Download the full version of the ebook at ebookname.

com

Computational Optimal Control Tools and Practice

1st Edition Dr Subchan Subchan

https://ebookname.com/product/computational-optimal-control-
tools-and-practice-1st-edition-dr-subchan-subchan/

OR CLICK BUTTON

DOWNLOAD EBOOK

Download more ebook instantly today at https://ebookname.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Nonlinear optimal control theory 1st Edition Leonard David

Berkovitz

https://ebookname.com/product/nonlinear-optimal-control-theory-1st-
edition-leonard-david-berkovitz/

ebookname.com

Optimal Traffic Control Urban Intersections 1st Edition

Slobodan Guberinic (Author)

https://ebookname.com/product/optimal-traffic-control-urban-
intersections-1st-edition-slobodan-guberinic-author/

ebookname.com

Batch Distillation Simulation Optimal Design and Control

2nd Edition Urmila Diwekar

https://ebookname.com/product/batch-distillation-simulation-optimal-
design-and-control-2nd-edition-urmila-diwekar/

ebookname.com

Lubrication Fundamentals Second Edition Dekker Mechanical

Engineering A.A. Wessol

https://ebookname.com/product/lubrication-fundamentals-second-edition-
dekker-mechanical-engineering-a-a-wessol/

ebookname.com
Ruins Revolution and Manifest Destiny John Lloyd Stephens
Creates the Maya First Edition William E. Lenz

https://ebookname.com/product/ruins-revolution-and-manifest-destiny-
john-lloyd-stephens-creates-the-maya-first-edition-william-e-lenz/

ebookname.com

McMillan on Options Second Edition Mcmillan L.G.

https://ebookname.com/product/mcmillan-on-options-second-edition-
mcmillan-l-g/

ebookname.com

Make It Jane Bull

https://ebookname.com/product/make-it-jane-bull/

ebookname.com

Reproducing the French Race Immigration Intimacy and

Embodiment in the Early Twentieth Century Elisa Camiscioli

https://ebookname.com/product/reproducing-the-french-race-immigration-
intimacy-and-embodiment-in-the-early-twentieth-century-elisa-
camiscioli/
ebookname.com

Fighting Like a Guerrilla The Indian Army and

Counterinsurgency War and International Politics in South
Asia 1st Edition Rajesh Rajagopalan
https://ebookname.com/product/fighting-like-a-guerrilla-the-indian-
army-and-counterinsurgency-war-and-international-politics-in-south-
asia-1st-edition-rajesh-rajagopalan/
ebookname.com
Linear and projective representations of symmetric groups
1 Reissue Edition Alexander Kleshchev

https://ebookname.com/product/linear-and-projective-representations-
of-symmetric-groups-1-reissue-edition-alexander-kleshchev/

ebookname.com
Computational Optimal Control
Computational Optimal
Control
Tools and Practice
Subchan Subchan
Cranfield University at Shrivenham, UK

Rafał Żbikowski
Cranfield University at Shrivenham, UK

A John Wiley and Sons, Ltd, Publication

This edition first published 2009


c 2009 John Wiley & Sons Ltd

Registered office
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ,
United Kingdom.

For details of our global editorial offices, for customer services and for information about how to apply
for permission to reuse the copyright material in this book please see our website at www.wiley.com.

The right of the author to be identified as the author of this work has been asserted in accordance with
the Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior
permission of the publisher.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print
may not be available in electronic books.

Designations used by companies to distinguish their products are often claimed as trademarks.
All brand names and product names used in this book are trade names, service marks, trademarks or
registered trademarks of their respective owners. The publisher is not associated with any product or
vendor mentioned in this book. This publication is designed to provide accurate and authoritative
information in regard to the subject matter covered. It is sold on the understanding that the publisher is
not engaged in rendering professional services. If professional advice or other expert assistance is
required, the services of a competent professional should be sought.

Library of Congress Cataloging-in-Publication Data

Subchan, S.
Computational optimal control : tools and practice / S. Subchan, R. Zbikowski.
p. cm.
Includes bibliographical references and index.
ISBN 978-0-470-71440-9 (cloth)
1. Control theory. 2. Adaptive control systems. 3. Mathematical optimization. I. Zbikowski, R. (Rafal)
II. Title.
TJ213.S789 2013
629.8’312–dc22 2009019927

A catalogue record for this book is available from the British Library.

ISBN 978-0-470-71440-9 (Hbk)

Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire.
Contents

Preface ix

Acknowledgements xv

Nomenclature xvii

1 Introduction 1
1.1 Historical Context of Computational Optimal Control . . . . . . . . . . . . . 2
1.2 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Outline of the Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 Optimal Control: Outline of the Theory and Computation 9

2.1 Optimisation: From Finite to Infinite Dimension . . . . . . . . . . . . . . . . 9
2.1.1 Finite Dimension: Single Variable . . . . . . . . . . . . . . . . . . . 10
2.1.2 Finite Dimension: Two or More Variables . . . . . . . . . . . . . . . 15
2.1.3 Infinite Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 The Optimal Control Problem . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3 Variational Approach to Problem Solution . . . . . . . . . . . . . . . . . . . 29
2.3.1 Control Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3.2 Mixed State–Control Inequality Constraints . . . . . . . . . . . . . . 33
2.3.3 State Inequality Constraints . . . . . . . . . . . . . . . . . . . . . . 33
2.4 Nonlinear Programming Approach to Solution . . . . . . . . . . . . . . . . . 34
2.5 Numerical Solution of the Optimal Control Problem . . . . . . . . . . . . . . 36
2.5.1 Direct Method Approach . . . . . . . . . . . . . . . . . . . . . . . . 36
2.5.2 Indirect Method Approach . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

3 Minimum Altitude Formulation 49

3.1 Minimum Altitude Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2 Qualitative Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2.1 First Arc (Level Flight): Minimum Altitude Flight t0
t
t1 . . . . 50
3.2.2 Second Arc: Climbing . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.2.3 Third Arc: Diving (t3
t
tf ) . . . . . . . . . . . . . . . . . . . . . 64
3.3 Mathematical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.3.1 The Problem with the Thrust Constraint Only . . . . . . . . . . . . . 64
vi CONTENTS
3.3.2 Optimal Control with Path Constraints . . . . . . . . . . . . . . . . . 66
3.3.3 First Arc: Minimum Altitude Flight . . . . . . . . . . . . . . . . . . 67
3.3.4 Second Arc: Climbing . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3.5 Third Arc: Diving . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.4 Indirect Method Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.4.1 Co-state Approximation . . . . . . . . . . . . . . . . . . . . . . . . 71
3.4.2 Switching and Jump Conditions . . . . . . . . . . . . . . . . . . . . 74
3.4.3 Numerical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.5 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.5.1 Comments on Switching Structure . . . . . . . . . . . . . . . . . . . 85

