-
Hi Daniel, I just commited changes of the print system to fix the display problems you reported. They will be incorporated in the next snapshot. Eberhard
-
Excalc: Fixup printing of indexed quantities for various frontends.
-
Hi Cristóbal, Thanks for the report. Indeed there is a necessary resimplification step missing when in a coframe statement variables are used that have an assigned value. I commited a correction to the coframe statement right now. The fix will be included in next snapshot. Eberhard
-
Excalc: Make sure that all substitutions are made while setting up a coframe.
-
Hi Daniel. Thanks for reporting this. I can confirm the problem and will look into the issue in the next few days. Eberhard
-
Hi Daniel, This is caused by an erroneous declaration of the variable metricu in the tri package. I have uploaded a correction to sourceforge for this. It will be fixed in the next snapshot. If you need it earlier you would have to read in or better compile tri.red with this correction your self. It will probably work as it is if you use the PSL based version of Reduce. Eberhard
-
Package tri: Change type of variable metricu to fluid in accordance to its origin in the package excalc. This prevented the simultaneous use of both packages under CSL.
-
Laplace package: Ensure that only unique kernels are made. subla on a s.f. may produce non unique kernels and can therefore not be used.
-
Assist package: correct a few places where non unique kernels were injected.
-
Just a very brief answer. As for a documentation of the inner workings of Reduce, there is [https://reduce-algebra.sourceforge.io/lisp-docs/insidereduce.pdf]. If you have questions please feel free to ask them on the developers lists. We are listening. Eberhard
-
With 'off mcd' there are no denominators (the denominators are represented by factors with a negativ power) and therefore rationaliziation does not kick in (and probably even more confusing , surds can internally now be represented as extended kernels which the rationaization code does not see). I would not recommend using 'off mcd' globally, only in places where it is absolutely neccessary. The off mcd mode could certainly be improved. However, it would mean substantial work. I guess it is not used...
-
Lets putting the display issue aside for the moment. if I understand you correctly you want 4/sqrt(2) be 'simplified' and that probably means for you having no sqrt in the denominator. This can be achieved by 'rationalization'. In Reduce you can say 1: on rationalize; 2: 4/sqrt 2; 2*sqrt(2) Eberhard
-
Changed the package rlfi to use an outputhandler. This allows now to also turn on Latex output from within a procedure. It was necessary to make an adjustment to the outputhandler mechanism itself. For the assgnpri slot now all three arguments to assgnpri are passed on (as a list).
-
I'm not sure, but it looks to me that you are after a numeric solution and not for an exact algebraic one. If that is the case there is the numeric package 1: load numeric; 2: s := ((e)**(0.0000000001471 * x) * (-5.9459)+27.9459 * (e)**(0.259 * x))/((e)**(0.259 * x)); (1471*x)/10000000000000 (259*x)/1000 - 59459*e + 279459*e s := ---------------------------------------------------------- (259*x)/1000 10000*e 3: num_solve(s-27,x=0,accuracy=6); {x=7.09776189657} Eberhard
-
Until the sum-operator is rewritten to handle quotients properly, you could try to use 'off mcd'. With mcd turned off the denominators are represented as factors with negative exponents and that should give you the simplification you are after. Eberhard
-
Did you look at the listvecops module? With this your example would go like 1: load listvecops; 2: listproc linreg ll; begin scalar lenx; lenx := LENGTH ll; return lenx; end; linreg 11: linreg({1,2,3}); 3 Eberhard
-
Finer control of preprocessing in the new Bezout resultant code.
-
Name changes in the new Bezout resultant code for interoperability with PSL. Improved exterior multiplication by fusing addition into the multiplication and use of a finger into the result to reduce the depth of traversals.
-
Correction of a rule for the Lerch_phi function.
-
Reworked implementation of the Bezout resultant. It uses the logcount function for parity calculation. This function is currently only available in CSL. The code is written being very light on stack use. It is completely untested, and not yet hooked into the general resultant code. The commit is mainly for giving it a safe place.
-
The error message indicates that one of the (abs mu) kernels have not been made unique and therefore are not recognized as equal in the gcd calculation. Infact there is a place in the procedure radf where a non unique abs kernel is produced. If one changes the relevant line in radf to (if !*precise and evenp n then reval {'abs,mvar u} the problem goes away. Eberhard
-
Excalc: Fix for a problem with defining a coframe where the coordinates are indexed variables.
