halvesize matlab,optim_private_nlsq.c 源代码在线查看 - ASUFIT－Matlab－全局拟合程序资源下载虫虫电子下载站...

心已赠人 2023-01-15 15:25 43阅读 0赞

/\* \* if SD'\*GradF>eps, \*/ if (mlfTobool( mlfGt( mlfMtimes(mlfCtranspose(SD), GradF), mlfEps()))) \{ /\* \* how='ERROR- GN not descent direction'; \*/ mlfAssign( &how, mxCreateString( "ERROR- GN not descent direction")); /\* \* end \*/ \} /\* \* temp=0; \*/ mlfAssign(&temp, mlfScalar(0.0)); /\* \* else \*/ \} else \{ mxDestroyArray(a\_); /\* \* if verbosity > 0 \*/ if (mlfTobool( mlfGt(verbosity, mlfScalar(0.0)))) \{ /\* \* disp('Conditioning of Gradient Poor - Switching To LM method') \*/ mlfDisp( mxCreateString( "Conditioning of Gradient Poor" " - Switching To LM method")); /\* \* end \*/ \} /\* \* how='CHG2LM'; \*/ mlfAssign(&how, mxCreateString("CHG2LM")); /\* \* levMarq=1; \*/ mlfAssign(&levMarq, mlfScalar(1.0)); /\* \* currstepsize=abs(currstepsize); \*/ mlfAssign(&currstepsize, mlfAbs(currstepsize)); \} /\* \* end \* end \*/ \} /\* \* \* if (temp) \*/ if (mlfTobool(temp)) \{ /\* \* % Levenberg\_marquardt Method N.B. EstSum is the estimated sum of squares. \* % GradFactor is the value of lambda. \* % Estimated Residual: \* if EstSum>fbest \*/ if (mlfTobool(mlfGt(EstSum, fbest))) \{ /\* \* GradFactor=GradFactor/(1+currstepsize); \*/ mlfAssign( &GradFactor, mlfMrdivide( GradFactor, mlfPlus(mlfScalar(1.0), currstepsize))); /\* \* else \*/ \} else \{ /\* \* GradFactor=GradFactor+(fbest-EstSum)/(currstepsize+eps); \*/ mlfAssign( &GradFactor, mlfPlus( GradFactor, mlfMrdivide( mlfMinus(fbest, EstSum), mlfPlus(currstepsize, mlfEps())))); /\* \* end \*/ \} /\* \* SD=-(JAC'\*JAC+GradFactor\*(eye(numberOfVariables,numberOfVariables)))\\(CostFunction'\*JAC)'; \*/ mlfAssign( &SD, mlfMldivide( mlfUminus( mlfPlus( mlfMtimes(mlfCtranspose(\*JAC), \*JAC), mlfMtimes( GradFactor, mlfEye( numberOfVariables, numberOfVariables)))), mlfCtranspose( mlfMtimes( mlfCtranspose(\*CostFunction), \*JAC)))); /\* \* currstepsize=1; \*/ mlfAssign(&currstepsize, mlfScalar(1.0)); /\* \* estf=JAC\*SD+CostFunction; \*/ mlfAssign( &estf, mlfPlus(mlfMtimes(\*JAC, SD), \*CostFunction)); /\* \* EstSum=estf'\*estf; \*/ mlfAssign( &EstSum, mlfMtimes(mlfCtranspose(estf), estf)); /\* \* if verbosity> 1 \*/ if (mlfTobool(mlfGt(verbosity, mlfScalar(1.0)))) \{ /\* \* num=\[num,sprintf('%12.6g ',GradFactor)\]; \*/ mlfAssign( &num, mlfHorzcat( num, mlfSprintf( NULL, mxCreateString("%12.6g "), GradFactor, NULL), NULL)); /\* \* end \*/ \} /\* \* end \*/ \} /\* \* gdnew=GradF'\*SD; \*/ mlfAssign(&gdnew, mlfMtimes(mlfCtranspose(GradF), SD)); /\* \* \* OLDX=XOUT; \*/ mlfAssign(&OLDX, XOUT); /\* \* % Save Variables \* FIRSTF=NewF; \*/ mlfAssign(&FIRSTF, NewF); /\* \* OLDG=GradF; \*/ mlfAssign(&OLDG, GradF); /\* \* GDOLD=gdnew; \*/ mlfAssign(&GDOLD, gdnew); /\* \* \* % If quadratic interpolation set PCNT \* if lineSearchType==0, \*/ if (mlfTobool(mlfEq(lineSearchType, mlfScalar(0.0)))) \{ /\* \* PCNT=1; MATX=zeros(3,1); MATL(1)=NewF; \*/ mlfAssign(&PCNT, mlfScalar(1.0)); mlfAssign( &MATX, mlfZeros(mlfScalar(3.0), mlfScalar(1.0), NULL)); mlfIndexAssign(&MATL, "(?)", mlfScalar(1.0), NewF); /\* \* end \*/ \} /\* \* else \*/ \} else \{ /\* \* % Halve Step-length \* how='Red\_Step'; \*/ mlfAssign(&how, mxCreateString("Red\_Step")); /\* \* if NewF==FIRSTF, \*/ if (mlfTobool(mlfEq(NewF, FIRSTF))) \{ /\* \* if verbosity>0 \*/ if (mlfTobool(mlfGt(verbosity, mlfScalar(0.0)))) \{ /\* \* disp('No improvement in search direction: Terminating') \*/ mlfDisp( mxCreateString( "No improvement in search " "direction: Terminating")); /\* \* end \*/ \} /\* \* status=1; \*/ mlfAssign(&status, mlfScalar(1.0)); /\* \* EXITFLAG = -1; \*/ mlfAssign(EXITFLAG, mlfScalar(-1.0)); /\* \* else \*/ \} else \{ /\* \* currstepsize=currstepsize/8; \*/ mlfAssign( &currstepsize, mlfMrdivide(currstepsize, mlfScalar(8.0))); /\* \* if currstepsize \*/ if (mlfTobool( mlfLt(currstepsize, mlfScalar(1e-8)))) \{ /\* \* currstepsize=-currstepsize; \*/ mlfAssign( &currstepsize, mlfUminus(currstepsize)); /\* \* end \*/ \} /\* \* end \*/ \} \} \} /\* \* end \*/ \} /\* \* XOUT=OLDX+currstepsize\*SD; \*/ mlfAssign(&XOUT, mlfPlus(OLDX, mlfMtimes(currstepsize, SD))); /\* \* if isinf(verbosity) \*/ if (mlfTobool(mlfIsinf(verbosity))) \{ /\* \* disp(\[num,' ',how\]) \*/ mlfDisp(mlfHorzcat(num, mxCreateString(" "), how, NULL)); /\* \* elseif verbosity>1 \*/ \} else if (mlfTobool(mlfGt(verbosity, mlfScalar(1.0)))) \{ /\* \* disp(num) \*/ mlfDisp(num); /\* \* end \*/ \} /\* \* \* end %----------End of Direction Update------------------- \*/ \} /\* \* if lineSearchType==0, \*/ if (mlfTobool(mlfEq(lineSearchType, mlfScalar(0.0)))) \{ /\* \* PCNT=1; MATX=zeros(3,1); MATL(1)=NewF; \*/ mlfAssign(&PCNT, mlfScalar(1.0)); mlfAssign(&MATX, mlfZeros(mlfScalar(3.0), mlfScalar(1.0), NULL)); mlfIndexAssign(&MATL, "(?)", mlfScalar(1.0), NewF); /\* \* end \*/ \} /\* \* % Check Termination \* if max(abs(SD))< tolX \*/ if (mlfTobool(mlfLt(mlfMax(NULL, mlfAbs(SD), NULL, NULL), tolX))) \{ /\* \* if verbosity > 0 \*/ if (mlfTobool(mlfGt(verbosity, mlfScalar(0.0)))) \{ /\* \* disp('Optimization terminated successfully:') \*/ mlfDisp( mxCreateString("Optimization terminated successfully:")); /\* \* disp(' Search direction less than tolX') \*/ mlfDisp(mxCreateString(" Search direction less than tolX")); /\* \* end \*/ \} /\* \* status=1; EXITFLAG=1; \*/ mlfAssign(&status, mlfScalar(1.0)); mlfAssign(EXITFLAG, mlfScalar(1.0)); /\* \* elseif (GradF'\*SD) < tolFun & ... \*/