4 Minimum Time Formulation 95

4.1 Minimum Time Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2 Qualitative Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2.1 First Arc (Level Flight): Minimum Altitude Flight . . . . . . . . . . 96
4.2.2 Second Arc (Climbing) . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.2.3 Third Arc (Diving) . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.3 Mathematical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.3.1 The Problem with the Thrust Constraint Only . . . . . . . . . . . . . 101
4.3.2 Optimal Control with Path Constraints . . . . . . . . . . . . . . . . . 103
4.3.3 First Arc: Minimum Altitude Flight . . . . . . . . . . . . . . . . . . 104
4.3.4 Second Arc: Climbing . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.3.5 Third Arc: Diving . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.4 Indirect Method Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.5 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.5.1 Comments on Switching Structure . . . . . . . . . . . . . . . . . . . 113

5 Software Implementation 119

5.1 DIRCOL implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.1.1 User.f . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.1.2 Input File DATLIM . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.1.3 Input File DATDIM . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.1.4 Grid Refinement and Maximum Dimensions in DIRCOL . . . . . . . 132
5.2 NUDOCCCS Implementation . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.2.1 Main Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.2.2 Subroutine MINFKT . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.2.3 Subroutine INTEGRAL . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.2.4 Subroutine DGLSYS . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.2.5 Subroutine ANFANGSW . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.2.6 Subroutine RANDBED . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.2.7 Subroutine NEBENBED . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.2.8 Subroutine CONBOXES . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.3 GESOP (PROMIS/SOCS) Implementation . . . . . . . . . . . . . . . . . . . 139
5.3.1 Dynamic Equations, Subroutine fmrhs.f . . . . . . . . . . . . . . 141
5.3.2 Boundary Conditions, Subroutine fmbcon.f . . . . . . . . . . . . 143
5.3.3 Constraints, Subroutine fmpcon.f . . . . . . . . . . . . . . . . . . 144
CONTENTS vii
5.3.4 Objective Function, Subroutine fmpcst.f . . . . . . . . . . . . . . 145
5.4 BNDSCO Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.4.1 Possible Sources of Error . . . . . . . . . . . . . . . . . . . . . . . . 146
5.4.2 BNDSCO Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.5 User Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

6 Conclusions and Recommendations 159

6.1 Three-stage Manual Hybrid Approach . . . . . . . . . . . . . . . . . . . . . 159
6.2 Generating an Initial Guess: Homotopy . . . . . . . . . . . . . . . . . . . . 160
6.3 Pure State Constraint and Multi-objective Formulation . . . . . . . . . . . . 161
6.4 Final Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

Appendix BNDSCO Benchmark Example 165

A.1 Analytic Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
A.1.1 Unconstrained or Free Arc (l  1/4) . . . . . . . . . . . . . . . . . . 167
A.1.2 Touch Point Case (1/6
l
1/4) . . . . . . . . . . . . . . . . . . . 168
A.1.3 Constrained Arc Case (0
l
1/6) . . . . . . . . . . . . . . . . . . 169

Bibliography 173

Index 179
Preface

Computational Optimal Control Reliable and efficient ways of finding the best possible
(optimal) solutions is a pervasive problem in science and engineering. In many such pro-
blems, especially in engineering, we can manipulate the optimised object/process through
limited (constrained) influences (control) at our disposal which we can vary over time
(dynamically). The dynamic control elicits dynamic responses from the optimised object/
process, often in ways which are difficult to guess or intuit. A judicious choice of optimal
control, obeying the constraints, must therefore be based on a systematic procedure. The
well-documented theory of optimal control is exactly the mathematical tool necessary for
that. Given a mathematical description of the optimised object/process, an optimisation cri-
terion (performance index) and constraints, the theory of optimal control gives mathematical
equations whose solution is the optimal control needed. In most realistic (and thus practical)
engineering problems, application of the theory leads to complex equations.
The complexity of the optimal control equations is not a flaw—it properly reflects the
mathematical details of the optimised problems; it is the details that make the problem real-
istic and hence practical. But the complexity of the optimal control equations means that, in
industrial practice, the theory must be accompanied by calculations with digital computers,
especially for advanced problems in aerospace and aeronautics. This blend of mathemati-
cal theory and numerical techniques is the essence of computational optimal control. Both
the theory and the numerical algorithms involved are rather non-trivial in nature, and their
interaction adds another layer of complexity.

Book Focus and Prerequisites This book focuses on informed use of computational op-
timal control rather than development of either theory or numerics. The aim of the book
is to provide a hitherto unavailable computational optimal control self-study textbook for
practising engineers, especially engineers working on challenging, real-world applications
in aerospace and aeronautical industries. Graduate and post graduate students who want to
specialise in advanced applications of optimal control should also find it of interest. The pre-
requisite knowledge is a general background in numerical analysis and ordinary differential
equations plus familiarity with the FORTRAN computer language, usually acquired during
graduate engineering studies. Some knowledge of optimal control would be helpful, but is
not essential—the relevant theory can be picked up while studying this text.

Case Study The main thrust of the book is to explain how to use computational optimal
control tools in engineering practice, employing an advanced aeronautical case study to pro-
vide a realistic setting for both theory and computation. The case study is focused on missile
x PREFACE
guidance in the form of trajectory shaping of a generic cruise missile attacking a fixed target
which must be struck from above. The problem is reinterpreted using optimal control theory
resulting in two formulations: (1) minimum time-integrated altitude and (2) minimum flight
time. The resulting trajectory has a characteristic shape and hence the problem is also known
as optimisation of the bunt manoeuvre. This eminently realistic and practical problem is quite
hard, because realistic missile flight dynamics and practical control and flight path constraints
are assumed. Due to its challenging nature, the bunt manoeuvre problem is an excellent illus-
tration of advanced engineering practice, without comforting simplifications found in many
textbooks. More importantly, the problem strongly exercises both the theoretical and numer-
ical aspects of computational optimal control, so it robustly tests the true value of theoretical
and software tools available to the practitioner. A detailed account of the actual performance
of these tools is given in this book.
The bunt manoeuvre problem in its minimum time-integrated altitude and minimum flight
time formulations is the only problem treated in this book. This allows a detailed, and often
tutorial, presentation with insights into the structure and nature of the optimal solutions and
also into the advantages and limitations of the available tools. Rather than moving from one
simple example to another, and learning little of real-world computational optimal control,
we prefer the reader to stay focused on one in-depth project. Introducing other challenging
examples would, in our opinion, distract the reader from the main aim of this book: informed
use of the tools of computational optimal control in advanced engineering practice.

Approach Each of the formulations of the bunt manoeuvre problem is solved using a
three-stage approach. In stage 1, the problem is discretised, effectively transforming it into
a nonlinear programming problem, and hence suitable for solution with the public-domain
FORTRAN packages DIRCOL and NUDOCCCS or the commercial FORTRAN packages
PROMIS or SOCS. The results of this direct approach are used to discern the structure of the
optimal solution, i.e. type of active constraints, time of their activation, switching and jump
points. The qualitative analysis of the solution structure, employing the results of stage 1
and optimal control theory, constitutes stage 2. Finally, in stage 3, the insights of stage 2 are
made precise by rigorous mathematical formulation of the relevant two-point boundary value
problems (TPBVPs), using appropriate theorems of optimal control theory. The TPBVPs
obtained from this indirect approach are then solved using the public-domain FORTRAN
package BNDSCO and the results compared with the appropriate solutions of stage 1. Addi-
tionally, a comparison is made with the results obtained by the commercial package GESOP
(a software environment with PROMIS and SOCS solvers) whose solution approach can be
considered as half-way between the approaches of stages 1 and 3.
For each formulation (minimum time-integrated altitude and minimum time) the influence
of boundary conditions on the structure of the optimal solution and the performance index
is investigated. Software implementation employing the public-domain packages DIRCOL,
NUDOCCCS and BNDSCO, and also the commercial packages SOCS and PROMIS under
the GESOP environment, which produced the results, is described and documented.