-
I assume you are using the CSL based version of Reduce. It looks like that you don't have write permissions to the Reduce image file on your machine. Eberhard
-
Are you only downloading the jlisp directory? Please checkout the complete trunk directory and try again. Eberhard
-
Are you using openjdk or java from oracle? My successful compilations were made with openjdk. Eberhard
-
Your video shows that you are using java 12. On a linux machine I had no problem to build the minireduce.jar or the reduce.jar with this version. So it looks like that your java installation on MacOS causes the problem. Unfortunately, I have no access to a MacOS machine. Eberhard
-
Excalc: Add the killing_vector cammand to the documentation.
-
Excalc: provide an example for Killing vectors in the test file. Correct a typo.
-
Excalc: Add a command to calculate the pde's that determine Killing vectors.
-
Just a guess. Could it be that you don't have permissions to write to the reduce image of CSL? Try to compile with redcsl -i- -o filename where filename should be a file you have permissions to write to. After compilation call redcsl with the same options. Eberhard
-
Well, you could define a coframe for your space-time in Excalc and calculate the Lie-derivative of the metric (you would need to do this by calculating the Lie-derivative of the basis of the coframe). This would give you differential equations for the components of the Killing vector which you need to solve some how. Eberhard
-
Dear REDUCE community, We are currently discussing if there is still a need for 32bit snapshots. It would make our life much easier if we could drop this architecture. It would of course still be possible to build one's own 32bit binaries from source provided at sourceforge. However, we will not be test building these versions as much. Please speak up if you have arguments for keeping 32bit binary snapshots. Thanks. Eberhard
-
Excalc: variation of a wedge product with respect to a 0-form was not handled correctly.
-
Excalc: for innerproducts do not assume that when the first argument is built up from natural vectors that also the second argument is expressed in the corresponding coordinates.
-
There must be something missing. You must have assigned something to f2 otherwise f2 would show up in the result you are looking for. Eberhard
-
Thanks for the report. You are of course right that the result shouldn't be zero. Way, way back I made the assumption that when you are using 'natural' vectors, i.e. vectors of the form '@ x', then every thing you will be manipulating is explicitly expressed in these coordiantes. I see that this assumption is not a good one. If you would have expressed your vector vr by tvector x1,x2,x3; vr := v1 * x1 + v2 * x2 + v3 * x3; then you would have gotten the result you expected 29: vr |_ f2; d(x1 _| f2)*v1...
-
Added an error message to protect for improper use of define_spaces in the cantens package.
-
Make the function noncomp1 work with matrices correctly. Minor changes to the excalc documentation.
-
There is currently no support for the musical isomorphisms in EXCALC. If you have a list of features you would like to see implemented, please send them to me (maybe privately) so they can be considere for future extensions. Thanks. Eberhard
-
You could do the following 1: load excalc; *** ^ redefined 2: pform {u,v,x,y,z}=0; 3: fdomain u=u(x,y,z),v=v(x,y,z); 4: d u; d x*@ u + d y*@ u + d z*@ u x y z 5: d v; d x*@ v + d y*@ v + d z*@ v x y z There is also some code for pullbacks in the EDS package. Eberhard
-
Thanks for the ouputs. They show that the calculation is done o.k.. That leaves just the printing code as the culprit. I could pin down where the problem is. The bad behaviour only shows up when the 'fancy-print' stuff is involved. If you hand the '-w' option to the CSL based Reduce the result gets shown as expected. I will fix it in the next few days. Eberhard
-
To narrow it down could you please (after you run the code) issue the following commands in your CSL Reduce lisp indxl!*; prop 'om; and let me know what is displayed?
-
Could you please tell me what you get after you run the program and then execute the following two commands? tr indvarprt; om(k, -l); Additionally, is it working when you use the PSL-based Reduce on your machine? Eberhard
-
Hm, thats weird. I get on my machine 1: load_package "excalc"$ *** ^ redefined 2: coframe e theta = r * d theta, e phi = r * sin(theta) * d phi$ 3: riemannconx om$ 4: write "Display the connection forms"; Display the connection forms 5: om(k,-l); theta ns := 0 theta phi phi cos(theta)*e ns := ----------------- theta sin(theta)*r phi theta - cos(theta)*e ns := -------------------- phi sin(theta)*r phi ns := 0 phi Eberhard
-
Thanks for reporting this. It is another case of not updated documentation. The example code should read load_package "excalc"$ coframe e t=d t,e x=d x,e y=d y with signature -1,1,1; pform f(k,l)=0$ antisymmetric f$ f(k,l) := 0$ f(-t,-x):=ex$ f(-x,-y):=b$ f(-t,-y):=0$ on nero$ f(k,-l); Apologies. The documentation will be adjusted in the next snapshot. Eberhard
-
Daniel, What do you mean by 'the connection itself is not displayed'? Is the result different from what is shown in the documentation? I don't see any difference. Eberhard
1 >