Re: Differential equations error with MathLink/JLink
- To: mathgroup at smc.vnet.net
- Subject: [mg27903] Re: Differential equations error with MathLink/JLink
- From: Ralph Gauges <ralph.gauges at eml.villa-bosch.de>
- Date: Fri, 23 Mar 2001 04:31:25 -0500 (EST)
- Organization: Tiscali / Nacamar
- References: <98vh54$8sr@smc.vnet.net> <99chhr$8bi@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Todd Gayley wrote:
> On 17 Mar 2001 06:17:56 -0500, Ralph Gauges <ralph.gauges at eml.villa-bosch.de> wrote:
>
> >Hi,
> >
> >I am rather new to mathematica and I have some problems solving
> >differential equations using any mathlink product.
> >First I tried the python module for mathematica and when that didn't
> >work, I tried JLink with the same result.
> >The problem is as follows. Whenever I try to specify a differential
> >equation like
> >
> >x'[t]==2*x[t]+3
> >
> >and want this to be evaluated, I get the error message :
> >
> >MathLinkException occurred: MLGet out of sequence
> >
> >I don't think I made a mistake since I just took the SampleProgram.java
> >from the JLink packet and changed one line:
> >
> > //ml.evaluate("2+2");
> > ml.evaluate("x'[t]==2*x[t]+3");
> >
> >Other stuff works just fine.
> >Has anyone else had these problems before and found a solution?
> >Thank you very much for the help
> >
> >Ralph
>
> Ralph,
>
> The result you will get from ml.evaluate("x'[t]==2*x[t]+3") is exactly what you will get
> if you type this into a Mathematica notebook--the original expression returned unmodified.
> Perhaps you want to use ml.evaluate("DSolve[x'[t]==2*x[t]+3, x[t], t]").
>
> Getting an "MLGet out of sequence" error in a J/Link (or MathLink) program generally means
> that you are trying to read the result in an inappropriate way. You just changed one line
> in SampleProgram.java, but you also need to change the way the result is read. That spot
> in the sample program calls getInteger() to read the result, but obviously an integer is
> not what comes back from your calculation.
>
> If you just want to display the result on the screen, you are better off using
> evaluateToOutputForm() instead of evaluate(), as evaluateToOutputForm() will return the
> result as a formatted string. If you want the result as an expression, so that you can
> operate on it in Java (not likely in this case), then you should use evaluate() and either
> pick apart the result with a series of "get" calls (using getNext() to tell you how to
> read each piece), or you can use getExpr() to read the whole result as an Expr and use the
> Expr methods to decipher its structure (this is the recommended technique).
>
> Remember, though, that it is usually easier to ask Mathematica to return exactly what you
> need rather than pick apart a complicated result on the Java side. In other words,
> something like this:
>
> ml.evaluate("x[t] /. Flatten[DSolve[x'[t]==2*x[t]+3, x[t], t]]");
>
> although this result will still be a symbolic expression:
> Plus[Rational[-3,2],Times[Power[E,Times[2,t]],C[1]]].
>
> --Todd Gayley
> Wolfram Research
Thanks, this getExpr seems to work a lot better than to fiddle around with all the get
methods. Now I will have to see how I get this over to python again since I haven't planned
on switching to java.
I guess trying to reproduce this error in java was not the best idea, but since the python
module is not supported I thought I could try.
In the python module the get Functions are mapped to python functions and I traced the
program flow to the error in the C files. There was a getNext statement which stated that the
next object was a function, but the following getFunction call caused the error described.
Maybe I can rewrite the module to use the getExpr instead of getting the individual parts.
Thanks a lot for the help
Ralph