Book Features As explained earlier in this preface, this book focuses on informed use of
computational optimal control for solving the terminal bunt manoeuvre, rather than devel-
opment of either the underlying theory or the relevant numerics. In this context, the main
features of this book are as follows:
PREFACE xi
• formulating trajectory shaping missile guidance as an optimal control problem for the
case of terminal bunt manoeuvre;
• devising two formulations of the problem:
– minimum time-integrated altitude
– minimum flight time;
• proposing a three-stage hybrid approach to solve each of the problem formulations:
– stage 1: solution structure exploration using a direct method
– stage 2: qualitative analysis of the solution obtained in stage 1, using optimal
control
– stage 3: mathematical formulation of the TPBVP based on the qualitative analysis
of stage 2;
• solving each of the problem formulations using the three-stage hybrid approach:
– stage 1: by using DIRCOL/NUDOCCCS solvers
– stage 2: by using the results of stage 1, understanding the underlying flight dy-
namics and employing optimal control theory
– stage 3: by using optimal control theory and the BNDSCO solver;
• analysing the influence of boundary conditions on the structure of the optimal con-
trol solution of each problem formulation and the resulting values of the performance
index;
• interpreting the results from the operational and computational perspectives, pointing
out the trade-offs between the two;
• using effectively DIRCOL, NUDOCCCS, PROMIS and SOCS (under the GESOP en-
vironment) and BNDSCO and documenting their use.

Book Organisation We have striven to write the book so that each chapter is as much as
possible independent of the others. Alas, we have not been able to attain the ideal of self-
contained chapters, but we hope that a certain degree of independence has been achieved.
We begin with the introductory Chapter 1 which is deliberately brief: it gives a very con-
cise historical context of the subject of computational optimal control, then defines the case
study investigated in the rest of the book and concludes with a detailed summary of the book,
chapter by chapter. Section 1.2 of Chapter 1 is an essential reference, as it describes the
mathematical details of the case study problem. A reader who is pressed for time may want
to glance at that section and move on.
Chapter 2 is a friendly (we hope) presentation of all the theory we use in the book. We
have made an attempt to produce a reasonably readable narrative, as opposed to a dry recita-
tion of formulae and theorems. It is not strictly necessary to read the whole of Chapter 2 in
order to follow later analyses, especially if the reader is familiar with the basics of the theory.
However, it might be useful at least to glance through the material—our hard-won experi-
ence shows that many “obvious” facts are far from clear, even for those who have already
encountered problems of computational optimal control.
xii PREFACE
Chapters 3 and 4 are the core of the book in that they consider each variant of the case
study analysed in this book. Each of these chapters is independent of the other, but both
should be studied carefully if the reader is to learn anything real from this book. An impa-
tient user may simply start reading the book from Chapters 3 or 4 and consult earlier chapters
if necessary. However, these earlier chapters are not in this book as a result of a contractual
obligation—we wrote them because we wished someone had written them when we em-
barked on the analysis of the terminal bunt problem. Perhaps reading those chapters will
save the reader some of the frustration we experienced in our encounter with computational
optimal control.
Chapter 5 is the part of our book which we strongly draw to the reader’s attention, because
it describes in practical detail—including code1 listings—real (and tested) software imple-
mentations of the analyses presented in Chapter 4. We have striven not only to explain how
to use various software packages, but also to share our (often hard-won) user experience. A
separate tutorial on the most challenging of the software packages, BNDSCO, can be found
in the Appendix.
Finally, we offer in Chapter 6 pragmatic conclusions based on our user experience with
the tools and practice of computational optimal control. Moreover, we suggest a few ways of
going beyond the approaches described in the book. If the reader is looking for “the bottom
line”, this is the place where we give it.
The book arose from a real-world, three-year project given to the authors by the UK Min-
istry of Defence, which was quite challenging and led, among others, to the first author’s
PhD thesis. Also, much of the material appearing in Chapter 3 appeared first in Subchan and
Żbikowski (2007a) while the bulk of Chapter 4 comes from Subchan and Żbikowski (2007b).

Bibliographic Comments There are three categories of optimal control books currently
available:
1. Introductory textbooks, focused mostly on theory:
(a) Entry-level texts
i. L. M. Hocking, Optimal Control: An Introduction to the Theory with Appli-
cations, Oxford University Press, 1991.
ii. E. R. Pinch, Optimal Control and the Calculus of Variations, Oxford Uni-
versity Press, 1995.
iii. A. E. Bryson, Dynamic Optimization, Addison-Wesley, 1999.
iv. D. S. Naidu, Optimal Control Systems, CRC Press, 2002.
v. D. G. Hull, Optimal Control Theory for Applications, Springer, 2003.
(b) More advanced texts, but dealing with linear problems only
i. F. L. Lewis and V. L. Syrmos, Optimal Control, John Wiley & Sons Ltd,
1995.
ii. A. E. Bryson, Applied Linear Optimal Control: Examples and Algorithms,
Cambridge University Press, 2002.
iii. B. D. O. Anderson and J. B. Moore, Optimal Control: Linear Quadratic
Methods, Dover, 2007.
1 The code listings presented in this book are available electronically from http://www.wiley.com/go/zbikowski.
PREFACE xiii
2. Advanced theoretical monographs

(a) L. S. Pontryagin, V. G. Boltyanskii, R. V. Gamkrelidze and E. F. Mishchenko,

The Mathematical Theory of Optimal Processes, John Wiley & Sons, Inc., 1962.
(b) A. E. Bryson and Y. C. Ho, Applied Optimal Control. Optimization, Estimation,
and Control Revised Printing, Hemisphere, 1975.
(c) G. Leitmann, The Calculus of Variations and Optimal Control, Plenum Press,
1981.
(d) J. Macki and A. Strauss, Introduction to Optimal Control Theory, Springer, 1982.
(e) V. M. Alekseev, V. M. Tikhomirov and S. V. Fomin, Optimal Control, Plenum
Press, 1987.
(f) T. L. Vincent and W. J. Grantham, Nonlinear and Optimal Control Systems, John
Wiley & Sons Ltd, 1997.
(g) R. Vinter, Optimal Control, Birkhäuser, 2000.
(h) M. Athans and P. L. Falb, Optimal Control: An Introduction to Theory and Its
Application, Dover, 2006.

3. More practically orientated monographs, including computational methods

(a) R. Bulirsch (ed.), Optimal Control: Calculus of Variations, Optimal Control

Theory and Numerical Methods, Birkhäuser, 1993.
(b) R. Bulirsch and D. Kraft (eds), Computational Optimal Control, Birkhäuser,
1994.
(c) J. T. Betts, Practical Methods for Optimal Control Using Nonlinear Program-
ming, SIAM, 2001.

Among the above books, the Bryson and Ho monograph, see 2b, must be singled out
as highly respected and widely used by practitioners—we benefited from it enormously. It
comprehensively presents the theory in a well-organised, clear, readable and systematic way
without distracting the reader by lengthy mathematical derivations. It is a ready theoretical
reference for any serious user of optimal control. However, this book offers more up-to-date
and expansive coverage of numerical methods than is included in Chapter 7 of Bryson and
Ho.
Titles 3b and 3a as edited compilations serve as research references, and are not intended
as textbooks. The book by Betts, see 3c above, is the only self-contained book on com-
putational optimal control and was written by a practitioner from Boeing. However, it is
dedicated to one approach to computational optimal control (direct method) and is focused
on one commercial FORTRAN package, namely SOCS. Finally, it mainly gives an in-depth
analysis of the numerical aspects of the direct method with one chapter briefly describing six
examples of varying difficulty. By contrast, our book can be considered a self-study guide to
real engineering practice of computational optimal control.

Shrivenham, England
2009
Exploring the Variety of Random
Documents with Different Content
influence it is at present a failure, though the enterprise has been
carried out regardless of cost, even in the most liberal manner.
The American mission at Teheran has really succeeded in making
some headway.
However, the at present (in regard to converts) abortive mission to
Julfa has in the educational department certainly done wonders, and
has given an impetus to the native schools, which previously, heavily
subsidised by successful Armenian emigrants, had done no work at
all, and were battened on by a set of hungry priests and mirzas, who
on some pretext or other sent away their pupils for five days out of
seven, and declared a holiday. Where the income went nobody
knew; this much was apparent, there was no result.
The long fasts of the Armenian Church are loyally kept by the
poorer of the Armenian community and by the villagers. They occupy
altogether a sixth of the whole year, and in them no eggs or meat
may be eaten, only vegetables, fruits, grain, and vegetable oil, but
wine and spirits are freely indulged in.
 CHAPTER XIII.
ISPAHAN.

Prince’s physician—Visit the Prince-Governor—Justice—The bastinado—Its

effects—The doctor’s difficulties—Carpets—Aniline dyes—How to choose—
Varieties—Nammad—Felt coats—Bad water—Baabis—A tragedy—The
prince’s view.

Almost as soon as I arrived in Julfa I received a visit from the

prince’s hakim-bashi, Mirza Abdul Wahab. This gentleman, a native
of Kashan, had received his medical education in Paris, and was an
M.D. of its University. He described his life in Ispahan as dull in the
extreme, that he was never off duty save when the Prince-Governor
was asleep, and that his anxieties on account of the vagaries of his
charge were great. The Mirza had spent seven years in France, and
had married there; he had also two native wives (his French wife
afterwards came to Teheran, but soon returned to France). He
complained of the many hours he had to stand, etiquette forbidding
any other attitude in the prince’s presence. He told me that he had to
read poetry to his Royal Highness for many hours each day.
“Not that I mind reading poetry,” said he, “but no one listens, which
is provoking in the extreme.”
I was very glad to have an acquaintance with whom I could
converse, for of course the hakim-bashi spoke French fluently. The
appointment as physician to the eldest son of the king and Governor
of Ispahan is a high employ, and the doctor hoped it would lead to
better things; but he did not like the being away from the capital. He
became shortly a Khan.
He invited me to call on the prince, and told me that his Royal
Highness would receive me at half-past eight a.m. the next day, or,
as he phrased it, two hours after sunrise. I promised to be punctual,
and duly presented myself at the appointed hour.
 I passed through a garden crowded with soldiers, servants,
persons having petitions to make, and the usual hangers-on of a
great man and his train. In a crowded outer room sat the Minister (or
real Governor), Mahommed Ali Khan, under whose tutelage the Zil-
es-Sultan is. The prince himself, being a mere youth, has no real
power, and everything is done by Mahommed Ali Khan. The hakim-
bashi now met me, and conducted me past a sentry into the private
apartment of the prince.
I took off my goloshes at the door, keeping my hat on, and making
a salute. The doctor introduced me in a few words, and the prince, a
good-looking youth of about eighteen years, motioned me smilingly
to a chair which had been placed for me opposite him. I asked after
his health in French, but he insisted on my talking in Persian, and
was much amused at the hash I made of it. He was a fine, good-
humoured youth, full of spirits.
After the first few minutes he threw off all his air of dignity and
talked and laughed merrily, asking many questions as to the
manners of Europeans, the Queen, climate of England, etc. He then
gave me his likeness, and told me that he photographed himself,
which was the case. I was regaled with tea, and took my leave,
breakfasting in the town with the hakim-bashi at his residence.
Here I saw for the first time the administration of justice in Persia.
The doctor was given the charge of the Jews of Ispahan: the Jews
had attended in a large crowd to complain of extortions practised
upon them by the soldiers who stood sentry at his gate. These men,
not content with exacting small presents from the poor people, had
insulted the wife and daughter of one of their number and severely
beaten them.
As we sat smoking the kalian at the open window, the crowd of
some hundred Jewish men and women shrieked and gesticulated;
while the two accused soldiers, who stood with the doctor’s servants,
vehemently protested their innocence. The hakim-bashi shouted, so
did the accused, so did the accusers, who wept, beat on their heads,
and prepared apparently to rend their already ragged garments.
 THE BASTINADO.

“Somebody must be beaten,” said the doctor, “and these Jews are
undoubtedly horribly persecuted.”
When the shouting was at its highest, the doctor called to the
sergeant of infantry and whispered in his ear. The two soldiers
turned pale, and the Jews proceeded to implore blessings on the
head of the doctor.
Presently a pole some eight feet long, with a transverse handle at
either end and a loop of rope in the middle, was produced, and,
kicking off their boots, the two soldiers lay down on the ground, and
each raised a foot; but the doctor was not to be appeased so easily,
and insisted on both feet of each man going into the loop. On this
being done, the noose was tightened by turning the pole by means
of the handles, and the soles of the soldiers’ feet were now upwards,
and a fair mark; two other soldiers held the ends of the pole, which is
termed a “fellek.”
The doctor now adjured the men to confess, as, if they did not, as
he put it to them, he should have to thrash them till they did, and
then have to punish them for the offence itself; whereas, if they
confessed, there would be only one beating and accounts would be
clear.
Both men confessed, though the value of a confession under such
circumstances may be doubted. Then the doctor’s servants drew
from his hauz a huge bundle of sticks some five feet long; they were
ordinary willow wands, switchy, and about twice the thickness of the
thumb at the butt; the bark was left on, and it appeared that they
were kept in water to prevent their breaking too easily.
Four of the soldiers now seized each half-a-dozen wands, and,
taking one in their right hands, awaited the signal. “Bezan!” (“Lay
on!”) exclaimed the hakim-bashi, and they proceeded to thrash the
bare soles of their comrades with the sticks; at first they struck fair
on the feet, but whenever the doctor’s eye was not on them, they
broke the stick over the “fellek” and substituted a fresh one.
The men now roared for mercy; some hundred sticks had been
broken over their feet, and, taking an average of four blows for each
stick, they had received four hundred, or two hundred each.
“Amān Agha!” “Mercy, Lord!” “Oh, hakim-bashi!” “Oh, merciless
Jews!” “Oh, Mussulmans!” “Oh, doctor, sahib!” “Oh, Lord, without
mercy!” “Oh, rascal Jews!” “Sons of dog fathers!” “Mer—cy!”
The hakim-bashi now addressed them—“Rascals, do you know
now that you are not to oppress the king’s subjects?”
“Ah,” replied one man, “but Jews—” He had better have been
silent, for the hakim-bashi raised his hand, and the beating
recommenced. I now interceded, and the men were led off, limping.
I asked the doctor if such beatings would not lame the men.
He replied, “Not in the least; they will be all right in two days, if a
little tender to-morrow. I have myself had quite as bad a beating from
my achōn (schoolmaster) when a boy. There is no degradation in the
punishment; all are liable to it, from the Prime Minister downwards.
What you have seen is merely a warning; one and two thousand
sticks are often given—I mean to say fairly broken over the soles of
the feet—and thicker sticks than mine; say, six thousand blows.”
 I asked what was the result of such beating.
“Well,” said the doctor, “I have known them fatal; but it is very rare,
and only in the case of the victim being old or diseased.”
I was told that it is really very much a matter of bribing the
farrashes (carpet-spreaders) who administer the punishment. As a
rule, a severe beating, such as is given by the king’s farrashes,
keeps a man in bed for weeks or months. Culprits much prefer it to a
fine. Here the doctor called one of his servants.
“Which would you prefer,” said he, “to lose a month’s pay or take
such a beating as those soldiers had?”
“The beating, of course,” replied the man.
“His pay is ten kerans a month,” said the doctor (seven-and-
sixpence).
Custom, I suppose, is everything; to our tender feet such a beating
would be very terrible, but Persians of the lower class walk much
barefoot; in fact, like our own tramps, unless the road be very stony,
one sees them on the march take off their boots and go bare, to
save shoe-leather or sore feet.
The doctor told me of the trials and troubles of his position, his
long hours of duty, and his many anxieties when his young charge
was ill. “Your arrival is a great thing,” he said; “you can speak as I
cannot dare to, and you can insist on proper directions being carried
out. At present, when the prince is indisposed, all the visitors and all
the old women prescribe, and as he tries all the remedies, he
becomes really ill.
“Then I have to telegraph his state to the king; then the king’s
French physician and his other hakims are ordered to suggest
remedies. You can fancy the result. Why, when I came here, the then
hakim-bashi was a young and rowdy prince, who, though a very
good fellow, kept the Prince-Governor permanently on the sick-list,
gave him two china-bowls of physic to take a day, and tabooed
everything that was nice. Of course I broke through all that, and, by
keeping him free from physic and on good plain food, he is a strong
and healthy youth.” I sympathised with the doctor, and took my
leave.
From the doctor’s house I went to the principal bazaar of the town
to buy carpets, for I had disposed of most of my own on leaving
Kermanshah, to lessen the weight of my luggage. I was shown
several hundred carpets, some four by seven yards, down to little
rugs a yard square. Some of the finer carpets, astonished me by
their beauty, and also their price—forty pounds was a usual figure for
a large and handsome carpet.
The finer and more valuable carpets were not new—in fact, few
really good carpets are made nowadays. At the time I am speaking
of (sixteen years ago) the magenta aniline dyes were unknown to the
carpet-makers of Persia, and all the colours except the greens were
fast. Nowadays the exact reverse is the case. A very brilliant carpet
is produced, and if a wet handkerchief is rubbed on it, the colours
come off; these are not fast, and the carpet is worthless.
The aniline dyes are particularly used in the Meshed carpets, and
as these are the showiest and most attractive, they are largely
exported. Of course a native will not look at them, for when he buys
a carpet he expects it to last at least a century: he is generally not
disappointed. One sees many carpets which are quite fifty years old
with hardly a sign of wear.
At the time of which I am speaking, carpets had very seldom been
exported from Persia, and consequently there was no rubbish
manufactured; now (1883) it is quite different; if a very good carpet is
wanted, an old one must be bought.
The carpets made for the European market are coarse, and the
weaving loose. Many, indeed, are made of fast colours, but gaudy
patterns only are used, and the fine and original patterns formerly in
vogue are disappearing. Of a couple of hundred carpets brought for
sale, perhaps there may be only six distinct patterns, though, of
course, the borders and arrangement of the colours may vary. The
favourite patterns are the “Gul Anar” and “Herati:” the latter is
certainly very effective, and is the pattern of nine-tenths of the
carpets exported.
 To choose a carpet, the first thing is to see if the colours are fast.
This is done by rubbing with a wet cloth. If the slightest tinge is
communicated to this, the carpet should be rejected. Then, if the
carpet is limp, and can be doubled on itself like a cloth, it is “shul-
berf” (loosely woven) and scamped. A carpet which is well woven (I
am speaking of new ones) is always stiff. Greens in the pattern
should be avoided, as they will fade to a drab, but this drab is not
unpleasing; white, on the contrary, in time becomes a pale yellow,
and is a good wearing colour, and should be chosen rather than
avoided.
The thinner and finer the carpet is, the greater is its value. The
size of the thread of the wool should be noticed, and the smaller it is
the better. It should be remembered that, in the question of price, a
thinner thread means a great difference in the amount of labour in
making.
The size, too, of the pattern should be noted, as a large pattern is
proportionately much cheaper. Again, the finer patterns being only
undertaken by the best weavers, one is more likely to get a good
carpet with a fine pattern than with a coarse. The general effect, too,
should be noted. This is never bad, but at times an eccentric pattern
is come across.
The softer the carpet is to the hand, the more valuable it is as a
rule, if it be not a Meshed carpet with aniline dye. These latter should
be avoided, as they always fade, and are of very small value.
One of the reasons why Oriental carpets last so long, is that chairs
are not used, and they are not walked on by boots, and so dirtied
and worn, but by bare feet. The carpet should now be doubled, and
the ends applied to each other. If one is broader than the other, it
shows careless work, and the carpet should be rejected as “kaj”
(uneven, or rather, crooked).
It must be then spread on a level floor and smoothed, to see if it
lies flat. Many carpets have “shatūr,” or creases; these never come
out. The carpet never lies flat, and wears in a patch over the “shatūr.”
 If all is yet satisfactory the carpet must be turned bottom upwards,
and the edges carefully examined; if any darns are seen in the
edges of the carpet it must be rejected, for the Persians have a plan
of taking out any creases by either stretching the edges, which often
break under the process, or, if there is a redundancy, cutting it out
and fine drawing it so skilfully that it is only detected on carefully
examining the back. Such carpets are worthless.
The top of the carpet should now be inspected; if the edging of
cotton at the top or bottom be blue with no white in it, the carpet is
rubbish, and merely a thing got up for sale, absolutely a sham. The
edge or finish should be either white cotton or black wool; the latter
is by far the best, but is seldom seen nowadays. The all-woollen
carpets are mostly made near Mūrghab, and by the wandering tribes
of Fars; they are very seldom exported, and are always of sad
patterns, often very irregular.
In making a carpet, the women who weave it will often run out of
the exact shade of wool used in some part of the pattern or even
ground-work; they will continue with another shade of the same
colour. This has a curious effect to the European eye, but the native
does not look on it as a defect.
The value in Persia of a carpet in the present day may, if perfect
(either new or old), be reckoned at from fifteen shillings to two
pounds a square yard. In the larger carpets nothing can be obtained
under a pound a square yard.
Of course there are a few carpets which have been made to order
for great personages which are worth more than the price I have
given, but these are not easily obtained and only at prix fou. By the
term carpet, I mean what Persians call kali, that is, in
contradistinction to farsch. Kali is our idea of carpet, that is, a floor-
covering, having a pile.
Farsch means floor-covering generally, and may be “nammad,” or
felt, or “gelim,” a thin, pileless floor-covering of coarse pattern, and
much used in Europe as a portière; in these “gelim” white greatly
predominates, and they soon get soiled and dirty; they are only used
in Persia by the villagers and poor.
 The farsch hamam-i, or bath carpet, is a finer species of gelim
made near Kermanshah; both sides are alike, the patterns are
elaborate and beautiful, and the colours very lovely, but they fade,
being mostly of aniline dye, and are harsh to the feel. Their only
recommendation is their extreme portability.
The nammad, or felts (carpets), are generally used by Persians to
go round the room and act as a frame to the carpet (kali), which
occupies the top and centre.
They are three in number for each room; two kanareh, or side
pieces, a yard to a yard and a half wide, and a sir-andaz, literally that
which is thrown over the head (of the apartment). The kanareh are
from half to two and a half inches in thickness, and are usually of a
light-brown or yellow-ochre colour, being ornamented with a slight
pattern of blue and white, or red and green, which is formed by
pinches of coloured wool inserted when the felt is made.
The best nammad are made at Yezd, and are often expensive;
they cost about thirty shillings a square yard, and will last a century;
they are two inches thick.
Nammad, however, are now getting out of fashion, for they will not
stand the wear produced by chairs, which are coming into common
use among the rich. Carpets are taking their place.
These nammad, or felts, are universally used as great-coats by
the peasantry, and are very good indeed as an outer covering, being
seamless. They are often made with bag-like sleeves with a slit at
the wrists, thus forming a glove, and when the peasant wants to use
his hands, they are thrust through the slit and the glove portion
turned back over the wrist. They are all in one piece.
The gelim, or tent carpets, are very suitable for travelling or rough
work, and being thin are easily dried. They wash well, and have no
pile.
There is yet another variety of carpet called jejim: this is very thin
and more like a plaid in consistency; it is used by horsemen, who
wrap their spare clothing in it and use it as a bed and carpet too.
 For about fifty pounds I was able to get enough carpets for all my
living rooms, and, owing to the steady rise in the price of carpets, on
my departure in nine years’ time on leave, I got as much as I gave
for them. Exactly the same as with horses after the famine, the
demand being greater than the supply on account of exportation,
prices rose considerably.
A good deal of illness occurring just at this time among the staff, I
had my attention directed to the water, which, being mostly from
surface wells, was much contaminated. I therefore engaged a water-
carrier from the town, purchased a skin and bucket for him, and the
staff were supplied with a skinful twice a day, for cooking and
drinking purposes, from the monastery well—a deep and good one.
The Persians are particular what water they drink, and invariably
employ a sakka, or water-carrier; but the Armenians generally have
a cesspool just outside their house door, and in its immediate
proximity the well is dug, often only ten feet deep. The result is
obvious.
Our superintendent being a married man, collars which I had cast
off for the last year, principally because I could not get them washed,
had to be worn; and I had to send them to Teheran by post to get
them washed, for in Ispahan the art of ironing was unknown; and the
American term for a shirt, “boiled rag,” was literally appropriate.
I made the acquaintance of three brothers who were Syuds, or
holy men, but who had the reputation of being freethinkers; these
men called on me and insisted on my breakfasting with them in the
town: they were wealthy landed proprietors and merchants. I found
their house beautifully furnished and their hospitality was great; they
discoursed much on the subject of religion, and were very eloquent
on the injustices perpetrated in Persia. They were nearly related to
the Imām-i-Juma, or high priest, a very great personage indeed, who
ruled the town of Ispahan by his personal influence. It was said that
any one who incurred his displeasure always, somehow or other, lost
his life.
Under the shadow of such a relation, the Syuds Hassan and
Houssein and their brother openly held their very liberal opinions.
They were, in fact, sectaries of the Baab.
This impostor has succeeded in establishing a new religion, the
tenets of which are very difficult to get at—a community of property
being one. Mahommedans state that a community of women is also
observed; this is, however, very doubtful.
The execution of their prophet, far from decreasing their numbers,
has had an opposite effect; many among the Ispahanis and Zinjanis
still secretly profess Baabiism.
A few years before my arrival in Ispahan (1867), a determined
attempt was made on the life of the present Shah by a few of the
fanatics of this sect, and the unsuccessful conspirators were put to
death with horrible tortures. (For details see Lady Shiel’s work.) In
these latter days (1880), when I was in Ispahan, a priest was
denounced by his wife as a Baabi. I saw him led to prison; he
avowed his Baabiism and declined to retract, though offered his life;
he, however, denied the statements of his wife and daughter, who
accused him of wishing to prostitute them to others of his co-
religionists.
On being taken to the public square for execution, after having
been severely bastinadoed, and when in chains, knowing his last
hour was come, he was offered his life if he would curse Baab.
He replied, “Curses on you, your prince, your king, and all
oppressors. I welcome death and long for it, for I shall instantly
reappear on this earth and enjoy the delights of Paradise.” The
executioner stepped forward and cut his throat.
A few days after his execution, my friends the three brothers were
arrested, their valuables looted by the king’s son the Zil-es-Sultan,
the then Governor of Ispahan, and by the Imām-i-Juma, the
successor of their former protector in the office of high priest of
Ispahan. Their women, beaten and insulted, fled to the anderūns
(harems) of friends and relations, but were repulsed by them for fear
of being compromised. They then came to the telegraph-office in
Julfa and sat in an outer room without money or food. After a few
days the relatives, rather than let the (to them) scandal continue of
the women being in the quarters of Europeans, gave them shelter.
The real cause of the arrest of these men was not their religion;
the Imām-i-Juma owed them eighteen thousand tomans (seven
thousand two hundred pounds); they were sent for and told that if
they did not forgive the debt they would be denounced and inevitably
slain. But habit had made them bold; they declined to even remit a
portion of the sum owing; they were politely dismissed from the high
priest’s presence, and a proposition made to the prince that the
whole of their property should be confiscated by him, and that they
should be accused of Baabiism and executed. This was agreed to.
They were sent for and taken from the prince’s presence protesting
their innocence, the youngest brother cursing Baab as proof of his
orthodoxy.
The next day all were savagely beaten in prison, and it was
generally given out that they would be executed; but being men of
wealth and influence, no one believed in this.
The English missionary in Julfa, the assistant superintendent of
the telegraph, and a few Armenians, addressed a letter to the prince
which, while apparently pleading their cause, really, I fear,
accelerated their fate (if it had any effect). The prince was furious,
and vouchsafed no reply.
I happened to see him professionally, and he asked me why I had
not signed this letter. I replied that I had not been asked to in the first
place; and that I should hesitate to mix myself up in the politics of the
country, being a foreign official. He appreciated my motives, and
asked if I knew the three men.
I replied that all three were my intimate friends, and I trusted that
their lives were not really in danger.
I never have been able to ascertain if his reply was merely given to
quiet me or not; it was this:—
“The matter is really out of my hands—it has been referred to the
king; he is very bitter against Baabis, as you know; nothing that
sahibs in Julfa may do will have any effect. Why, sahib, what would
your Prince of Wales say if he were interviewed, and letters written to
him about confessed criminals by obscure Persians? The
missionary, the missionary, he only troubles me to make himself
notorious.”
I explained that these Syuds were really personal friends of the
missionary as well as my own.
“All disaffected people are friends of missionaries, as you very well
know.”
I again asked him if they would be spared or not?
“I can tell you nothing more,” he said; “one has cursed Baab, he
will not die. As for the others the king will decide; for me, I wish
personally to kill no one; you have known me long enough to know I
dislike blood. I am not the Hissam-u-Sultaneh” (the king’s uncle, a
very severe Governor). He changed the subject and declined to
return to it. I cannot tell if the two elder brothers had been offered
their lives or not. I went back to Julfa hoping that they would all be
spared. The town was in great excitement. Next morning at dawn
their throats were cut in the prison, and their bodies flung into the
square. The prince had not dared to execute them publicly for fear of
a tumult.
Their houses were looted, and part of their estates; the Imām-i-
Juma’s share of the plunder was large, and he never repaid the
eighteen thousand tomans. Such was Persia in 1880. The youngest
brother, who had cursed Baab, was spared, and afterwards
reinstated in part of his family property.
 CHAPTER XIV.
JULFA AND ISPAHAN.

Julfa cathedral—The campanile—The monk—Gez—Kishmish wine—The bishop

—The church—Its decorations—The day of judgment—The cemetery—
Establishment of the Armenian captives in Julfa—Lost arts—Armenian
artificers—Graves—Story of Rodolphe—Coffee-house—Tombstone bridges—
Nunnery—Schools—Medical missionary—Church Missionary establishment—
The Lazarist Fathers.

The sights of Julfa are very few—the cathedral, or Egglesiah

Wang, and the schools of the Church Missionary Society, the
cemetery, and the nunnery, being the only objects of interest.
The Egglesiah Wang, or “big church,” is a part of the monastery of
Julfa. At the entrance, which is by a stuccoed doorway surmounted
by a Latin cross in a mud wall, is a sort of stone drinking trough,
something like our old English fonts; it is embedded in the wall. The
door is of great thickness, so that in disturbed times the monks
would be safe against attacks of Mussulmans; and for the same
reason the entrance is narrow and winding. On emerging from this
short passage, one comes to the outer court, in one corner of which
are the graves of a few Europeans who have either died in Julfa or
been brought here for burial. These are noted in Sir F. Goldsmid’s
‘Telegraph and Travel,’ and some of the Latin inscriptions are
translated into English verse by him. A large campanile of imposing
appearance and peculiar (qy. Russian) style, stands in the centre; it
is new and well made, and consists of a brick tower standing on
stone columns, and containing three bells. The rest of the courtyard
is occupied by logs of wood from the monastery garden brought here
to season previous to being sold to the Julfa carpenters. Armenians
are very unromantic. The monastery church has a door opening into
this courtyard, but entrance is usually effected through a passage, in
which are the tombs of former bishops; these are mostly mere blocks
of stone let into the ground, but the two last bishops have more
ambitious mural monuments; the last, Thaddeus, having a black
marble tablet, probably cut in India, with an inscription (in English),
setting forth his virtues; over all is his photograph. There is an inner
entrance in this passage of railings, and the walls have some small
trap-doors through which the monks in stricter days used to confess
the laity—at least the females.
A few miserable daubs on plaster in this passage represent Saint
Michael weighing good and evil spirits, etc. To the right is a low door,
which leads to the church; at the extremity of the passage is a large
courtyard, which contains the apartments occupied by the
Arachnoort and the little bishop Christopher.
There is no pretension to magnificence in these. The rooms of
Christopher are small, and comfortable in a humble manner; a few
religious engravings and paintings of saints that Wardour Street
would not look at, hang on the walls; it is carpeted with cheap rugs,
and a mattress and a couple of cushions form the furniture; in the
corner is the tall, silver-headed ebon staff of the little bishop; in a
recess stand his conical hat and hood. A room quite without
ornament forms his bedroom, and his property, a few years ago
considerable, he has made over to the Church.
The little bishop is a gouty man, and does not indulge, though
there are legends that in his youth his potations were pottle deep; he
does not even smoke, but he snuffs—a thing that most of the old
people in Julfa do. A jovial old man in a skull-cap and flowing black
robes, he insists on regaling one with gez; a sort of sweetmeat,
prepared from the gezanjebine, a mawkish exudation from a plant
found in the desert near here, and akin to manna; it is mixed with
sugar and made up into round cakes with almonds or pistachios. It is
impossible to break these cakes with the finger; they will bend freely,
but on striking them with a hammer or another cake, they fracture at
once. Ispahan is celebrated through Persia for this sweetmeat, and
large quantities are sent away in every direction. In Julfa and
Ispahan the gez is always offered on the arrival of a guest, and
urgently pressed on one; it is considered impolite to refuse. The
flavour is merely sickly sweet, and it sticks to the jaws like butter-
scotch; it is white in colour, and very cloying. The monk, too, always
has a glass of good old Kishmish wine for his friends. The Kishmish
grape is the smallest in Persia; it is a bright yellow colour, and very
sweet; it is, when dried, what we call the Sultana raisin. The wine is
a golden yellow, delicious when quite new, but terribly heady. It is a
great favourite with the Armenians, as it is quickly intoxicating. As a
rule it will not keep well, but when it does is not to be despised. A
glass of arrack is offered as an alternative, and this is more suited to
the native taste; it is, as a rule, what is called in India “fixed
bayonets.”
The monk is a laughing philosopher, and generally has some store
of local yarns; in fact, he is a sort of “vieulx Parchemins,” and his
tales would have astonished and delighted the author of the ‘Contes
drolatiques.’
Passing under the guidance of the monk we ascend on the further
side of the courtyard a long staircase and enter a huge empty room
newly carpeted, which brings us to the curtained doorway of the
bishop’s private apartment. On the walls, decorated with many
figures in cut plaster of the Russian eagle (for the bishop is a
Russian subject, and wisely takes care that the Persian authorities
shall know it),[15] hang some twenty daubs in oil of saints and sacred
scenes; these are more pretentiously framed than those in the
monk’s room, but of equal value. A high chair, considerably
ornamented with native carvings, is the bishop’s habitual seat, and at
its side is a table covered with well-bound books, which at my first
visit considerably impressed me; but I found out afterwards that they
were always the same books, so my respect for the literary
attainments of the Arachnoort somewhat diminished.
After a decorous interval the bishop enters, a handsome man—a
man who would create a furore in England—a man with large,
dreamy, black eyes, which he uses as much as the late Mr. Fechter,
a pale and interesting face, and a long, silky beard, well combed,
and black as the raven’s wing. From his neck hang an amethyst
cross and a large portrait of the Virgin, in an oval enamel surrounded
by paste. Clad in black lined with violet, his tall conical cap and
flowing black hood give a fine stage picture, which is completed by a
gentle raising of the hand (a white and delicate hand) as if to bless; a
soft, whispering, almost purring voice, completed the charm of a man
who in some other sphere would have doubtless achieved the
success usually attained by great personal attractiveness. A sort of
smile, as of a superior nature compassionating itself, spreads over
his handsome face, and in a whisper he asks after one’s health. The
glossy beard is stroked, the black eyes are rolled; coffee is brought,
a kalian, and the visitor retires, after much bowing on both sides.
The church alone remains to be seen, for the monastery itself is
not in use, the cells being filled with firewood and corn. The church is
not large, and is divided by a row of coarsely-painted wooden rails
into two compartments, the outer and larger one, which is
surmounted by a dome, being decorated with large paintings in oil of
the events in Bible history from the creation. There is nothing
particularly remarkable in these; most are copied from well-known
pictures, while others are amusing in their naïveness. The general
effect is good, a sort of gorgeousness being produced by so many
yards of brightly-painted canvas. All round the walls are modern tile-
work, presenting a florid pattern of green leaves on a white ground.
The general effect of this is not bad. The episcopal throne is placed
just beyond the railings, and consists of an elaborately-carved and
ornamented chair, covered by a wooden domed canopy, gilt and
painted in gaudy colours. A few feet in front of this is a raised
platform, some four feet from the ground, running back into a recess.
This can be curtained off at pleasure; a gaudy curtain hangs at either
side of it. At the extremity of the recess is the altar; there is only one
in the church. It has a sort of cabinet for the host, and has numerous
smaller platforms above it, each a few inches high; on these are
coarsely painted a few figures of saints.
All round the church run various pictures of martyrdoms, some of
them horrible in their grim realness, others as intensely ridiculous.
Here are shown the various sufferings of Ripsimeh virgin and martyr,
also Gregor; these are the chief saints in the Armenian calendar.
Illustrations are also seen of the parables and miracles. One of
these is the man who had the beam in his eye seeing the mote in his
brother’s. The mote is depicted as a moat, and the beam as a huge
beam of wood.
A painful daub, framed, and meant for the Entombment, is gravely
exhibited as a Raphael, and once it was intimated to me that a good
offer would not be refused. It is even copied on the outer wall near
the bishops’ tombs.
But the bouquet of the whole collection is the great picture of the
Day of Judgment. All the persons of the Trinity are depicted, and the
heavenly hosts are shown with the delights of heaven. These are,
however, in the upper part of the picture, and of small size; but in the
lower part, that representing the pangs of hell, the artist has given
free vent to his taste for horrors.
New ideas for bogey might be derived from his very vivid
treatment of the devils. George Cruikshank, the devil-drawer par
excellence, is nowhere with the Eastern artist. These devils are life-
size, and so are the nude male and female figures suffering
torments. The mouth of hell is represented as a yawning beast,
vomiting fire and smoke, into the jaws of which the nude wicked are
tumbling. As a popular preacher is reported to have said, “The devil
feeds you on fire, and if you don’t take it properly you get touched up
with the spoon.” This is actually represented.
The position of the picture is well chosen, being over the door. All
the congregation must see it on going out; and if they feel certain of
getting their deserts, it must make them uncomfortable indeed.
Julfa was once a very large place, having twenty-four well-
populated parishes, and the Armenians were extremely prosperous.
A large village, with valuable lands, and an energetic trading
population, the agricultural portion of the community being market
gardeners, within a couple of miles of the then capital Ispahan; it was
a very different place from the Julfa of to-day, which contains merely
a population of old men, women, and girls, the better description of
male having all emigrated. Ispahan, too, is now merely a vast ruin,
with small local trade and few wants. Shah Abbas the Great brought
away the entire population of Julfa on the Araxes, which now marks
the Russo-Persian frontier on the road between Tabriz and Tiflis, and
is now merely a village of a few hovels; and, giving them lands in the
immediate and best part of the environs of Ispahan, in fact its
present site, called it Julfa. The far-seeing monarch sought to
introduce the thrifty trading habits of the Armenian among his own
subjects, and to give an impulse to the commerce of his country, and
the Julfa artisans in those days were not to be despised; travelling
east and west, they brought many arts from Europe, India, and
China. The weaving of shawls at Kerman and Yezd is still an
important trade, and only the connoisseur can detect the difference
between the Cashmere shawl and its imitation, that of Kerman:
probably the European would prefer the Kermani one. Silk weaving
was doubtless brought from China to Yezd. Coarse imitations of the
Chinese porcelain are to this day common in Persia, but the art is
dead. Enamelling, which the Armenians practised, and even
patronised—for in the Persian collection in the South Kensington
Museum may be seen a large enamelled tray, quite a unique
specimen, which bears an inscription saying it was made for ⸺,
prince of the Armenians—is a dying art.
As jewellers these people attain great proficiency. Any really
difficult work was always brought by the native gold or silver smiths
of Ispahan to be finished by an Armenian of Julfa, one Setrak. The
trade of watchmaker, or rather watchmender, is almost monopolised
in Persia by Armenians; and my former dispenser was a very good
drug-compounder, having received his instruction when a convict in
India, serving his time after committing a burglary with violence.
The cemetery lies on a bare and stony plain, under a lofty hill
called the Kūh Sufi. When Ispahan was the capital this plain was all
under cultivation by irrigation, the remains of the canals being yet
visible: here lie the inhabitants of Julfa, and also a few Europeans.
Each ancient grave is marked by a huge block of stone of a cube
form, the upper face being, however, generally larger than the under
one. Some are nine feet long, a yard high, and two feet wide. Many
of the stones have Dutch, Latin, and French inscriptions. One of
these latter is the well-known one of the watchmaker to Shah Abbas
the Great. “Cy gît Rodolfe” is the inscription it bears; and here lies
Rodolphe, who was a great favourite of the king, Abbas the Great.